Archived

This topic is now archived and is closed to further replies.

johannula

Basic flaw in H3000 wind instrument calibration?

Recommended Posts

I will be racing next summer on a boat with H3000 (herc perf) instruments and Expedition.

I have always felt that the wind calibration in H3000 system is basically flawed. The main reason is the lack of leeway adjustment in wind calculations. Instead there is emphasis on upwash, which should enter the calculations.

Most boats that I have raced have had fractional rig and the wind sensor on a vertical mast head unit 1 m above the top of the mast. I do not think there will be several degrees of upwash as B&G implies in their calibration procedure.

Last summer I had Nexus (NX2) instruments with Expedition. Expediton compensates for leeway when making the wind calculations. We had very precise TWA and TWD (no "twd tacking") readings on Expedition with no upwash compensation on NX2 system. (some problem with HPC compass, but that is already discussed on this forum).

 

Unluckily B&G is very vague providing information on how the actual wind calculations are made. It states, that heeling is considered when calculating TWA and TWS for geometric error. No statement about TWD. No statement about leeway compensation.

B&G also speak for regular wind calibrations to get the instruments show valid readings. The problem is that the beans and the potatoes get mixed. I would prefer to calibrate only for the MHU offset. If the system can handle leeway compensation, you don't need anything else in non-shear, regular gradient conditions.

When shear and unusual gradient are present I would prefer to enter them separately in the system and have the compensations logged for later analysis. (Perhaps later another post on how to deal with shear).

Am I missing something or is B&G having the flaw in their way of wind calibration? I can deal with it in Expedition, but would still like to know.

Share this post


Link to post
Share on other sites

Thanks PhonyHRF for sharing your experiences.

 

Nick White states in the Expedition user manual:

 

"Please note that True wind angle displayed in Expedition and StripChart includes the effects of leeway wherever possible. Polar data from designers generally present TWA in this manner (TWA to the boat's centerline plus leeway equals TWA track) as do Ockam systems. B&G is also likely to migrate to this convention in the future. For systems that do not add in the effect of leeway into TWA (B&G, Silva, etc.), leeway is added to the TWA received from the instrument system wherever possible (Italics added). There are several benefits to this. For example, calculations of VMG have to include leeway and it makes the optimal routing function easier to use. This is why you may notice that Twa as reported by Expedition may be a few degrees wider than Twa reported by the instruments.".

 

If I get it right from Nicks text, B&G is not using leeway adjustments in wind calculations and leeway will be added to the data from H3000 system inside Expedition.

 

With Nexus NX2 system I could see 3-4 deg difference in TWA comparing the NX2 display and Expedition number box. That was equal to the amount of leeway. I think it will be the same in B&G H3000 unless I send "corrected TWA" to the displays via external channels.

Share this post


Link to post
Share on other sites

PhonyHRF,

 

your suggestion sounds good. Unfortunately I have to wait until last part of April, for the ice to go away in the Baltic. If you live in warmer climate, please let me know how it went.

 

I will probably send a note to the B&G guys about their calculations. WTP2 user guide had a nice diagram how the calculation goes.

Share this post


Link to post
Share on other sites

Hello,

 

As I understand it, H690/790/2000 and H3000 do not include leeway in TWA. This is neither a good nor a bad thing, it is just the way it is treated.

 

so twd = hdg + twa

 

Most other systems that have a leeway estimate (B&G wtp, Bravo, Ockam, Cosworth) do include it in TWA. In this case you need to remove the remove the leeway estimate from twa to compute twd as above. I believe Nexus does not include leeway in twa.

 

Or you could use course = hdg - leeway, but either way it reduces to the same thing.

 

The main reason we include leeway in twa is that it makes things a lot simpler later on - laylines, optimal routing etc.

 

Also, most (all?) VPPs include leeway in the values. For reference, a couple of interesting VPP values for leeway at tws=12kt:

 

Farr 40 3.4

Bene 36.7 4.1

Bene 44.7 3.5

Share this post


Link to post
Share on other sites

Hello,

 

As I understand it, H690/790/2000 and H3000 do not include leeway in TWA. This is neither a good nor a bad thing, it is just the way it is treated.

 

so twd = hdg + twa

 

Most other systems that have a leeway estimate (B&G wtp, Bravo, Ockam, Cosworth) do include it in TWA. In this case you need to remove the remove the leeway estimate from twa to compute twd as above. I believe Nexus does not include leeway in twa.

 

Or you could use course = hdg - leeway, but either way it reduces to the same thing.

 

The main reason we include leeway in twa is that it makes things a lot simpler later on - laylines, optimal routing etc.

 

Also, most (all?) VPPs include leeway in the values. For reference, a couple of interesting VPP values for leeway at tws=12kt:

 

Farr 40 3.4

Bene 36.7 4.1

Bene 44.7 3.5

 

 

Hi Nick and thank you for the input.

 

Correct me, if I am wrong, but I see that the lack of leeway compensation while calculating TWA and especially TWD a major problem. It will cause error (TWD tacking 2*leeway) in TWD calculations when tacking. B&G H3000 try to correct this by TWA correction table attributing to "upwash" or what ever. I find this barking at the wrong tree as the leeway is the cause of the error and easily corrected by taking leeway into calculations.

 

If I get it right there are five major sources of error in wind calculations when the MHU is situated well over non-turning mast top (no wind shear, average gradient):

 

MHU offset (mechanical)

MWA (measured wind angle at MHU) geometric error caused by healing (compensated in H&G hercules and herc. performance processors, also in Nexus)

Leeway (not compensated by H3000 processors or Nexus)

Boat speed errors

Compass offset error and especially non-zero deviation table

 

All the above can be calibrated in Expedition, when needed. You should know, how the instrument system is doing the calculations so that you don't over compensate it in Expedition (or other tactical software). It is unfortunate that he companies are so reluctant to give away exact information of their calculations (at least B&G and Nexus).

 

I have found that upwash is not usually a great problem with highly situated MHU. How do you see it?

Share this post


Link to post
Share on other sites

Having been lurking for a few days I decided to take a look at my B&G H3000 Motion manual - Page 82 shows you how to enter the leeway coefficient, so I think it is safe to assume that the main processor has this information available to them. Page 101 of the manual states that leeway is incorporated into the Course (CSE) as it is a combination of heading & leeway. Page 105 says that dead reckoning positions include leeway. Page 113 describes how you can display the leeway angle and turn leeway calibration for wind on & off (I assume they mean TWA but I am not sure?). Page 239 of the manual also documents how leeway can be damped.

 

The net-net of this would lead me to believe that leeway IS LIKELY incorporated into TWA & TWD in B&G, but to be honest I haven't really checked into this in great detail. At the next chance I get I am going to play with the K value and see if it does anything...

Share this post


Link to post
Share on other sites

I agree with you that the manual doesn't always reflect what is actually happening - I will see if I can track this down through my contacts at B&G and see what they say. In the meantime, if this is true what is one supposed to do -- ignore the TWA and TWD data from the B&G and use Expedition to output your own channels for TWA and TWD, etc...

 

Unless I am missing something, for most grand prix type yachts leeway could be between 2-3 degrees close hauled upwind resulting in a major instrument problem if you don't take this into account...

Share this post


Link to post
Share on other sites

Regarding the Expedition tags, with the latest B&G firmware (can't remember the specific version) and the latest version of Expedition (8.0.3 if I remember) sending tags seems to work pretty well on my H3000 Motion system - so much so that I have started using this feature more. I will PM you a little later with my contact info - it would nice to be able to converse with someone who is using B&G intensely...

Share this post


Link to post
Share on other sites

hey, if you have contacts can you ask them to fix the display bugs so they do not freeze when you send Expedition tags? and if you have some serious contacts can you PM me please? i have a pretty significant amount of documented bugs in the H3K platform.

Thanks!

 

Hey PhonyHRF, PM me if you want to unload your b&g stuff.

Share this post


Link to post
Share on other sites

Mistaken,

 

It would be great to know your firmware version....there appears to be a new version just out on the B&G website, but as usual B&G do not explain what the changes are...

Share this post


Link to post
Share on other sites

Having been lurking for a few days I decided to take a look at my B&G H3000 Motion manual - Page 82 shows you how to enter the leeway coefficient, so I think it is safe to assume that the main processor has this information available to them. Page 101 of the manual states that leeway is incorporated into the Course (CSE) as it is a combination of heading & leeway. Page 105 says that dead reckoning positions include leeway. Page 113 describes how you can display the leeway angle and turn leeway calibration for wind on & off (I assume they mean TWA but I am not sure?). Page 239 of the manual also documents how leeway can be damped.

 

The net-net of this would lead me to believe that leeway IS LIKELY incorporated into TWA & TWD in B&G, but to be honest I haven't really checked into this in great detail. At the next chance I get I am going to play with the K value and see if it does anything...

 

Nice reading Mistaken,

 

I did not notice the important information on page 113 about turning the leeway calibration on wind on/off. It is still unclear what this actually means as there is no other text on the topic in the manual (that I could find with search functions).

 

Jorma

Share this post


Link to post
Share on other sites

Not to be cynic, BUT the manual states quite a few things that should work one way, and don't…just like their support engineers at least they are consistent in one thing J

 

 

 

I tried different Kc values and did notice TWA and TWD direction changes which lead me to believe that it did take into account leeway BUT then I went out again and tried different Kc values from 0 to 45 (completely out of the norm) but did not see the variance following the Kc values which should had...

 

I tried to get an official answer on this from the B&G guys but never got them to…so until then, I will put $50 on Nick and 5cents in the manual and B&G.

 

 

 

PhonyHRF,

 

did you check, if you had the p. 113 switch (see above) on or not, when doing the experiments? Did you monitor the actual leeway reading when you changes K-value - did it change to something out of limits value?

 

Jorma

Share this post


Link to post
Share on other sites

I would like to return to the upwash.

 

Has anyone information on upwash? How much of it can you expect on a 40 ft racer, fractional rig, while beating on 12kn wind and 1.2m vertical MHU? Is it possible to eliminate upwash totally by installing the MHU higher?

Share this post


Link to post
Share on other sites

PhonyHRF,

 

have you estimated how much upwash you have when beating in about 12 kn wind (counting out leeway effects)?

 

The idea of having wind instruments lower, close to the deck, is interesting. I have thought it often.You could see those also on BMW-Oracle.

 

I think you could get better gradient estimates based on two TWS readings. For TWA, I think the flow is so much disturbed downstairs that you can not make accurate estimates of the "undisturbed wind field" outside effects of the boat. Perhaps you could use the lower TWS + TWA figures for trimming guides. Ideas?

Share this post


Link to post
Share on other sites

Phony HRF,

 

it gets kind of complicated. I'll come back later - some thinking to do.

Share this post


Link to post
Share on other sites

PhonyHRF,

 

I finally got the time to give you the answer I promised. The question was :"how to find the amount of upwash from the log data". I would expand this to other errors/corrections of TWD.

 

I am supposing a B&G (H3000) - Expedition system and that your boat speed, compass (especially deviation table) and MHU offset are well calibrated. I also suppose no strong currents and non-shear wind - otherwise it gets still more complicated. I will also not go into up-wash and TWS relationship for the same reason (although I should).

 

I am no pro-expert on this thing - so if some one can make it simpler (also with clearer language)- I will be pleased.

