Electronic Foil Control Systems.

[BalticBandit]

I have managed to get afloat again after a month of poor weather and other commitments.Key change this time has been the addition of a data logging capability and the ability to send the servo to max to generate a disturbance to evaluate the control.

This data should allow me to now start optimizing the control.

First impressions are that the servo is responding far too much to chop but not actually managing to control the height that well.

I need to separate chop from height changes without adding a lag to the control loop.

The GPS logged 10 NM of sailing during an hour afloat with top speed of 17.1kts I did not push the speed as the control looked like it would become unstable at higher speeds.

View attachment 87155

Trying to read the data plot and wondering if I'm getting it right.

Servo drives to Max UP, followed by a rapid flap move to almost the opposite extreme and then slowly with chop driven oscilations stabilizes

Nose pitches up, and then slowly comes back down

All the time the ride hight is dropping slowly and then gently comes back up?

Or am I misreading the data?

 

[clivee1]

The servo drives to max lifting the transom and causing the bow to pitch down, A button on the tiller extension starts the data log and drives the servo to max.

Remember that the servo is exclusively driving the rudder foil.

The gyro shows max rate of rotation at ~0.4 seconds.

The bow drops from ~3.5 degrees bow up to 1 degree bow down at 0.6 seconds, and then recovers to ~7.5 degrees bow up.

The height drops from ~800mm to ~300mm and then starts to recover.

The servo drive shows a lot of high frequency movement induced by chop effecting the height reading, and the phase advance software in the control system.

 

[Munter]

Great post Clive. The ability to capture and analyse data should help refine the system far faster than is possible with current techniques. I wonder if the top moth designers have considered some form of data capture to help them develop mechnical systems? I think it would be a great advantage to really know what is happening with the control surfaces rather than just incidentally observing them while sailing.

 

[atypicalguy]

Great post Clive. The ability to capture and analyse data should help refine the system far faster than is possible with current techniques. I wonder if the top moth designers have considered some form of data capture to help them develop mechnical systems? I think it would be a great advantage to really know what is happening with the control surfaces rather than just incidentally observing them while sailing.

Yes I could use an accelerometer about now to get a handle on pitch, even if it were not linked to the control system in any way (mine is all mechanical). It's interesting that Clive's boat recovers to 7.5 degrees bow up - that is a lot of bow up to deal with.

 

[clivee1]

I got to sail again this weekend with a vertical accelerometer added to the electronics so that short term height measurements could be derived from the vertical accelerations and only tied to the wand over a couple of seconds. This is to allow the control to ignore chop but respond quickly to real changes in flying height.

Previous tests had shown that a vertical accelerometer would need compensation for pitch and heel.

12' of heel gives a height error of 430mm after just 2 seconds, so I added heel measurement using a horizontal transverse accelerometer.

The boat flew well in light conditions and I managed to complete a club race and finish well ahead of the rest of the fast handicap (not asymmetric) fleet.

The height control felt that it would become unstable at higher speeds. The top speed I recorded on the GPS was 17kts.

Subsequent review of the data log showed significant fluctuations in the inertially measured height at 1 – 2 Hz frequency that had no correlation to the height as measured by the wand.

I had expected the inertially measured height to be flat with higher frequency fluctuations on the wand measurement as it bounced over the chop.

I think that changes in coarse as I steer for balance are causing lateral accelerations that are giving false heel readings and consequently feeding through to the vertical acceleration.

The 1 -2 Hz frequency is consistent with the rate at which one steers for balance and adjusts the main sheet.

Running through the numbers at 15kts and 12' of heel a 12' change in course over 2 seconds will produce a height error of 400mm. This will become more significant as the speed increases.

I think that for the vertical accelerometer to work so that I can fly smoothly through chop, I need heel compensation and that this will require a roll axis gyro as well as a transverse accelerometer so that heel can be separated from course changes.

Clive

 

[Major Tom]

I got to sail again this weekend with a vertical accelerometer added to the electronics so that short term height measurements could be derived from the vertical accelerations and only tied to the wand over a couple of seconds. This is to allow the control to ignore chop but respond quickly to real changes in flying height.
Previous tests had shown that a vertical accelerometer would need compensation for pitch and heel.