 

 

First find out the following as they were during the logging (details follow later):

 

A. What was calibrated in B&G and how they were done

 

B. How that was transferred into Expedition (some automatic transformations like leeway correction for TWA?)

 

C. What was calibrated and calculated in Expedition and how were they done

 

Then you should have an overview what is happening to your data from the sensor to the display and log. It can get quite complicated as you can do calibrations in the instuments system or in the tactical software - data might change in both systems and also while transferring between them. Things are also not so well documented (especially B&G) so there is also left room for speculation.

 

Once you have a clear idea of the above, you should start looking at the log.

 

The correctness of TWD and TWA; (upwash best seen while on beat)

 

1) TWD should be independent of the heading. When you tack the TWD should not "tack", it should remain stable. If "TWD tacking" is left after leeway compensation, the possible cause is upwash

 

 

2) When tacking the sum of TWA:s should match the tacking angle from the heading, if the TWA:s are leeway compensated. You can also use COG in non tidal conditions. If the sums do not match, this would be a place to look for the upwash - unfortunately you have to be very careful with the chain of calibrations described above.

 

 

3) Unsymmetrical TWA:s are mainly due to wind-shear and possibly MHU offset error. Upwash should be symmetric.

 

 

 

Finding the calibrations:

 

A. Inside B&G

 

- Look at the TWA correction table inside B&G (Hanbook p. 75). How does the first line (upwind) look like? As there should not bee leeway compensation inside B&G, this line should represent the correction of the combined effects of upwash and leeway (and something else?). Shear goes into MHU (MWA) offset correction (unfortunately).

 

B. Between B&G and Expedition

- I am not quite sure what happen here automatically without choosing variables on "Exp cals and calcs". If I get it right leeway correction for TWA is done automatically in certain systems (incl B&G H3000). Nick around?

 

C. Inside Expedition

 

- check table inside Expedition- Exp cals: "TWA". Are there some calibrations of TWA and have you been using them while recording your logs?

 

- check if TWA&TWS and TWD are checked in Exp Calcs (all the TWA/TWS calculations are made inside) Expedition and if I get it right it omits the TWA-correction table inside B&G as TWA/TWS calculations are done in Expedition.

 

 

With all this information you might grasp some ideas of how much upwash you have in your boat in certain conditions.

 

 

We are discussing in our team how to deal with the complexities described above. At the moment I favor an approach where most calculations/calibrations are done inside Expedition as I am not please the way B&G is mixing several things in one basket.

 

I think there is one correction you cannot turn off in B&G and that is the correction for error in MWA caused the heel of mast. You have to know, so you don't use it second time in Expedition.

 

I like the leeway correction for TWA in Expedition. I am quite unsure whether you need corrections for upwash at all, if you have a tall vertical bar for the MHU. I have not found any definitive data.

 

Shear and gradient I want to deal separately from other stuff and not get it mixed with MWA offset as in B&G. MWA offset is error in apparent wind, shear in true wind.

Share this post


Link to post
Share on other sites

I think it is very inaccurate to try to calculate upwash from past instrument data. I don't think 1.2 m MHU height is enough to totally get rid of upwash on a almost 20 m mast. Actually I think the forward pointing MHU may be as good on a beat, but clearly worse on a run (when you don't have upwash).

 

The upwash depends on your sailplan and how you use it. Trimming the sails, sail shape, the amount of roach etc. changes the upwash.

 

If I would like to find the amount of the upwash at the MHU, I would do one of the following:

 

1. Do some CFD calculations. Probably panel method is good enough. (Since you are Finnish, ask WB-Sails! They have many of those and can easily see the amount of upwash at different locations)

 

2. Do some wind tunnel testing. (Again ask WB-Sails! Though I don't know if they have measured the wind profile during the tests)

 

3. Use several or adjustable MHU's to measure AWA at different distances from the mast top.

 

Have you read this? http://www.avomeripurjehtijat.fi/index.php?option=com_docman&task=doc_download&Itemid=49&gid=156

Share this post


Link to post
Share on other sites

So tend to agree with Joakim that any number I could come up would be so inaccurate that would be useless. Anecdotally, I went from forward pointing MHU to the short vertical MHU (54"), to the 72" MHU and do remember measuring significant differences especially in wind speed and some in direction on anything with a TWA 50<. Which leads me to believe that the readings I get with the 1.2m MHU are a lot more accurate than the forward pointing MHU.

 

How long was your forward pointing MHU and mast? Mine is over 0.6 m from the forward face of the mast, which is only 12 m above freeboard (28' boat). There are two clear problem areas in my instruments (Nexus NX2). First there is a clear (easily 2 kn in 15 kn wind) discrepancy between downwind and upwind true wind speed. This can be compensated, but only with a computer. This is clearly caused by the location of the MHU. During downwind it is "behind" the sails and thus sees a clearly lower AWS as the true one. The second problem is the clearly (about 5 degrees) too low AWA and TWA while beating with a lot of depowering. Up to about 14 knots TWA is as accurate as I can tell, but then errors start to grow, most likely due to diminishing upwash. I don't have a compass sensor. I use COG (no currents around here!) and get quite stable TWD up to 14 knots. Thus I guess my instruments work like the document I linked tells, upwash and leeway cancel each other until depowering diminishes upwash.

 

If I had a compass sensor, the error would be different in TWD. The TWA would not change, but TWD would be less accurate under 14 knots and more accurate above. If I then would add leeway correction, TWA would be wrong below 14 knots, but more accurate above 14 knots. Only after adding both leeway and upwash corrections would I get equal (or better?) performance under 14 knots and better above 14 knots (if upwash could be dependent on TWS, which is not the case in NX2).

 

If you want to measure upwash while sailing, you need to have very good conditions (not much shifts, no wind shear) and sail like racing. AND you also need to have a very accurately calibrated compass and leeway in order to use TWD change as a guideline to upwash. Since I don't have the compass, I have measured TWA from the GPS tracks and compared those to TWA reported by the instruments.

Share this post


Link to post
Share on other sites

The B&G standard seems to be a bit shorter than the Nexus I have and I guess the rig was much bigger as well, thus the MHU was much more disturbed than mine. Getting much noise behind a symmetric spinnaker is not at all surprising, since there will be vortex shedding behind it (the same phenomenon that makes the jib unstable on a run). Did you notice some difference in damping also on a beat? That would be unexpected.

 

There are two tables for B&G wind correction. One is for TWA (upwash, mainly upwind) and the other for downwind TWS (not used below 90 degrees TWA).

Share this post


Link to post
Share on other sites

This is clearly caused by the location of the MHU. During downwind it is "behind" the sails and thus sees a clearly lower AWS as the true one.

 

This went wrong. The MHU sees too high AWS due to wind accelerating around the sails. Thus TWS will also be too high without correction.

Share this post


Link to post
Share on other sites

I think it is very inaccurate to try to calculate upwash from past instrument data. I don't think 1.2 m MHU height is enough to totally get rid of upwash on a almost 20 m mast. Actually I think the forward pointing MHU may be as good on a beat, but clearly worse on a run (when you don't have upwash).

 

The upwash depends on your sailplan and how you use it. Trimming the sails, sail shape, the amount of roach etc. changes the upwash.

 

If I would like to find the amount of the upwash at the MHU, I would do one of the following:

 

1. Do some CFD calculations. Probably panel method is good enough. (Since you are Finnish, ask WB-Sails! They have many of those and can easily see the amount of upwash at different locations)

 

2. Do some wind tunnel testing. (Again ask WB-Sails! Though I don't know if they have measured the wind profile during the tests)

 

3. Use several or adjustable MHU's to measure AWA at different distances from the mast top.

 

Have you read this? http://www.avomeripu...emid=49&gid=156

 

Hei Joakim,

 

thanks for a nice posting. I agree with you on many points.

 

I gave the answer on how to look at the log for up-wash, because I was asked about it. I also think there are easier ways of estimating up-wash , than going through old log files (usually missing calibration data - at least mine are). My answer was long because, I finally saw how complex the calibration schema is on an instrument+ tactical software system.

 

I have tried to contact WBS and North S guys for the simulation data, but they all seem to be on holidays now. It will be interesting to hear their thoughts.One nice thing about the simulations is that you can look at the tip vortex. In a fractional rig, there is only mast and the tip of main sail up there. The tipvortex is quite strong down-wind from mast while beating. What is the joint effect of upwash and tip vortex on vertical MHU - that is what I hope to hear?

 

Two or more wind instrument on different heights would be nice - also for detecting gradient and shear.

 

I was familiar with the ims_perf-document, but I had concentrated mainly on the "Wallying" theory and numbers. I had omitted the estimates of upwash and its relation to lift. It has a nice formula 4*CL for estimation of up-wash. Unfortunately it does not state the height or relation to the mast top.

 

I think the easiest way of finding a rough estimate of upwash on beat would be of having well calibrated boat speed log and compass. MHU offset should be calibrated, but all upwash calibrations set to zero. Leeway compensation should be applied and also the geometric correction for heel of the mast (either in the instruments or in the tactical software). After this, all (most?) TWA and TWD errors should be due to upwash. Then back to the 2 methods I suggested:

 

1) TWD tacking when the boat tacks. Divide the TWD change by 2 and you have an estimate of upwash

 

2) Calculate the sum of TWA:s and subtract the tacking angle from if . Tacking angle can be acquired from change in course (heading+leeway) or from COG. Divide the difference by 2 and you have another estimate of upwash . These two should be close.

Share this post


Link to post
Share on other sites

I do not update the TWS table all that much since I get pretty consistent and stable readings at all angles, but the TWA table is a different story and animal; I continuously work on it using the TWA AutoCal facility, but it is a huge pain to get good readings but I believe it pays. When I race with a crew you should see the faces they make when asked to come early so we can do these calibrations….when solo I do runs for an hour and a half or so depending on the day.

 

 

 

PhonyHRF,

 

you hit right on the target! The problem you describe was the reason for me to start this thread. I dislike the way B&G put together upwash and leeway-compensation and also imply every-race calibration. My hope was that we could get away with up-wash with the vertical MHU. All the other wind-angle calibrations do not need any day-to-day calibration.

:

- Leeway compensation (need a heel sensor),

- MHU off-set

- Correction for the geometric error caused my mast heel (need a heel sensor),

 

 

Gradient and shear are different things, but we'll have later another thread for them.

 

Last summer I was racing on a Nexus NX2 + Expedition system with a vertical MHU and a heel sensor. We had pretty constant TWD reading while tacking without any upwash corrections. TWS reading were also quite stable with varying wind angles.

 

I still have some hopes that you could get away without upwash corrections. If there is something to adjust (like shear or even up-wash), I'd like to use a specific correction for the thing.

We'll see how it goes.

Share this post


Link to post
Share on other sites

Jorma, i think we are missing something in the translation...you know going from whatever you speak to English and then from English to whatever i mumble....

I believe that the TWA AutoCal facility (in the B&G) is the "easiest" way to correct for upwash, gradient and shear. If i understand correctly Upwash is influenced by all kinds of variables, like traveler position, mast shape and position, main and foresail shape, heel, sea conditions, gradient, shear, boat trim, etc, etc. so by using the AutoCal you are somewhat including these variables (that only very well-funded programes like some AC, and TP52 for example can actually measure and quantify)

if you have the H3k configured to include heel and Kc (leeway coefficient) when doing the runs, then the only thing left to do is to add the leeway (or use Expedition) to the TWA. and if you want you can also add the design estimated leeway to each of the TWA in then polar to "force" the B&G to estimate targets accounting for leeway.

 

but them more i think about this the more questions i have…because doesn't the B&G back calculate TWA from AWA and calculates AWA from course (which includes leeway if available and further corrects for heel with the hercules, pitch, roll and yaw with the motion) wouldn't this mean that it is taking into account all these variables since they are included in the wind calcs?

 

It would be so freaking easy to document all this in a flow chart diagram on how each variable is used, just like Ockman!

 

PhonyHRF,

 

I agree about the language (any suggestions?). Still I feel, that we are getting the message across.

 

I understand you favoring the B&G TWA and MHU autocal functions. They have many positive sides. They are like "mixture of spices" - many things lumped together, easy to use (no offence).

I have been looking for a different way of doing the calibrations: one at a time - many small boxes where each spice is its own box. For example I would like to adjust for MHU-offset and shear separately, because shear does change quite often during longer races (and you don't stop for calibrating ;-) ). I would love to have a small red screw labeled "Shear" (a calibration parameter in the system) that would only change "offset in TWD".

 

 

Both ways have their merits.

 

I tried to get the flowchart from the B&G support - didn't get even a reply.

 

Jorma

 

By the way : I think the "back calculations" go the other way in B&G: from TWA to AWA. The "raw-AWA" or measured AWA is called MWA in B&G.

Share this post


Link to post
Share on other sites

What do you mean by AWA/AWS being calculated from true wind? This is not possible, since you can only measure apparent wind.

 

You are only measuring AWS, AWA, BSP, heading and heel, all the rest must be calculated. After calculations (or iteratively) you can then make corrections to the original measurements and get new calculated values.

Share this post


Link to post
Share on other sites

That is just expressing the same thing in a different way. Both correct for upwash (and error in leeway correction and other factors) by changing the actual measurement of AWA. AWS is not touched on a beat. The correction is just expressed in TWA units in B&G and AWA units in most other instruments. It is just the same since AWA and TWA are connected by simple trigonometry through BSP. All instruments are calculating TWA and TWS using measured AWA and AWS. There is no other way to do it!

 

After you apply the upwash, leeway and any other (what ever they call them) corrections, both AWA and TWA are changed in the calculation. It is then matter of politics whether you call AWA the angle between apparent wind and keel line or velocity vector and should it be corrected with upwash or not. I really don't even care, since you don't really need an accurate AWA on a beat, you only need accurate TWA and TWS to have the polars and TWD accurate. Thus the corrections are used to get accurate TWA and TWS and as a side product you get an accurate AWA as well.

 

The same applies to AWS/TWS correction on downwind. You can apply the correction to TWS or AWS to get the same results, but still all is calculated using AWA and AWS.

 

Well it's only 45 minutes to the beginning of new year. HAPPY NEW YEAR!

Share this post


Link to post
Share on other sites

Joakim,

 

The B&G terminology is a bit funky and your explanation isn't quite right. B&G talks about three different "kinds" of wind - measured, apparent and true. What is commonly called AWA/AWS is what B&G calls measured wind.

 

To get from Measured to Apparent one needs to correct for heel, upwash and MHU angle. Boat speed is then incorporated to get to True info.

 

But, B&G upwash corrections are all based on TWS matrices and that's why, at least in my understanding, the Apparent is the last thing calculated. As a result, their math does Measured -> True -> Apparent as that's the only way the displayed apparent wind information can include upwash corrections.

 

Weird, but, I believe, true ...

Share this post


Link to post
Share on other sites

 

But, B&G upwash corrections are all based on TWS matrices and that's why, at least in my understanding, the Apparent is the last thing calculated. As a result, their math does Measured -> True -> Apparent as that's the only way the displayed apparent wind information can include upwash corrections.

 

Weird, but, I believe, true ...

 

If you are using true wind based corrections:

 

Measured AWS/AWA/BSP -> Uncorredted TWS/TWA -> Corrections -> Corrected AWA/AWS/TWA/TWS

 

If you are using apparent wind based corrections:

 

Measured AWA/AWS -> Corrections -> Corrected AWA/AWS + BSP -> Corrected TWS/TWA

 

You can also do a mix of the two:

 

Measured AWA/AWS/BSP -> Part of the corrections (say heel) -> Partly corrected AWA/AWS/BSP -> Partly corrected TWS/TWA -> Rest f the Corrections -> Corrected AWA/AWS/TWA/TWS

 

It may even be iterative prosess, since your AWA/AWS/TWA/TWS are changing in the corrections, thus the corrections may change as well depending on are they defined using the corrected or uncorrected values.

 

When B&G says it calculates AWA and AWS from TWA/TWS, it is just a mathematical detail, since AWS/AWA are trigonometrically linked to TWA/TWS. If you alter one of those four, all the rest are altered at the same time. But all is based on apparent wind + some corrections.

 

Then it gets even more confusing when someone applies more corrections in a PC software.

Share this post


Link to post
Share on other sites

Simulation information on upwash:

 

I finally got to talk with Mikko Brummer of WB-Sails, who is one of the top professionals in sail simulation and sail making in our country. He was kind to share some of his experience. He had run simulations on what happens in the mast top. (See the picture).

 

Mikko's estimate on upwash was 3 degrees in AWA (AWS=15 kn, beat) with regular main sail and some more on a fathead main sail. The higher vertical wind sensor was 1000mm above the mast top. With normal main sail the vertical MHU had less error than forward pointing.

 

When transferred to TWA, this would mean 5-6 degrees. That is way too large to omit. From these simulations you can clearly say that upwash is present also 1-1.5 m above mast top and it should be taken into calibrations.

 

If I get it right, the "upwash" in the are of mast top is mainly caused by the special flows in that area (including tip vortex) and not so much by "bending of the wind" before the yacht as often presented when talking about upwash in lower hight. "Upwash" at mast top is a sum of error causing flows nicely packed together for calibration purposes (my thinking, don't blame Mikko).

 

So I was wrong before - now I know.

 

Jorma

post-41841-080953600 1294343146_thumb.jpg

Share this post


Link to post
Share on other sites

Joakim,

 

The B&G terminology is a bit funky and your explanation isn't quite right. B&G talks about three different "kinds" of wind - measured, apparent and true. What is commonly called AWA/AWS is what B&G calls measured wind.

 

To get from Measured to Apparent one needs to correct for heel, upwash and MHU angle. Boat speed is then incorporated to get to True info.

 

But, B&G upwash corrections are all based on TWS matrices and that's why, at least in my understanding, the Apparent is the last thing calculated. As a result, their math does Measured -> True -> Apparent as that's the only way the displayed apparent wind information can include upwash corrections.

 

Weird, but, I believe, true ...

 

That does seem weird, it would make more sense to have the matrices be based on measured wind speeds than calculated true wind speeds. Otherwise you would have circular logic in the equations.

Share this post


Link to post
Share on other sites

Good discussion here, I've had a discussion offline with Jorma and will try to answer his question in an understandable way for the wider audience… please be aware that ( a ) this is based on latest software versions and ( b ) it varies between Hydra, Hercules, Hercules Performance and Hercules Motion [i have tried to be clear where/when]

 

1) Is it possible to turn on leeway correction so that both TWA and TWD are leeway compensated?

Yes, essentially the methodology is very simple (which makes the lack of fine detail in some documents even more irritating I am sure!):

 

H3000 contains two separate functions "Heading" [self explanatory] and "Course" [Heading + Leeway Angle], the Hercules systems give the user the option to use either Heading or Course in the TWD calculation. Effectively this is a "Use Leeway?" setting.

 

Because B&G Hercules Performance and Motion systems* "back-calculate" the displayed TWA, TWS, AWA and AWS from the corrected TWD number this means leeway gets included in onward calculations - note that there are another set of numbers (MWA, MWS, raw boatspeed etc.) that create the original TWD value [some are hidden from the user].

 

*important note: not Hydra or Hercules, these don't use the Performance Wind Filter (PWF) [yet!] so only back calculate AWA/AWS from TWA/TWS, so they don't calculate TWA from TWD

 

2) Where can this be turned on?

To switch between using Heading and using Course in TWD use Cal Val 2 under Leeway [0= use heading, 1= use course (i.e. include leeway in calcs)]. From memory I believe the text that is displayed for Cal Val 2 is "USE_LWAY". Cal Val 1 under Leeway is the co-efficient. Remember you need a heel sensor to get leeway.

 

3) If the Leeway correction is turned on, are all the corrections in the "True Wind Angle correction table" (p. 75) due to upwash (also when autocal utility is used)? My original problem was that, otherwise that table mixes leeway and up-wash effects, which I do not like.

I think you had a pretty good run through the various effects on wind calculations in your posting, essentially these are the main influences on wind measurement (everything else is less significant and gets "lost in the noise"):

 

· MHU alignment (physical sensor position [measured wind angle] relative to CL of boat)

· Heel/Trim (Angular position of the sensor in airflow)

· Distortion of the airflow around the sensor due to the local effects of the mast/sail plan - generically referred to as "upwash"

· Mast Twist

· Mast Rotation

· Induced wind of the wind sensor due to sea state and manouveres

· Boat speed errors [a]

· Compass non-linearity [deviation]

· Compass offset [essentially irrelevant to this discussion as it doesn't introduce tack-to-tack errors in TWD, just an offset]

 

The main reason to use a single correction table is that most people do not have time or give the effort to separate all these items and a combined lookup table gives the best result for least effort on behalf of the user. However - if you correct for heel/trim [ b ] and leeway [c][d] then the correction tables will be essentially a combination of (i) upwash (ii) mast twist [which can be significant in some classes].

 

(Here I am ignoring rotating mast boats and assuming steady state conditions, i.e. no motion induced wind [corrected with Hercules Motion or ignored in this case].)

 

Notes:

 

[ a ] Not just actual speed errors – on systems that use damped boat speed into the wind calculations then acceleration and deceleration can seriously affect the True Wind calculations (due to lagging boat speed readings)

[ b ] Hercules systems have this feature, requires heel or heel/trim sensors (see below)

[ c ] As per above discussion, use course

[ d ] One warning on using course instead of heading – leeway is not hugely accurate when slow maneuvering, in pre-start etc. So if you are using and trending TWD in these conditions – which we all should – then be aware that errors may creep in.

 

4) There is a notion in the user guide (p. 14) that the Hercules system will correct wind data for Heel angle". Does this refer to the geometric effect of MWA getting more narrow, when mast is heeled (separate from correction of leeway)? Is it automatically turned on? Can it be turned off?

Yes, Hercules (and above) systems can correct wind data for the geometry of the sensor.

 

It is off by default (on the basis that some customers do not use Heel sensors).

 

This is controlled by Cal Val 2 under AWA - called "HEEL_CRN" [0= off, 1= correct for heel(/trim)]

 

(5) Is there any way to compensate for wind shear in H3000 (other than using MHU offset, which I do not like to do)? You can leave shear as it is (as you suggest in your manual correctly). I would like to go the other way correcting it, because then laylines will be right and also the performance indicators (vmg%, vmc%, pol%) will work OK. When racing in Baltic we quite often have 5-10 deg shear, sometimes much more.

H3000 doesn't have a direct wind shear number (WTP and Deckman both do) and to be honest we often discuss this and struggle with the many different points of view on whether H3000 should have this internally. Generally any time we discuss it we raise further questions - e.g. if you have 10 degrees of shear and you offset the TWD to compensate for it, is your performance polar (crew's targets) still valid? Experience says with that much shear you see a few % fall off the polar performance and hence need to do a polar% correction also - at that point you get into running separate navigation and performance polars (so you, as navigator, can adjust the navigation polar for laylines etc. while the crew are still striving to hit their optimum performance targets) - then you find that there are so many tweaks it is much easier to use Deckman or Expedition at that level. I am sure we will look again in future…

 

Sorry for the long post, hopefully it is all understandable.