12' of heel gives a height error of 430mm after just 2 seconds, so I added heel measurement using a horizontal transverse accelerometer.

The boat flew well in light conditions and I managed to complete a club race and finish well ahead of the rest of the fast handicap (not asymmetric) fleet.

The height control felt that it would become unstable at higher speeds. The top speed I recorded on the GPS was 17kts.

Subsequent review of the data log showed significant fluctuations in the inertially measured height at 1 – 2 Hz frequency that had no correlation to the height as measured by the wand.

I had expected the inertially measured height to be flat with higher frequency fluctuations on the wand measurement as it bounced over the chop.

I think that changes in coarse as I steer for balance are causing lateral accelerations that are giving false heel readings and consequently feeding through to the vertical acceleration.

The 1 -2 Hz frequency is consistent with the rate at which one steers for balance and adjusts the main sheet.

Running through the numbers at 15kts and 12' of heel a 12' change in course over 2 seconds will produce a height error of 400mm. This will become more significant as the speed increases.

I think that for the vertical accelerometer to work so that I can fly smoothly through chop, I need heel compensation and that this will require a roll axis gyro as well as a transverse accelerometer so that heel can be separated from course changes.

Clive

Reading through all this it looks as we may have under estimated how ahead of it's time a bow mounted wand is!

 

[atypicalguy]

I got to sail again this weekend with a vertical accelerometer added to the electronics so that short term height measurements could be derived from the vertical accelerations and only tied to the wand over a couple of seconds. This is to allow the control to ignore chop but respond quickly to real changes in flying height.
Previous tests had shown that a vertical accelerometer would need compensation for pitch and heel.

12' of heel gives a height error of 430mm after just 2 seconds, so I added heel measurement using a horizontal transverse accelerometer.

The boat flew well in light conditions and I managed to complete a club race and finish well ahead of the rest of the fast handicap (not asymmetric) fleet.

The height control felt that it would become unstable at higher speeds. The top speed I recorded on the GPS was 17kts.

Subsequent review of the data log showed significant fluctuations in the inertially measured height at 1 – 2 Hz frequency that had no correlation to the height as measured by the wand.

I had expected the inertially measured height to be flat with higher frequency fluctuations on the wand measurement as it bounced over the chop.

I think that changes in coarse as I steer for balance are causing lateral accelerations that are giving false heel readings and consequently feeding through to the vertical acceleration.

The 1 -2 Hz frequency is consistent with the rate at which one steers for balance and adjusts the main sheet.

Running through the numbers at 15kts and 12' of heel a 12' change in course over 2 seconds will produce a height error of 400mm. This will become more significant as the speed increases.

I think that for the vertical accelerometer to work so that I can fly smoothly through chop, I need heel compensation and that this will require a roll axis gyro as well as a transverse accelerometer so that heel can be separated from course changes.

Clive

who sells gyros like that and how much do they cost?

 

[SimonN]

Clive

Thanks for the update on a really inetersting project. I can understand why you are doing it and it is certainly an interesting challenge. However, do you think that you are going to end up with something that is actually significantly enough better than the current wand systems? Or is part of the purpose of the project the actual technical challenge itself.

 

[clivee1]

who sells gyros like that and how much do they cost?

Analog devices:

~£50 1 off.

The device is 7mm x 7mm x 3mm

ADXRS300_gyro.pdf

 

[clivee1]

Clive
Thanks for the update on a really inetersting project. I can understand why you are doing it and it is certainly an interesting challenge. However, do you think that you are going to end up with something that is actually significantly enough better than the current wand systems? Or is part of the purpose of the project the actual technical challenge itself.

I think that ultimately it could significantly out perform a mechanical system.

I do not know if I will single-handedly manage to get that far, and the technical challenge is in itself worth while.

Key advantages are:

It allows significant fairing as there is no moving parts on the main foil.

It can be more intelligent during and before take off.

it can provide controlled damping. A simple proportional system will be prone to oscillations.

It can provide a reactive pitch torque to counter gusts. At the moment Moths steer some pretty wild courses downwind to maintain control.

It can separate ride height changes from chop.

It can optimize itself for upwind / downwind / marginal / full on conditions.

In all similar applications electronic solutions have ultimately become better and cheaper than the mechanical alternative.