 

Alan

 

p.s. PhoneyHRF – I reserve the right to change my mind about how it works tomorrow...!!

Share this post


Link to post
Share on other sites

It would be great to know your firmware version....there appears to be a new version just out on the B&G website, but as usual B&G do not explain what the changes are...

 

Something we absolutely have to improve 100% - the technical dealers get technical bulletins but we should publish software version summaries on the website.

 

The latest version of H3000 CPU software (r2.01 IIRC) provides a variant of H-Link communications on Hydra and Hercules level processors specifically for the new B&G Zeus navigation system - allowing them to get advanced calibration tables like the True Wind tables and to share data that isn't available by other standard means (e.g. daggerboard height...).

 

The key feature summary on the dealer bulletin summarises the following bullets (you'll like the last one - I have to go to the office an see what we changed now!):

 

  • Zeus navigation compatibility
  • H-Link LT communications in Hydra and Hercules (note: H-Link "LT" does everything except Polar transfer and remote channels which remain Performance/Motion features)
  • All H-Link communications have additional support including bridging of Heading data from Zeus (NOT for Pilot use..!)
  • Improved leeway compensation in True Wind calculations

 

Normally production releases also wrap up a few minor bug fixes, though the only one I remember was a change to a default value for rotating mast (volts/degree).

 

Regards

 

Alan

Share this post


Link to post
Share on other sites

This isn't really on topic, but I am really impressed with the way Garmin distributes their updates.

Share this post


Link to post
Share on other sites

 

H3000 contains two separate functions "Heading" [self explanatory] and "Course" [Heading + Leeway Angle], the Hercules systems give the user the option to use either Heading or Course in the TWD calculation. Effectively this is a "Use Leeway?" setting.

 

Because B&G Hercules Performance and Motion systems* "back-calculate" the displayed TWA, TWS, AWA and AWS from the corrected TWD number this means leeway gets included in onward calculations - note that there are another set of numbers (MWA, MWS, raw boatspeed etc.) that create the original TWD value [some are hidden from the user].

 

 

So how do you correct for leeway? It is easy to understand the difference of Heading and Course (=COG, if there is no current), but you should alter AWA as well, since AWA is measured "against" Heading and it should be measured against Course, when you calculate TWA/TWS.

 

If "upwash" = leeway (which is quite often the case), AWA and TWA should not change at all in the corrections, but TWD does change, since TWA is referred to different baseline (Heading vs. Course).

Share this post


Link to post
Share on other sites

 

Mikko's estimate on upwash was 3 degrees in AWA (AWS=15 kn, beat) with regular main sail and some more on a fathead main sail. The higher vertical wind sensor was 1000mm above the mast top. With normal main sail the vertical MHU had less error than forward pointing.

 

When transferred to TWA, this would mean 5-6 degrees.

 

What was the size of the rig? These things are proportional and 1 m above is very different for a 10 m and 20 m rig.

 

3 degree error in AWA at AWS =15 kn and TWA ~40 is about 4.5 degrees in TWA. Yes it is a big error for calculating targets from polars. The polars should change about 7% due to this error.

Share this post


Link to post
Share on other sites

 

Mikko's estimate on upwash was 3 degrees in AWA (AWS=15 kn, beat) with regular main sail and some more on a fathead main sail. The higher vertical wind sensor was 1000mm above the mast top. With normal main sail the vertical MHU had less error than forward pointing.

 

When transferred to TWA, this would mean 5-6 degrees.

 

What was the size of the rig? These things are proportional and 1 m above is very different for a 10 m and 20 m rig.

 

 

This was on a X-41 model. It has a P+BAS+FBD about 19,35 m.

 

You should also consider mast twist - according to measurements in the DYNA-yacht project, the effect of the twist of the mast would be about the same amount as "upwash".

post-11158-080731000 1294404972_thumb.jpg

Share this post


Link to post
Share on other sites

We have covered quite many of the error sources of wind calibration.

 

I received some more information on mast twist: Estimates of mast twist on non-rotating modern rig beating fully powered (15 kn) in flat water would be around 2-3 degrees increase in AWA. The results in one simulation and a wind tunnel experiment plus some expert opinions are in the same magnitude. This would end up in increase of 4-5 deg in TWA

 

 

Mast twist and upwash are both working the same way i.e. increasing AWA and end up in about 9 degr addition in TWA. Leeway is around 5 degrees and will decrease TWA the same amount.

 

 

One topic that has not been addressed is the increase of AWS (apparent wind speed) due to the upwash (going upwind). Estimates from one simulation and one wind tunnel are around 3% increase when beating. This increase works to decrease TWA (about 1 degr.) and increase TWS about 5% (if my quick-and-dirty calculations are right).

 

 

When you add all these together you end up in (4.5 + 4.5 - 5 - 1) =+3 degrees (addition in TWA) and when you calibrate you should remove this (subtract 3 degr).

In B&G H3000 all these TWA errors will be corrected in the "True wind angle correction table". When I looked for it in the manual (data is only for an example there), I found from the cell (upwind, 15 kn) number -2.5. So it looks that we are at least in the right ball park. (I did not look beforehand - honestly!)

 

 

How about the increase in TWS itself besides the angular aspect? 5% increase in TWS - is it large to be dealt with? It will have an effect on polars and if you are playing your polars tight, you should consider what to do.

If I understand it right, there is no way of correcting this in H3000. There is the "True wind speed correction table", but it is intended for correction in downwind work (spinnaker caused increase in TWS). Starting angle of correction can be set (usually around 165 deg) and the correction is interpolated to zero at 90 degr TWA.

 

 

If you want to correct this upwind TWS error, I think the only way is to incorporate it into your polars (the same as correcting for polar height (10m) vs MHU height (20m)).

 

 

The last thing in this lengthy note is, that I've learnt that there are two things both referred as "upwash":

 

 

1. Upwash "downstairs": the bending of wind in front of the boat before it meets the sail (sailing books have commonly picture of this)

 

 

2. Upwash "upstairs": the flows around mast top. They are complex and quite another set of flows (tip vortex etc) than the former. Alan put it (opened my eyes): " Distortion of the airflow around the sensor due to the local effects of the mast/sail plan - generically referred to as "upwash". "

 

 

Making this distinction helped me to understand, why you cannot avoid upwash ("upstairs") when calibrating.

Share this post


Link to post
Share on other sites

 

Mast twist and upwash are both working the same way i.e. increasing AWA and end up in about 9 degr addition in TWA. Leeway is around 5 degrees and will decrease TWA the same amount.

 

 

Leeway will not cause an identical change in TWA, it will cause an identical change in AWA, since leeway also causes "mast twist" by twisting the whole boat and decreasing AWA. Thus in your example leeway would cancel out mast twist and upwash in AWA and TWA, but you would get TWD error unless you corrected for leeway in the TWD calculations. Using TWD=TWA+HDC+leeway and no corrections for AWA and TWA should give everything correct in your example. This is basically how my set up works, since I use COG, which includes leeway "automatically". It seems you can not do that in B&G, since it will use leeway also to correct AWA.

Share this post


Link to post
Share on other sites

Mast twist and upwash are both working the same way i.e. increasing AWA and end up in about 9 degr addition in TWA. Leeway is around 5 degrees and will decrease TWA the same amount.

 

 

Leeway will not cause an identical change in TWA, it will cause an identical change in AWA, since leeway also causes "mast twist" by twisting the whole boat and decreasing AWA. Thus in your example leeway would cancel out mast twist and upwash in AWA and TWA, but you would get TWD error unless you corrected for leeway in the TWD calculations. Using TWD=TWA+HDC+leeway and no corrections for AWA and TWA should give everything correct in your example. This is basically how my set up works, since I use COG, which includes leeway "automatically". It seems you can not do that in B&G, since it will use leeway also to correct AWA.

 

 

Hi Joakim,

 

just a short comment on calculations. You can calculate this also from AWA side, but the easiest is:

 

Suppose TWD =90 deg (east), Hdg =135 deg and the boat has no leeway. TWA = TWD - Hdg = -45 deg. Then introduce leeway that causes the boat "slide sideways" so that the true direction of its movement is 5 deg lower (leeway= 5 deg). The TWD and hdg will stay the same but the actual course = hdg+leeway= 140. If you want to have "leeway adjusted TWA" your reference changes from heading to course (hdg+leeway). Course equals COG, if no water movement is present. Your final formula for TWA (leeway adj)= TWD - course= TWD - hdg - leeway = -50 deg.

If leeway increases by x deg, then adjusted TWA will increase with same amount if everything else stays the same.

 

You can calculate this also starting from apparent wind. You will end up the same. There will be a new wind component caused by leeway (perpendicular to hdg) that you have to take into consideration when doing the AWA calculations.

Share this post


Link to post
Share on other sites

just to add another variable....why not include set and drift to the wind calcs if you have it. here in the PNW where currents of 3knots+ are not uncommon and they mess the calibration way more than upwash in some cases

 

Because you are sailing in the water, thus only interested in the wind speed and direction against the water. It gets more complicated, if the current changes during the race.

Share this post


Link to post
Share on other sites

i would respectfully disagree, include it just as leeway and the wave filter for TWData, with today's GPS technology a lot more easy and accurate than correcting for upwash which is fairly constant for each wind speed. For Example here you can be beating in the port tack with a current of 3 knots right at the bow and when you tack to starboard the current is now at 70-90 degrees in you beam. I have not made any math on this but I would guess that this current effect has a lot bigger effect in the TWD than leeway.

 

Yes current can have a big effect on the ground wind direction and speed. But you are stuck to water and move in it. Thus your performance is only related to the wind over the water not wind over ground.

 

At least the Nexus Race Software calculates the ground wind also, but that is separate to TWD.

Share this post


Link to post
Share on other sites

Think of a situation of wind over ground = 0 knots and current 3 knots. Can you sail (=have speed over water and steer)? Do you feel any wind? Where is head to wind?

 

Then think of another situation of wind over ground 3 knot and current 3 knots in the same direction (well opposite). Can you sail (=have speed over water and steer)? Do you feel any wind? Where is head to wind?

 

Do you want to see the wind over ground or the wind over water in your instruments?

Share this post


Link to post
Share on other sites

you can get really complicated with any of this, but in the simple view..., you really only want to see "ground" wind when you are comparing observations on your boat, with GRIB, or other WX forecasts, because forecast winds are ground winds.

 

in most other cases, the so-called "sailing wind" which does not remove the effect of current, is most useful - e.g., when selecting sails, when comparing to target speeds, etc.

Share this post


Link to post
Share on other sites

not sure i totally understand your post...,

 

but - the wind generated by the current, isn't "noise", no matter how big the current is...

 

the sailing wind is the _only_ wind your boat sees..., it has no knowledge of the ground wind..., and i don't really see why you want to know what it is in the situation you describe.

 

you seem to be thinking that the ground wind (wind referred to a fixed reference) is the "true" wind that you should be concerned with on the boat..., and in general it isn't.

 

your boat only sees the wind referred to the (sometimes moving) water, and a change in that wind due to a change in the current shouldn't really be treated any differently than a change in wind due to a puff, or shift.

 

of course, it's good to know whether a change in the sailing wind is because of a change in ground wind, or a change in current, but subtracting current from your sailing wind is not a good way to do this.

Share this post


Link to post
Share on other sites

It would be quite nice with TWD needing no correction for currents, because everything was water referenced. If only the buoys would float the same way with sea anchor .

 

As they are tied to the bottom things get a little bit more complicated and we might end up using georeferencing. Your boat sees only the water, but you are looking at the next mark.

 

As we do not have any real currents in open waters in the Baltic, I'd like to ask an elementary question:

 

How do you get around the buoys with large current on the race course. How do you find your laylines (I know Expedition can do it) and how figure what is the fastest way upwind with current coming like somewhere between head wind and cross wind?

 

You need drift and set, coordinates for the marks, but do you have any use for GWS and GWA?

 

Is there any new aspect for wind or other calibration in this scene?

 

 

Jorma

Share this post


Link to post
Share on other sites

I like B&G's term of "measured" wind.

 

It is interesting that improved instruments are taking sailors to where other earth scientists have long tread. The concepts of aleatory variability and epistemic uncertainty are rearing their ugly heads!

 

For those not in the field, "aleatory variability" is multi-syllable words for "The natural world is filled with random variability.". The concept is that no matter how perfectly you know something, there is still probablistic variability at work. Heisenberg's Uncertainty is an an example. Epistemic uncertainty on the other hand is a multi-syllable word for "We don't know all the forces at work, our instruments aren't perfect, our concepts aren't perfect, our models aren't perfect or we may be using the wrong model, our data has gaps, our data is too voluminous and comes in too fast for us to process, on and on.". Basically, epistemic uncertainty is what we seek to reduce in order to accurately measure aleatory variability!

 

When you get to the point where you can quantify that your epistemic uncertainty is the same magnitude as (or less than) the aleatory variability, you are reaching a point of diminshing returns and the difficulty in determining one from the other becomes exponential.

Share this post


Link to post
Share on other sites

Thanks PhonyHRF for two nice posts,

 

1) You tell that your wind data gets more stable when including drift and set in the calculations. If the current parameters are calculated at the boat, what you do is, that when correcting for them, you change all the wind calculations to groud reference (GPS-based: COG,SOG+wind)) and cut out the water reference (hdg, BS + wind). Things will get more stable with a good GPS set, but breaking the water reference would cause problems. How do you use in practice the GW-data?

 

You also brought up the question of using two GPS sets and doing post processing on logs. "Mistaken" was bringing this up in his post on "On deck compuer"- Topic (here). How do you do it? It might be a good idea to start a new topic/thread on this matter. I have 10Hz GPS that is working fine on starts but not very accurate for measuring SOG.

 

 

2) Nice flow chart.

The more I have thought these wind calibration problems, the more I have been puzzled why B&G H3000 is using the corrections mostly on true wind (TW) side and not the AW side as most of the errors are errors in the AW (mast twist, upwash). (AW is here the traditional AWA, AWS - not the B&G AW-back calculated AWA, AWS) Only MHU off-set is calibrated on the AW side. Leeway correction is also calculated starting from AWA side, although a change of reference is also made.

 

In your flow chart you have included these corrections to be made on TW-data. I am also not a physicist, but I feel that as they are AW-errors, they should be corrected on the same side (as Ockam does).

One possibility of course, is that the TWA correction table is only "on surface" a TWA/TWS-table. The data might go to the AW-side after some calculations. TWA/TWS table is easier for people to apprehend than AW/AWS table.

 

Maybe some intermediate variables are used, like in WTP2 (see picture). You can find it at the end of WTP2 manual here .

 

Hope we get Alan to comment. We definitely need him.

post-41841-065350900 1294951747_thumb.jpg

Share this post


Link to post
Share on other sites

Does it say in your literature what the precision and accuracy (different things) is of the different sensors you are using? The linearity of the response? If reported, most electronic instruments would have this as a percent of full range output. A very good instrument might be 1% or even 0.5% (very, very good instruments at 0.1%) of full range output. What is the output range of the anemometer? Wind direction indicator? Knotmeter? Heel indicators? How is that lack of precision screwed? How fast is it updating? Fast enough to smooth out natural variability?

 

You guys are going back and forth about 1 degree corrections (assumed). Are your instruments really good enough, in total, to benefit from these minute corrections? I ask because this is the exact kind of question I have to answer every day in my own work.

Share this post


Link to post
Share on other sites

Does it say in your literature what the precision and accuracy (different things) is of the different sensors you are using? The linearity of the response? If reported, most electronic instruments would have this as a percent of full range output. A very good instrument might be 1% or even 0.5% (very, very good instruments at 0.1%) of full range output. What is the output range of the anemometer? Wind direction indicator? Knotmeter? Heel indicators? How is that lack of precision screwed? How fast is it updating? Fast enough to smooth out natural variability?

 

For my instruments (Nexus):

 

MHU: angle better than +-2 degrees, speed 5% of reading (not MAX, which is 50 m/s).

LOG: 1% of reading (not MAX, which is 30 knots)

 

I don't have a heel sensor and I don't think its accuracy is that important, since 1 degree error in heel will not show much in other calculations.

 

Those are the given transducer accuracies, BUT much more important is the installation of them. E.g. having the log transducer off the centerline and/or too close to keel or stem totally spoils the accuracy. I have a good location for it and on a W/L course comparing integrated SOG to integrated boat speed gives typically less than 1% difference. I haven't touched the calibration since 2008 and still get the same accuracy. The update rate is 3 Hz, but needs dampening to give anything useful.

 

The +- 2 degrees of AWA is already quite much, but here as well the installation has a big effect due to disturbance from sails. Also I think that this +-2 error does is in the nonlinearity and it doesn't vary from day to day. Thus AWA=26 will be 26 even the next day. Also 5% for AWS is quite much, but here again it is not random error, it is an error in the linearity + offset and thus could be calibrated.

Share this post


Link to post
Share on other sites

Jorma, i do not understand what problem you are referring to when you cut the water out. you still have apparent wind data for sail choice for example, and then you have Ground wind data; which in places with very little set and drift is exactly the same as you measure with what we currently call TW data. Grown wind data and Tw data read almost the same (less than 0.01 Standard deviation in 7 minutes tacks, 35 runs across multiple days (over 20) across wind ranges from 5knots to 20 knots) when sailing in the lake…can you explain what problems you are referring to? I would like to think about them..

 

If you have currents, GW and TW will be totally different. GW is useless as basis for polars, since you are sailing with moving water. GWA could be anything on a beat, but TWA will always be about 40 degrees (+- shear effects). Please consider the questions I had earlier.

Share this post


Link to post
Share on other sites

 

regular polar data (high end racing programs can "simulate" this and get specific polars, or racing program with the know-how can get polars for different conditions based on historical data which is something i do to compensate for seas condition, wind weight, racing with a crew or solo, sail inventory) assume no set and drift., no shear, no gradient so if anything removing set and drift from the wind cals would give you a more accurate reading, would it not?

 

BTW, for me very little set and drift is <0.4 Knots

 

Please consider the case no ground wind and 3 knots current. Any polars based would predict target speed to zero, if your input is zero wind (the ground wind). But you are happily sailing in 3 knots of true wind. Which one is more accurate?

 

Sure if the current is very small GW is almost the same as TW, thus you haven't done any harm to your polars in most cases. But why do you want to correct for the very small current?

Share this post


Link to post
Share on other sites

For my instruments (Nexus):

 

MHU: angle better than +-2 degrees, speed 5% of reading (not MAX, which is 50 m/s).

LOG: 1% of reading (not MAX, which is 30 knots)

 

I don't have a heel sensor and I don't think its accuracy is that important, since 1 degree error in heel will not show much in other calculations.

 

Those are the given transducer accuracies, BUT much more important is the installation of them. E.g. having the log transducer off the centerline and/or too close to keel or stem totally spoils the accuracy. I have a good location for it and on a W/L course comparing integrated SOG to integrated boat speed gives typically less than 1% difference. I haven't touched the calibration since 2008 and still get the same accuracy. The update rate is 3 Hz, but needs dampening to give anything useful.

 

The +- 2 degrees of AWA is already quite much, but here as well the installation has a big effect due to disturbance from sails. Also I think that this +-2 error does is in the nonlinearity and it doesn't vary from day to day. Thus AWA=26 will be 26 even the next day. Also 5% for AWS is quite much, but here again it is not random error, it is an error in the linearity + offset and thus could be calibrated.

 

That is very odd that they would report error as a percentage of reading rather than full scale outputyuhh1. But then, why would marine instruments report their instrument error in the same way that every other industry, from Fluke to scientific instruments, do? :angry:

 

Unfortunately, you can't calibrate out precision range of an instrument, it is intrinsic and is after you calibrate offsets. It's not only linearity though (which is related to accuracy, not precision), there are other errors that can happen to analog instruments with temperature, humidity, cabling (unless it is a current limited system), etc.. And it does change (linearity, etc.), it changes with use, that is why critical electronic instruments such as load cells, etc. require regular re-calibration certification. And if anyone thinks marine wind instruments are built superior to industrial and scientific instruments.... Don't even get me started on taking an initial zero reading to get an accurate offset!

 

So, yeah, +-2 degrees means just that, it will be somewhere in there, and you are right, that is an awful lot of error. The same with 5% AWS. You are also right that those errors are based on perfect installations, so that is best case.

 

I'm not saying that this isn't a valuable discussion, and I am enjoying it as well. I'm just saying their are diminishing returns in reducing data, hoping to get a target that is smaller than the inherent error of your instruments. You are better off getting a more accurate instrument (if one is available). My engineer clients do this all the time and it drives me nuts, they see a certain number at the absolute low end of the scale, and think it is accurate, but they are misleading themselves. Their/our problem is they are interested in two different scales, a macro scale and a micro scale and the environment is too harsh for the micro scale instrument to survive. They really want highly accurate low pressure data in an environment which also sees very high pressures. They are trying to accurately determine 6kpa with an instrument 1% accuracy over FRO of 2000kpa where pre and post reading zero offsets are up to 60kpa. Not going to work (and these are professionals).

 

It's difficult to discuss this not knowing people's knowledge level, but most lay people don't understand there is a difference between accuracy of measurement and precision of measurement in electronic instruments, the same as with any instrument and it typically comes down to full scale output. Professionals forget too sometimes, blinded by the digital numbers.

 

Now, there are other ways to attack this issue - including probablistic methods of data analysis - but that is a whole nother topic and I don't want to get all "Bayesian" on anybody.

 

What really seems discussed here is not a "calibration" issue, so much as a "calculation" issue. That is why I like B&G's naming of "measured" wind. That is accurately descriptive. It is the pure reading you get out of the instrument after you calibrate it.

 

I second PF's frustration. It would be great if wind instruments were standard analog 4-20ma, this would reduce cabling issues and the frequency would be down to your A/D which could more easily be upgraded. Or instruments that used different methods of wind strength/direction detection. Failing that, the current analog instruments could be better engineered to reduce mass damping effects as well.

 

But, the most expensive thing is the displays anyway, so....

Share this post


Link to post
Share on other sites

I consider that 3 knots of apparent not true just as i would when under motor. you are suggesting that the current is pushing you forward (i.e. >90 degrees toward where you want to go; off course you could in some boats make bigger angles, and theoretically at all alngles with a wing, but then one would use ground wind and apparent wind reference) otherwise you would be being pushed backwards and you would be seeing a negative wind and sailing away from where you want to go with the boat as reference , and i want to correct for all set and drift.

 

 

Think of the following situation. You want to go south, current pushes you 3 knots to south and the is no GW. Before you hoist any sails you have SOG 3 knots south and your instruments are showing no wind, since they show GW. But there is 3 knots of apparent wind from south. Hoist the sails and start beating 45 degrees to the TRUE wind. Now you are heading SW at 2 knots boat speed, but your SOG is 4.64 knots and COG is 198. How much benefit is there from GWA, which is totally unstable, due to no GW?