Clive

 

[Basiliscus]

The inputs to my electronics are the wand, A solid state gyro mounted in the pitch plane which measures the rate of pitch rotation, an accelerometer mounted in the pitch plane that will respond to changes in pitch and acceleration of the boat. This is used to provide a low frequency reference to the gyro. as it is safe to assume that the long term acceleration of the boat is zero, and a vertical accelerometer.

It would be best to use the accelerometer for high-frequency height control and use the wand for low-frequency height control. The accelerometer will be subject to bias, scale factor, and alignment errors that will result in unbounded (ie, very, very large) errors if the acceleration is integrated to get vertical velocity and height. However feeding back wand position to the height estimate will stabilize the estimate against these errors. At the same time, the accelerometer will respond faster to a change in height than will the wand, making it most suitable for adding lead to the height control.

I have 2 nested control loops. I drive the rudder foil servo to try and achieve a target pitch based on the pitch measurement that comes from a hardware integration of the gyro out put and the pitch plane accelerometer, and the rate of change of pitch that comes directly from the gyro.The target pitch is set by the height control loop as a function of the measured height from the wand and a software integration of the height.

I want to use the vertical accelerometer to measure the vertical velocity and acceleration to provide phase advance to the height control loop. I also want to double integrate the acceleration to generate a second height reading that will allow me to separate the vertical motion of the boat from the contours of the waves.

My initial attempts to do this have not worked as small changes in heel and pitch change the measured vertical acceleration and the double integration rapidly produces significant height errors. Whilst I can correct for the error due to pitch I need to measure heel to compensate for that, This is my next step.

Both control loops have non linear gain functions so that when running close to target setup they do not respond significantly to small deviations however the responce gets much more agressive if the boat is seriously out of shape.

I suggest three nested loops. The inner most loop would use acceleration feedback. The middle loop would use pitch rate feedback, and the outer loop would use wand feedback. Each loop should command a rate of change in the controlled variable proportional to the error in the controlled variable. For example, the controlled variable for the outermost loop is height. The difference between the desired height and the height measured by the wand is the height error. The outer loop should multiply the height erorr by a gain to calculate a vertical velocity command. The next loop would then respond to the vertical velocity command. The vertical velocity loop would compute the vertical velocity error, then multiply the vertical velocity erorr by a gain to form a vertical acceleration command. The innermost loop would use the acceleration feedback to form the acceleration error. The vertical acceleration control would then be comanded by a gain times the acceleration error. Unfortunately, this approach would be best suited to direct lift control from a main flap than by control by the tail, but the principle would still be the same. Form the erorr, use the error to comand the rate of change.

Pitch attitude corresponds to height rate because at a constant speed the flightpath angle is proportional to the vertical velocity divided by the speed. So both the pitch rate gyro and the accelerometer are somewhat related to the second derivative of the height. But there is a difference because the lift on the foil is also dependent on the angle of attack, which is affected by the pitch attitude.

Another way to go would be to use two nested loops, with the outer loop being a height loop, as above. The inner loop would be a combination of accelerometer and pitch rate gyro feedback to form the controlled variable.

What you really need is a dynamic model that allows you to express the relationship between the motion of the boat and the values measured by each of the sensors, and the effectiveness of your control surface. Then you can estimate the quantities you want to control and determine how to move the surface. Check out NACA TR-918.

Regarding the question as to whether a second servo on the main foil flap would be beneficial:Undoubtedly the height control could be improved and in no way would it be diminished. However in the horizontal plane we do not need a rudder and a centreboard trim tab to steer our boats so why should we in the vertical plane.

If you were trying to achieve precise positioning in the horizontal plane, you'd want direct side force control then, too.

There are really two coupled modes to the boat's motion you're trying to control - a pitch mode and a heave mode. With one control, you can only get at one mode by the way in which it is coupled to the other mode. With two controls, you can control each mode directly. And with the additional control you can satisfy an additional objective, such as minimizing drag.