 

Then another situation. You want to go south, current pushes you 3 knots east and there is 3 knots of south GW. No your GW instruments show beat to the mark, but true wind is from SE and 4.24 knots. How to get to the mark? Set heading towards the mark (south) and you are beating at say 3 knots, but your SOG 4.24 knost and COG SE. What the heck, you have to lower your course and you find out that you have COG toward mark (south), when your heading is 217. Now your TWA is 82 and your boat speed 5 knots with code zero. SOG is 4 knots. How useful it is to know that GWA is 37 degrees and GWS is 3 knots compared to TWA 82 and TWS 4.24 knots?

 

I have never sailed in over 1 knot current. If I would, I would use true wind and true wind polars against the water and the just add drift and set to the polars. Then there would be two different polars. E.g. in the latter example the first ones tell you just like without any current, that at 4.24 TWS and 82 TWA you should have a boat speed of 5 knots. Then the other set of polars tell you that while you have TWA=82 your will have SOG=4 and COG=180 in the current we are having now.

Share this post


Link to post
Share on other sites

That is very odd that they would report error as a percentage of reading rather than full scale outputyuhh1. But then, why would marine instruments report their instrument error in the same way that every other industry, from Fluke to scientific instruments, do? :angry:

 

Unfortunately, you can't calibrate out precision range of an instrument, it is intrinsic and is after you calibrate offsets. It's not only linearity though (which is related to accuracy, not precision), there are other errors that can happen to analog instruments with temperature, humidity, cabling (unless it is a current limited system), etc.. And it does change (linearity, etc.), it changes with use, that is why critical electronic instruments such as load cells, etc. require regular re-calibration certification. And if anyone thinks marine wind instruments are built superior to industrial and scientific instruments.... Don't even get me started on taking an initial zero reading to get an accurate offset!

 

So, yeah, +-2 degrees means just that, it will be somewhere in there, and you are right, that is an awful lot of error. The same with 5% AWS. You are also right that those errors are based on perfect installations, so that is best case.

 

They really want highly accurate low pressure data in an environment which also sees very high pressures. They are trying to accurately determine 6kpa with an instrument 1% accuracy over FRO of 2000kpa where pre and post reading zero offsets are up to 60kpa. Not going to work (and these are professionals).

 

I second PF's frustration. It would be great if wind instruments were standard analog 4-20ma, this would reduce cabling issues and the frequency would be down to your A/D which could more easily be upgraded. Or instruments that used different methods of wind strength/direction detection. Failing that, the current analog instruments could be better engineered to reduce mass damping effects as well.

 

Wind speed and boat speed are measured by counting revolutions of the impeller. Thus it would be pointless to use the maximum scale as a reference for error. There is no maximum to the transducer (well it may be broken in too fast flow) in the same sense as for a load cell or a pressure transducer.

 

+- 2 degrees means that there is remaining error of that amount in the calibration curve. There is a calibrated value for each 45 degrees, thus the error is less than +-1 at those points and may be up to +-2 in between the calibration points.

 

Even the industrial wind sensors use the same way of reporting accuracy (in degrees, m/s or % of reading) and are about as accurate, but much more expensive: http://www.vaisala.com/en/meteorology/products/weatherinstruments/windsensors/Pages/default.aspx

 

All of these, wind speed, wind angel and boat speed, are digital. Speed is counted from pulse frequency and angle from pulse PWM. There is no reason for these to change in time electronically. What may happen is that the impeller becomes dirty or the bearing gets more friction or play. which may change angle PWM.

 

I think 4-20 ma analog wouldn't be any better, mainly just worse. There are no AD converters in Nexus systems. Just digital pulses or data in RS-485 network.

 

I have done quite a lot with different kind of measurement systems. I have e.g. designed and build my own altimeter, which uses a high accuracy absolute pressure sensor. It has a resolution of 0.2 m and accuracy is a few meters after calibration to current barometric pressure and temperature. I also make very accurate inclinometers, which have a resolution of 0.001 degrees and accuracy for angle difference is 0.5% of difference + 0.01 degrees. Offset error is about 0.1 degrees for absolute angle, but that is not important, since it is used to measure the difference of two angles.The calibrated measuring range is +-12 degrees.

Share this post


Link to post
Share on other sites

Wind speed and boat speed are measured by counting revolutions of the impeller. Thus it would be pointless to use the maximum scale as a reference for error. There is no maximum to the transducer (well it may be broken in too fast flow) in the same sense as for a load cell or a pressure transducer.

 

I don't want to argue here because you obviously have some knowledge of the subject, but I do want to want to clarify. Yes, there is a maximum scale as a reference for error. It reads the pulses and reads the time, there is an error (however small) in what triggers a pulse, there is an error, however small, in how it reads the time. There is a maximum amount of pulses it can read in a certain time and that is FRO. I have a very hard time believing that the accuracy at 1kt is 0.05kts and 0.1kts is 0.005kts. If so, that is one hell of an instrument!

 

+- 2 degrees means that there is remaining error of that amount in the calibration curve. There is a calibrated value for each 45 degrees, thus the error is less than +-1 at those points and may be up to +-2 in between the calibration points.

 

I don't know how Nexus instruments work (see below) but for B&G (and this thread I thought was about B&G wind instruments) wind angle is a potentiometer (my understanding). Its precision is its precision. Even with calibration points every 10 degrees, it will only be as precise as it is. You can't make up precision with increased calibration points. A yardstick is not a micrometer.

 

Even the industrial wind sensors use the same way of reporting accuracy (in degrees, m/s or % of reading) and are about as accurate, but much more expensive: http://www.vaisala.com/en/meteorology/products/weatherinstruments/windsensors/Pages/default.aspx

 

Couldn't find their specs at the link, but I trust you on that.

 

All of these, wind speed, wind angel and boat speed, are digital. Speed is counted from pulse frequency and angle from pulse PWM. There is no reason for these to change in time electronically. What may happen is that the impeller becomes dirty or the bearing gets more friction or play. which may change angle PWM.

 

Again, you are talking Nexus, not B&G, so it is apples and oranges. I don't know what method they use with Pulse Width Modulation (I'm assuming that's what you mean), but the same principle applies. A physical (analog) reading (wind vane angle) is digitized. The precision of the orientation of the wind vane itself is probably the limiting factor.

 

I think 4-20 ma analog wouldn't be any better, mainly just worse. There are no AD converters in Nexus systems. Just digital pulses or data in RS-485 network.

 

4-20ma systems are as accurate as any "digital" system. I have two right in front of me that is accurate to 0.001% of FRO. One at 0-100kpa, the other at 0-2000kpa. I admit they have their downside and are in most cases are being replaced with digital systems, but more due to size, cost, and manufacturing simplicity, not due to increased accuracy or reliability. In fact the 4-20ma systems are more reliable and robust than their digital replacements.

 

Obviously you have experience with creating digital instruments. That is why I said it is difficult to discuss this when you don't know people's experience/knowledge. Many of those sensors you can get through Sparkfun for Arduino. But that is not what we are talking about here.

 

Anyway, there is a range of knowledge here and some benefit from having more information.

 

My point is the same. The precision of the instruments is the limiting factor.

Share this post


Link to post
Share on other sites

I have done quite a lot with different kind of measurement systems. I have e.g. designed and build my own altimeter, which uses a high accuracy absolute pressure sensor. It has a resolution of 0.2 m and accuracy is a few meters after calibration to current barometric pressure and temperature. I also make very accurate inclinometers, which have a resolution of 0.001 degrees and accuracy for angle difference is 0.5% of difference + 0.01 degrees. Offset error is about 0.1 degrees for absolute angle, but that is not important, since it is used to measure the difference of two angles.The calibrated measuring range is +-12 degrees.

 

Did you experiment with a custom inclinometer / heel meter for the B&G system? would a DIY work?

 

I did some quick tests with a accelerometer connected to the Heel input, and it seems to work, now only work out callibration, offset, different curves, damping, etc.

 

Erik

Share this post


Link to post
Share on other sites

Hi guys (and girls if any),

 

Interesting discussion going on here. So many topics floating around, that I don't know where to start.

 

The topics that we are dealing at the moment:

1) Original B&G calibration theme (including its logic - flow chart) - still waiting for Alan

2) Ground wind issues

3) Quality of measurements: (accuracy, precision, linearity, frequency of measurements etc.)

4) How to improve the use of GPS - pre/post processing

5) Calculations moved to non-system computers - how to go that way?

 

 

My suggestion would be to move the 3) quality of measurements and 4) GPS (perhaps also 5 calculations moved to non-system computers) to their own threads. I was planning any way for a topic "How do I know my numbers are right- quality control of calibration".

Votes? We can also continue here, if you prefer.

 

People are presenting their backgrounds so here is mine: My daytime/winter work in the humanistic/medical field and you know what I am doing in the summers(and part of the night time during winters - as now). I did work in research several years ago. I am familiar with quality controls when measuring such things as "interpersonal relationships" or "coherence of ego". After those, measuring this wind thing does not feel hopelessly difficult.

 

I will comment on the themes above in separate posts to keep things easier to follow. We start new threads, if the idea is supported.

 

Jorma

Share this post


Link to post
Share on other sites

I think there is a problem that the post is called H3000 wind instrument "calibration" when it is really talking about H3000 wind "calculation". They are different things.

 

I could see a few different threads on this topic. GPS is certainly a whole different topic and deserves it's own thread. Comparative wind calculations of different systems, Expedition, Nexus, NKE, B&G, etc. also deserves it's own thread. Quality of wind and other transducers also deserves it's own thread.

 

The problem is it is hard to get the ball rolling!

Share this post


Link to post
Share on other sites

I think there is a problem that the post is called H3000 wind instrument "calibration" when it is really talking about H3000 wind "calculation". They are different things.

 

I could see a few different threads on this topic. GPS is certainly a whole different topic and deserves it's own thread. Comparative wind calculations of different systems, Expedition, Nexus, NKE, B&G, etc. also deserves it's own thread. Quality of wind and other transducers also deserves it's own thread.

 

The problem is it is hard to get the ball rolling!

I favor this highly:

"Comparative wind calculations of different systems, Expedition, Nexus, NKE, B&G, etc. also deserves it's own thread"

Jorma

Share this post


Link to post
Share on other sites

On ground wind:

 

Things get complicated when the water starts moving, also with TW-approach ( like laylines and optimum course).

 

PhonyHRF pointed out that when there is no current GW works OK and is the same as TW. So why use it then? I think PhonyHRF gave us a good reason for using GW: it is more accurate. If this is true, then in non-current environments there might be some advantage of using GW-variables. The current varibles (drift and set) would be only error terms that would be corrected out in GW calculation as their true value should be zero. The polars and laylines would work all right. This is empirical question - which way is more accurate and precise?

 

Beware of currents though. Even 0.4 kn cross current (BS=6kn) can change your COG and depart it from Hdg+leeway by 4 degrees and have effect of TW and GW number departing from each other.

 

To Joakim: If I understand right, you have a hybrid system on your boat, where BS is water referenced (paddle wheel), but course (leeway+hdg) is ground referenced (COG from GPS). You would have quite peculiar TW figures with your system in 3 kn cross currents. This is not to mock your boat, you have done marvelously in races - just to show that you can do well also by GPS based (ground referenced) instruments in low current conditions like in Baltic. The errors caused by small currents are well compensated by accuracy of COG compared to low-end compass.