The control of height using tail control will always have a lag compared to controlling height with the main foil flap. The reason is to change height, you need to change the vertical velocity, and to change the vertical velocity you need to change the vertical acceleration by changing the lift on the foil. If you have a flap on the foil, when you change the flap deflection, you change the lift and create an immediate change to the vertical acceleration. If you change the lift through the changing the pitch attitude of the boat to control the angle of attack of the foil, then you first have to generate a pitch acceleration to generate a pitch rate to generate a change in pitch attitude. Only once the pitch attitude changes do you change the lift on the foil and change the vertical acceleration and start to affect the height.

But it's worse than that. With tail control, as opposed to a canard, the tail has to drop to increase the angle of attack. So the boat actually has to descend slightly before it can start to rise. This has significant implications for the stability of control loops.

It's simply not possible to get as tight a height control using only the rudder foil compared to controlling only the main foil flap. The lags in the height response to pitch control mean that the boat will go unstable if the height control is cranked up too high.

However, height control using the flap is also subject to serious limitations. The flap only has limited authority. The lift on the foil is a function not only of angle of attack but of speed. Since the lift has to equal the weight, as the speed increases, the flap has to be deflected to offset the increase in lift due to speed. Eventually the flap will be fully saturated just trying to maintain the trim of the boat, and even before then, any deflection for trim takes away from the authority available to counter dynamic disturbances. Tail control doesn't have quite as much limitation in this regard. There's not much difference in angle of attack required to climb or descend at a constant angle, so once the pitch attitude is set, the tail control can return to near its trim value as the height changes. So over the long term, the tail can be much more powerful than the flap.

The combination of characteristics means the flap and tail controls can be used in complementary ways. The flap can null out high-frequency disturbances from small waves and the tail can adjust the pitch trim so that over the long term the flap deflection returns to neutral. This maximizes the flap's effectiveness for dynamic ride control, and the steady state flap deflection can correspond to the optimum angle for minimizing foil profile drag.

Most importantly however: I found last year sailing with a mechanical system that I seemed to be hitting a hard wall regarding max speed. Contrary to all the text books the drag appears to be going up far faster than V^2.

Pitch control is very important for hydrofoil peformance, especially if you are using surface piercing foils or if the flap doesn't span the full width of the forward foil. At constant lift, the induced drag should decrease with the square of the speed - this is the whole point of using hydrofoils at high speed. But if the effective span shrinks with speed, the induced drag can actually go up. Pitch control can be important for flying lower and increasing the effective span to reduce induced drag more than the increase in parasite drag due to extra wetted area.

I have an unsubstantiated theory that around 18Kts we are starting to see true cavitation around unfair parts of the structure and that the size of the cavitation bubbles grows rapidly causing the rapid increase in drag.

That's quite possible. At areas where the local flow velocity is increased due to interference, you can get cavitation at 18 kt. You can also start to get cavitation if you have sharp pressure peak, such as at a corner.

 

[aus_stevo]

That's quite possible. At areas where the local flow velocity is increased due to interference, you can get cavitation at 18 kt. You can also start to get cavitation if you have sharp pressure peak, such as at a corner.

is this why bladerider , mach2 and the new fastacraft foils all have the bulbous fairing at the t-joint junction? rather than just for structure

 

[ferrero]

Clive, very interesting design and congratulations on some terrific development work.I'm most interested in your opinion of the various altitude sensors that could be used in this application-ultrasound ,micro radar etc.

Thanks for having the guts, courage and determination to make such a trmendous contribution to foiler development!

Over thinking it, go with a 3 axis accelerometer and a contact switch. You know when you are on the surface, integrate the acceleration twice and you have height off the water. It should be a lot less expensive than either other option, plus you won't have to filter out the waves.

Interested to see what his solution was.

You don't get the accuracy from an accelerometer unless you spend 10 grand pluss on the unit and have some very fancy filters. Integrating the output twice would basically give you garbage.

 

[ferrero]

We've already been through the rule 52 shenanigans when foils were first brought into mothing - the forward motion of the boat that moves the wand and drives the flap is interpreted as 'manual power' ie. it's not a stored energy device, so the existing mechanical wand devices clear the rule.
Electronics OTOH require stored energy of some description (unless it's run by solar or wind power and even then I suspect that the device would need to have no capacitance in the circuit as a capacitor is technically a means of storing energy...) and hence does violate R52. For that matter most of the modern canting keel maxis also violate R52 as they all use engine power to cant the keel... how do they get around it?