Jorma

Share this post


Link to post
Share on other sites

I don't want to argue here because you obviously have some knowledge of the subject, but I do want to want to clarify. Yes, there is a maximum scale as a reference for error. It reads the pulses and reads the time, there is an error (however small) in what triggers a pulse, there is an error, however small, in how it reads the time. There is a maximum amount of pulses it can read in a certain time and that is FRO. I have a very hard time believing that the accuracy at 1kt is 0.05kts and 0.1kts is 0.005kts. If so, that is one hell of an instrument!

 

Yes the given 5% accuracy is not valid for the whole range. At the same time the error does not depend on the FRO. Maybe it is actually something like 5% of reading + 0.1 m/s. You can compare this to a tape measure. A 1 m tape measure is not more accurate than a 30 m tape measure for measuring 1 m, They both have the same offset error, reading error and then they have a percentage error caused by thermal expansion or manufacturing error.

 

Measuring time very accurately is easy with processors and the output shown is already an average of several pulses. I don't think that is a problem at all. The problem is that the impeller rotational speed is not totally linear with wind/water speed and there is also an offset due to friction.

 

It is totally different with a load cell or a pressure transducer, since they are very different for different full scale. Their accuracy is logically %FRO, since FRO determines the mechanical output (e.g. cross sectional area of a load cell) and also voltage output and thus possible errors in AD.

 

I don't know how Nexus instruments work (see below) but for B&G (and this thread I thought was about B&G wind instruments) wind angle is a potentiometer (my understanding). Its precision is its precision. Even with calibration points every 10 degrees, it will only be as precise as it is. You can't make up precision with increased calibration points. A yardstick is not a micrometer.

 

I don't know how B&G works, but there are different types of errors. Some are more or less random, there might be some hysteresis, some depend e.g. on temperature and some stay the same. I think the angle error of Nexus (and potentiometer type) stay mostly the same. By that I mean that every time the transducer is in the same position it will read the same, but that same may be 2 degrees off the accurate angle. This kind of error could be removed with more dense calibration.

 

 

4-20ma systems are as accurate as any "digital" system. I have two right in front of me that is accurate to 0.001% of FRO. One at 0-100kpa, the other at 0-2000kpa. I admit they have their downside and are in most cases are being replaced with digital systems, but more due to size, cost, and manufacturing simplicity, not due to increased accuracy or reliability. In fact the 4-20ma systems are more reliable and robust than their digital replacements.

 

But 4-20 mA system is just a data transfer system. It would likely have a AD and DA converter in the transducer end and then again an AD converter at the instrument end. As long as the converters are more accurate than the actual transducer its not an accuracy problem, but I can't really see what would be the benefit of it in boat instruments, which mostly have digital output to begin with. It would just add costs. Also the standard 4-20 mA current loop uses quite much power and each transducer needs to have their own loop.

 

My point is the same. The precision of the instruments is the limiting factor.

 

My view is that the limiting factor is how the environment effects the measurements. You are not able to measure wind not effected by the boat, water not effected by the boat and wind is not the same from water surface to mast top. These have a much bigger effect on the accuracy than the actual transducer error in a laboratory environment.

Share this post


Link to post
Share on other sites

 

Did you experiment with a custom inclinometer / heel meter for the B&G system? would a DIY work?

 

 

No, I don't have a B&G system and the inclinometer I make is not intended to be used with any instrument system. It is an USB-device for stability measurements.

Share this post


Link to post
Share on other sites

 

To Joakim: If I understand right, you have a hybrid system on your boat, where BS is water referenced (paddle wheel), but course (leeway+hdg) is ground referenced (COG from GPS). You would have quite peculiar TW figures with your system in 3 kn cross currents. This is not to mock your boat, you have done marvelously in races - just to show that you can do well also by GPS based (ground referenced) instruments in low current conditions like in Baltic. The errors caused by small currents are well compensated by accuracy of COG compared to low-end compass.

 

 

Yes, well almost. I'm using boat speed and true heading as a basis for TW. Boat speed comes from the paddle wheel and true heading from a 4 Hz GPS with some own damping through NMEA. Thus the instruments doesn't know it is COG, which it would only update at 1/2 Hz while true heading gets updated at maximum of 5 hz (if I remember correctly) in a Nexus system. I intended to upgrade the system with a HPC compass transducer, but it has worked so well without it that I haven't done it.

Share this post


Link to post
Share on other sites

On Quality of measurements: (accuracy, precision, linearity, frequency of measurements etc.)

 

I think we have missed one important thing in this quality of measurements (QM) thing. That is separation of measurements/instruments as such and the complexities of the phenomenon (wind) being measured/calculated. I think Heriberto has been pointing to this direction (and Joakim just posted the same as I am dealing with bellow)

 

 

1) The QM by instruments

 

These should be measured in standard conditions like wind tunnels. The QM (accuracy, precision, linearity etc ) of the instrument should come from these measurements. Of course you can talk about "QM in the field conditions", but that should not be mixed with the bellow.

 

 

2) The complexities of the phenomenon

 

I focus on wind, because it is the most complex. At least partially turbulent flow (professionals - bare with my terminology - and correct if necessary) is a complex thing with random phenomena of different scales. When the wind and sails meet the flow is quite varying in different parts of the sail. We have only one place to measure at - it is like judging a movie by one frame.

 

Modeling these complexities is the main challenge in the "calibration" of sail instruments - it is not the accuracy of instruments as such (1.) - although there could be made improvements also in angle and speed measurements as such.

 

What do I want to measure with wind instrument? To me the answer is: the orientation of the boat to the "undisturbed" wind flow/field, direction and speed of the wind. The problem is that I have no way of having my MHU in the "undisturbed flow" 50 m in front of the boat. I am doomed to disturbed flow and most of the calibration is modeling this difference and trying to calculate it away in order to have as good estimate of the undisturbed flow as possible. The ever changing nature of the wind does not make things any easier (I will come to this later as it has consequences in the frequency of measurements and damping).

 

Jorma

Share this post


Link to post
Share on other sites

On topic: Calculations moved to non-system computers - how to go that way? and

"Comparative wind calculations of different systems, Expedition, Nexus, NKE, B&G, etc:

 

 

I think this is very important topic for several reasons.

 

I have thought a lot of moving the calibrations and calculations from instrument system to Expedition and the showing the results on displays. Nick has advised against it for reasons that are only partially clear to me. People have also some problems transferring data to displays. Reliability of this system is the main weak point.

 

You would need only a simple/economical central unit with good sensors and fast bus. (as PhonyHRF has suggested). The calculations would be done in a powerful/low cost PC .

 

Expedition has all the calibrations of B&G (?) and more (like AWS*AWA, TWD-off set). There is no problem of bringing in the data necessary for calculations. The bottle neck is, when sending back the data for displays. Manufactures reserve this feature (remote/external channels) to their high-end (expensive) models (B&G Herc perf up, and Nexus NXR).

 

If this move to non-system computers would come true, tactical programs could be developed to have more advanced options for calculations (to Heriberto: even Bayesian!) as there is vast computing power available.

 

It would be very interesting to have calculation logics of different programs /instruments shown here. There was a post of WTP2 flow chart and PhonyHRF is building one of H3000. How about Expedition? The manual is not explicit about wind calculations/calibrations.

 

Nick, would it be possible to share with us the flow-chart/logic of wind calculations/calibrations in Expedition?

 

Jorma

Share this post


Link to post
Share on other sites

My view is that the limiting factor is how the environment effects the measurements. You are not able to measure wind not effected by the boat, water not effected by the boat and wind is not the same from water surface to mast top. These have a much bigger effect on the accuracy than the actual transducer error in a laboratory environment.

 

I don't entirely agree with your characterization of many things here, for example, 4-20ma systems are not "data transfer", they are complete analog up until the single AD box. The box reads current rather than voltage so that voltage drop over long cable lengths do not affect the measurement. An example would be a masthead unit on a long cable.... Neither here nor there, let's drop that.

 

 

B&G systems are analog transducers. As I said, I know nothing of Nexus, but if they say +- 2 degrees, you cannot improve the precision with added calibration points. Even at the calibration points themselves, the precision is +- 2 degrees. Add a hundred calibration points, the precision is still +- 2 degrees. That is the limit of how precise the instrument is. Just like a compass. "Accuracy" is another question.

 

As Jorma says, the best system would be one where you could read everything completely undisturbed from the boat. In fact, you would have readings at every level up to and above the masthead, and down below the waterline to below the total depth of the keel. Systems like that don't exist, yet, which can take both bulk and discrete measurement. That is why I like that B&G calls what they get "measured" wind. It isn't really the apparent wind, it is what you measure. After that, you add and subtract different values for upwash, heel angle, mass damping by the impellers and wind vane, leeway, on and on and on. I would prefer if they used the measured wind to come up with a matrix for what those other parameters were. This could include speed through water (impeller) and SOG (gps), etc., then add additional parameters to determine wind to bearing, wind to course, true wind, ground wind, etc..

Share this post


Link to post
Share on other sites

I don't entirely agree with your characterization of many things here, for example, 4-20ma systems are not "data transfer", they are complete analog up until the single AD box. The box reads current rather than voltage so that voltage drop over long cable lengths do not affect the measurement. An example would be a masthead unit on a long cable.... Neither here nor there, let's drop that.

 

B&G systems are analog transducers. As I said, I know nothing of Nexus, but if they say +- 2 degrees, you cannot improve the precision with added calibration points. Even at the calibration points themselves, the precision is +- 2 degrees. Add a hundred calibration points, the precision is still +- 2 degrees. That is the limit of how precise the instrument is. Just like a compass. "Accuracy" is another question.

 

As Jorma says, the best system would be one where you could read everything completely undisturbed from the boat. In fact, you would have readings at every level up to and above the masthead, and down below the waterline to below the total depth of the keel. Systems like that don't exist, yet, which can take both bulk and discrete measurement.

 

Is measuring wind speed and boat speed really analog in some system? I know that measuring wind angle is analog in some system, but never heard of analog speed measurement. I have never used a 4-20 mA system, but I have used some digital current loops. I have also understood that quite a few 4-20 mA transducers are more or less digital. E.g. in speed measurement you need to convert frequency to current.

 

If +-2 degrees is the reported accuracy of a angle measurement device, that means that at any angle you are not more than 2 degrees off the real angle and hopefully that even includes the whole operating range (temperature, speed, voltage etc.). This means that most of the time the system is far more accurate. Likely at the calibration points it is +- 0.5 degrees at calibration conditions. Just like making a deviation table for a compass the more denser you calibrate the more accurate it gets up to a limit of other errors like measuring the real heading at calibration points and the effect of temperature and heel to heading. Since the corrections at 45 degree intervals seem to be in +-0-4 degrees in Nexus MHU's, I believe that the +-2 degree accuracy specification mainly comes from the incomplete calibration. Say if you have a -2 degree correction at 0 degrees and then +1 at 45 degrees, it probably just linearizes to get -0.5 degrees at 22.5 degrees. But the real error at 22.5 degrees may be -2.5 - +1.5 to be inside the specs. Of course there are other sources of error as well.

 

When it comes to transfering the angle data from mast top to instruments not much accuracy is needed. 2 degrees of 360 is almost 1%. You can design many different analog or digital system that can achieve much better accuracy than that. I don't know how it is done in B&G, but as said Nexus uses PWM, which comes from two opto sensors that see a rotating ball with black and white pattern from two different angles. Vaisala uses 6 opto sensors to read a Grey coded angle, thus they have a resolution (and accuracy) of 6 bits = +- 2.8 degrees. Probably Vaisala transducers get more accurate with some averaging.