If it is just stored energy that is the problem does that mean you are'nt allowed shock chord?! How about having a wind up power system with little crank arm you wind up before you get to the start? Same deal as stretching a bit of elastic, you are just using the energy in a cleverer way.

 

[the_skiff_devil]

We've been through the rule 52 issue already. Everyone uses shock cord, so it'd be very hard for someone to protest on 52 for it.

 

[Kenny Dumas]

How 'bout some really small pressure sensors on the back of the foil (to avoid the Bernoulli pressures) to sense depth / elevation?

http://www.pmctransducers.com/tranducers.html

Gotta be some really cheap ones out there from the automotive world. Figure ~1.5 psi at 3 ft. depth with typical 1% accuracy and easy to use since they're just resistor bridge strain gages with a DC output. Should be fast enough, just depends on the DAC.

 

[atypicalguy]

That's quite possible. At areas where the local flow velocity is increased due to interference, you can get cavitation at 18 kt. You can also start to get cavitation if you have sharp pressure peak, such as at a corner.

is this why bladerider , mach2 and the new fastacraft foils all have the bulbous fairing at the t-joint junction? rather than just for structure

I have no insider knowledge Astevo but I think they have them because they are all two part foils and there isn't enough depth to make the connection robust otherwise. You need more depth/material locally at the T and a bulbous fairing is the simplest way to achieve that. And it is a proven design now.

But if you build single piece foils I'd say one is better off without it.

 

[Mal Smith]

That's quite possible. At areas where the local flow velocity is increased due to interference, you can get cavitation at 18 kt. You can also start to get cavitation if you have sharp pressure peak, such as at a corner.

is this why bladerider , mach2 and the new fastacraft foils all have the bulbous fairing at the t-joint junction? rather than just for structure

I have no insider knowledge Astevo but I think they have them because they are all two part foils and there isn't enough depth to make the connection robust otherwise. You need more depth/material locally at the T and a bulbous fairing is the simplest way to achieve that. And it is a proven design now.

But if you build single piece foils I'd say one is better off without it.

It may have only been done on moths for structural reasons, but bulbs have been used to control cavitaion at the foil root e.g on the Boeing Jetfoil (there is a paper on it somewhere). So maybe it serves both purposes.

 

[ferrero]

That's quite possible. At areas where the local flow velocity is increased due to interference, you can get cavitation at 18 kt. You can also start to get cavitation if you have sharp pressure peak, such as at a corner.

is this why bladerider , mach2 and the new fastacraft foils all have the bulbous fairing at the t-joint junction? rather than just for structure

I have no insider knowledge Astevo but I think they have them because they are all two part foils and there isn't enough depth to make the connection robust otherwise. You need more depth/material locally at the T and a bulbous fairing is the simplest way to achieve that. And it is a proven design now.

But if you build single piece foils I'd say one is better off without it.

It may have only been done on moths for structural reasons, but bulbs have been used to control cavitaion at the foil root e.g on the Boeing Jetfoil (there is a paper on it somewhere). So maybe it serves both purposes.

The cavitation number on flow over a moth foil is too low for cavitation to be an issue. You are talking around 40 knots before it is an issue. The bulb helps structure but can also help reduce junction drag if you do it right.

 

[ebnelson]

The inputs to my electronics are the wand, A solid state gyro mounted in the pitch plane which measures the rate of pitch rotation, an accelerometer mounted in the pitch plane that will respond to changes in pitch and acceleration of the boat. This is used to provide a low frequency reference to the gyro. as it is safe to assume that the long term acceleration of the boat is zero, and a vertical accelerometer.

It would be best to use the accelerometer for high-frequency height control and use the wand for low-frequency height control. The accelerometer will be subject to bias, scale factor, and alignment errors that will result in unbounded (ie, very, very large) errors if the acceleration is integrated to get vertical velocity and height. However feeding back wand position to the height estimate will stabilize the estimate against these errors. At the same time, the accelerometer will respond faster to a change in height than will the wand, making it most suitable for adding lead to the height control.