 

You can measure wind and water speed far away from boat using ultrasonic or laser doppler systems, but these are expensive and you need a lot of software development to use that data. Well actually the ultrasonic system is not that expensive, but it can only measure the speed towards the sensor. A very nice feature of ultrasonic doppler speed measurement is that you actually get a profile with a single measurement. You can get speed towards the sensor at any distance from the sensor (limited by power and noise). I have been involved on a project already in the mid 90's where speed profiles were measured with ultrasonic devices in water. You got the profile from very close to the sensor to about 1 m with a rather simple system.

 

There are probably no commercially available system to be easily used with any instrument system. But the technology has been there for at least 20 years.

Share this post


Link to post
Share on other sites

 

You would need only a simple/economical central unit with good sensors and fast bus. (as PhonyHRF has suggested). The calculations would be done in a powerful/low cost PC .

 

If this move to non-system computers would come true, tactical programs could be developed to have more advanced options for calculations (to Heriberto: even Bayesian!) as there is vast computing power available.

 

 

Any of the data shown on instruments today is based on very simple calculation. There is no need to use powerful PC's or other processors for such a simple tasks, that can be easily handled by a 10 dollar microcontroller, which is much more reliable and takes only 1 mA.

 

Another thing that is misleading is the different buses and their update rates. They may have been important 20-30 years ago, but today you can have any bus transfer much more than you can digest and even a well configured NMEA 0183 is good enough for most tasks. E.g. Nexus bus is just a 9600 baud binary RS-485, that is 10 000 times slower than the ethernet we use at our officies. There is not much data to be transfered (radar, sonar and chart images are an exception to this) and a few Hz update rate is more than enough.

 

I could rather easily make my own instrumentation using some transducers and doing the calculation within my own program and doing all the calibrations and corrections just as I like. Then you could e.g. use NMEA displays to show the data. I even know how to connect Nexus displays directly to any microcontroller or PC and how to output all the data as you like to them. I just don't think it is worth the effort doing that, since my system only costed about 2000 euros and delivers me what I expect from it. If I really needed to have much more accuracy and had the time to do all the calibration, then I would much rather do it completely in my own program, since that would give me access to everything and put out all the guessing what the external system does.

Share this post


Link to post
Share on other sites

Airmar has an ultrasonic boat speed sensors. I am considering taking the plunge, but they aren't cheap, about $700. They output as 5.5 hz pulse per knot and an update rate of 2 seconds. It will also optionally convert the data to NMEA 0183 if you prefer.

 

This would avoid the mass damping, fouling, friction and other errors of your impeller boat speed. Airmar also has an ultrasonic wind instrument for around $1000. It is pretty new on the market, and it is unclear how well it works in rain. It has three-axis accelerometers, and three-axis compass and yaw rate gyro and even a gps reciever to help calculate internally true wind speed and angle. It claims less than 2 degree accuracy on TRUE wind.

 

My opinion is this is the direction that all wind and boat speed sensors are going to go in coming years.

Share this post


Link to post
Share on other sites

Here is an account of using the PB200 Airmar wind instrument on a sailboat. Here is an exerpt:

 

For my testing, I used dual Raymarine ST60 wind instrument displays, one for the PB-200 and the other for the mechanical wind and vane sensor. The output of the PB-200 was used to feed my E-Series chartplotter and an S1G autopilot set in follow the wind mode.

 

When sailing upwind in stable winds under a PB200 enhanced autopilot, I found the autopilot was able to stay "in the groove" in the waves and wakes. During a period of 15 minutes that the AWA was closely observed, the PB200 showed variations of no more than 1 degree from my chosen AWA of 45 degrees as the autopilot aggressively followed the wind. Meanwhile, the mechanical driven display showed AWA variations as much as 12 degrees. Had those variations been fed to the autopilot, the sailboat would have zig-zagged a bit, the sail trim would have spent less time near optimal sheeting and the boat would have covered considerably less distance in the same time.

 

When sailing downwind with an asymmetric spinnaker, the autopilot also performed extremely well. While the PB200 showed AWA variations of 3 degrees as my boats autopilot responded well to staying 130 degrees off the wind, the mechanically driven display showed variations of as much as 35 degrees, especially as each 1-2 foot following wave caught up to us.

 

If you haven't found the excellent marine electronics blog Panbo, you should take a look.

Share this post


Link to post
Share on other sites

Airmar has an ultrasonic boat speed sensors. I am considering taking the plunge, but they aren't cheap, about $700.

 

This would avoid the mass damping, fouling, friction and other errors of your impeller boat speed. Airmar also has an ultrasonic wind instrument for around $1000. It is pretty new on the market, and it is unclear how well it works in rain. It has three-axis accelerometers, and three-axis compass and yaw rate gyro and even a gps reciever to help calculate internally true wind speed and angle. It claims less than 2 degree accuracy on TRUE wind.

 

 

These are quite different ultrasonic measuring systems that I was talking about. The speed sensors still measures quite close to hull and only in one spot. It has been available for some years and its been used by most instrument systems. I don't see any need for it, but then again fouling is not that bad around here. My boat speed is very accurate from well below 1 knot to at least 9 knots (my boat isn't faster than that expect for short surfing).

 

The ultrasonic MHU doesn't seem to be that accurate: http://www.airmartechnology.com/uploads/brochures/PB200.pdf

The 2 degreee error is RMS error and in wet condition (rain or even moisture) it becomes 8 degrees. Also the wind speed accuracy is 5 knots RMS in wet conditions.

 

What do they mean with RMS error anyway. Is that just the noise of the measurement or is it the RMS difference to real value? An RMS value means that the error can occasionally be about +-3 times the RMS value. Actually you can also see +-6 degrees in the wind angle error plot.

 

Note that this measures wind speed inside the unit. Thus is very different from the sensors I was talking about. They can measure far away from the mast. Since the sensors is quite big and heavy you need to have it close to mast top, thus in very bad environment for accurate measurements.

 

Vaisala has had a similar unit for many years with clearly better specifications: http://www.vaisala.com/Vaisala%20Documents/Brochures%20and%20Datasheets/MET-WCO-WMT700-datasheet-B210917EN-C-LoRes.pdf

Share this post


Link to post
Share on other sites

Started a new thread on quality control of calibration.

 

 

Jorma

Share this post


Link to post
Share on other sites

Nice picture Phony HRF!

 

I have about same questions. You have placed "heel correction" quite late. Why so ? It is an AWA error caused by geometry and should be corrected quite early in the chain.

 

Jorma

Share this post


Link to post
Share on other sites

Quick comment on sensors - I have had very good success with the Airmar Ultrasonic boat speed sensors (they work at low speeds, don't foul or get stuck, and seem to produce reliable and accurate data). I can't say the same for their wind sensors (sorry) - they are probably fine for a power boat, but in my experience they are not even close to being accurate enough for a grand prix level racing boat...

Share this post


Link to post
Share on other sites

Quick comment on sensors - I have had very good success with the Airmar Ultrasonic boat speed sensors (they work at low speeds, don't foul or get stuck, and seem to produce reliable and accurate data). I can't say the same for their wind sensors (sorry) - they are probably fine for a power boat, but in my experience they are not even close to being accurate enough for a grand prix level racing boat...

 

What instruments do you have?

 

H3K Motion, dual 5Hz DGPS with custom interpolation system

Share this post


Link to post
Share on other sites

Wow - a lot of questions to which answers are probably outside the scope of what I can talk about in a public forum. However, I can give you the general drift -- I started with two SimRad MX500 D/GPS control units each attached to their own MX421B-10 Smart Antennas setup in a BRIM/RAIM configuration. I then have been hacking the firmware of these units to make things "better". At present the output of my system is high-speed NMEA to Expedition, but I am thinking of switching to a LAN connection.

 

I started down this path when I became unhappy with the quality of GPS data I was receiving and the downstream effects of bad data. I have now come to the point that while I could improve my GPS system further, its accuracy and speed far exceed most anything else in my B&G system. My next project is to see if I can hack the system to create a highly accurate compass using two arbitrarily positioned DGPS sensors and thus eliminate or improve on my gyro...

Share this post


Link to post
Share on other sites

Error in measurement - Defining, measuring and dealing with

 

There are two kinds of error in measurements

 

1) Systematic error (connected to accuracy - often called validity (criterion validity) in social sciences)

 

In a simple situation there is always same amount of difference in the measurement when compared to true value or "golden standard". If this can be measured it is easy to correct by simple calculation (adding/subtracting). The unit is the same as the measurement ( like kn). The problem in our situation is to find the golden standard for comparison, especially with wind instrument. Systematic error can also be different in different parts of the scale (non-linearity).

 

Systematic error can be calibrated out, random error not.

2) Random error (connected precision - called reliability in social science)

 

Random error can be isolated when measuring a constant phenomenon. If the precision was high all the measured findigs would be the same. Usually this is not the case.

 

The method of expressing the amount of random error ( or precision) is something that is often omitted in specifications and also in our discussion. -+2 degrees in TWA - what does that mean? All observations will fall within this limit? 95%, 50% or what? The conclusions will be different.

 

RMSE (Root mean square error) has been used by some technical specifications and also in this thread. Definition here (ugly looking formula). If I understand right, it is about the same as standard deviation (SD) of observation in a steady state (measuring the same thing several times). I would suggest that we used SD when assessing precision (like when looking at our logs).

 

SD is easily measured from Expedition stripchart. Find a flat section in the measurement and SD can be easily acquired. If you want to improve your precision by taking a mean of readings, you have to go to Excel or some other program. SD in Expedition comes from raw numbers and damping has no effect.

 

Repeated measurements is the best remedy for random error as the errors will cancel each other out partially. Taking a mean over 10 measurement points will make a more precise assessment, especially if the phenomenon stays stable that time. Increase of precision by adding measurements will tapper of (logarithmic). 10-20 points is OK. If frequency of measurement and logging is high, this is easier as the phenomenon has no time to change. I think the only justification for going to higher (over 5 Hz) frequencies of measurement and log update rate is to increase precision.

 

What would I like to know from instrument specification about accuracy and precision. Lets take boat speed as it has applicable "golden standard" (SOG and laboratory measurements). I would like to know its accuracy expressed as distance from "true value" over whole measuring range (a line plot) where also the SD (precision) would be expressed. The same goes for specification of my boats BS quality. If someone prefers percentage, he can calculate it as he wishes. Such a plot expresses the amount of error much more precisely than one percentage.

 

Jorma

Share this post


Link to post
Share on other sites

Mistaken and PhonyHRF,

 

you have interesting information on Airmar ultrasonic wind instrument and high-tech GPS.

 

Could you give some more precise information, what is gained/lost with them.

 

Like with wind instrument: can you estimate the precision of AWA and AWS (raw AWA, not back calculated). Find a steady state and report time interval, mean, SD, frequency of update (in log) and sea state if known. In that way I can compare to mine.

 

On GPS same thing on SOG and COG. How does it compare to the standard thing?

 

I would be very interested in gains you have achieved with your improved instruments. I can give my numbers for comparison.

 

 

Jorma

Share this post


Link to post
Share on other sites

i have the airmar in my boat and LOVE it. I am looking how can i connect the airmar wind into my b&G and replace the wind vane.

 

PhonyHRF,

 

So I misunderstood these comments? You have the airmar ultrasonic BS-instrument? No wind instrument?

 

Jorma

Share this post


Link to post
Share on other sites

I will send a PM to you.

 

Which forum would you recommend in GPS Discussion for racing sailors interested in GPS?

 

Jorma

Share this post


Link to post
Share on other sites