I have 2 nested control loops. I drive the rudder foil servo to try and achieve a target pitch based on the pitch measurement that comes from a hardware integration of the gyro out put and the pitch plane accelerometer, and the rate of change of pitch that comes directly from the gyro.The target pitch is set by the height control loop as a function of the measured height from the wand and a software integration of the height.

I want to use the vertical accelerometer to measure the vertical velocity and acceleration to provide phase advance to the height control loop. I also want to double integrate the acceleration to generate a second height reading that will allow me to separate the vertical motion of the boat from the contours of the waves.

My initial attempts to do this have not worked as small changes in heel and pitch change the measured vertical acceleration and the double integration rapidly produces significant height errors. Whilst I can correct for the error due to pitch I need to measure heel to compensate for that, This is my next step.

Both control loops have non linear gain functions so that when running close to target setup they do not respond significantly to small deviations however the responce gets much more agressive if the boat is seriously out of shape.

I suggest three nested loops. The inner most loop would use acceleration feedback. The middle loop would use pitch rate feedback, and the outer loop would use wand feedback. Each loop should command a rate of change in the controlled variable proportional to the error in the controlled variable. For example, the controlled variable for the outermost loop is height. The difference between the desired height and the height measured by the wand is the height error. The outer loop should multiply the height erorr by a gain to calculate a vertical velocity command. The next loop would then respond to the vertical velocity command. The vertical velocity loop would compute the vertical velocity error, then multiply the vertical velocity erorr by a gain to form a vertical acceleration command. The innermost loop would use the acceleration feedback to form the acceleration error. The vertical acceleration control would then be comanded by a gain times the acceleration error. Unfortunately, this approach would be best suited to direct lift control from a main flap than by control by the tail, but the principle would still be the same. Form the erorr, use the error to comand the rate of change.

Pitch attitude corresponds to height rate because at a constant speed the flightpath angle is proportional to the vertical velocity divided by the speed. So both the pitch rate gyro and the accelerometer are somewhat related to the second derivative of the height. But there is a difference because the lift on the foil is also dependent on the angle of attack, which is affected by the pitch attitude.

Another way to go would be to use two nested loops, with the outer loop being a height loop, as above. The inner loop would be a combination of accelerometer and pitch rate gyro feedback to form the controlled variable.

What you really need is a dynamic model that allows you to express the relationship between the motion of the boat and the values measured by each of the sensors, and the effectiveness of your control surface. Then you can estimate the quantities you want to control and determine how to move the surface. Check out NACA TR-918.

Regarding the question as to whether a second servo on the main foil flap would be beneficial:Undoubtedly the height control could be improved and in no way would it be diminished. However in the horizontal plane we do not need a rudder and a centreboard trim tab to steer our boats so why should we in the vertical plane.

If you were trying to achieve precise positioning in the horizontal plane, you'd want direct side force control then, too.

There are really two coupled modes to the boat's motion you're trying to control - a pitch mode and a heave mode. With one control, you can only get at one mode by the way in which it is coupled to the other mode. With two controls, you can control each mode directly. And with the additional control you can satisfy an additional objective, such as minimizing drag.

The control of height using tail control will always have a lag compared to controlling height with the main foil flap. The reason is to change height, you need to change the vertical velocity, and to change the vertical velocity you need to change the vertical acceleration by changing the lift on the foil. If you have a flap on the foil, when you change the flap deflection, you change the lift and create an immediate change to the vertical acceleration. If you change the lift through the changing the pitch attitude of the boat to control the angle of attack of the foil, then you first have to generate a pitch acceleration to generate a pitch rate to generate a change in pitch attitude. Only once the pitch attitude changes do you change the lift on the foil and change the vertical acceleration and start to affect the height.

But it's worse than that. With tail control, as opposed to a canard, the tail has to drop to increase the angle of attack. So the boat actually has to descend slightly before it can start to rise. This has significant implications for the stability of control loops.

It's simply not possible to get as tight a height control using only the rudder foil compared to controlling only the main foil flap. The lags in the height response to pitch control mean that the boat will go unstable if the height control is cranked up too high.

However, height control using the flap is also subject to serious limitations. The flap only has limited authority. The lift on the foil is a function not only of angle of attack but of speed. Since the lift has to equal the weight, as the speed increases, the flap has to be deflected to offset the increase in lift due to speed. Eventually the flap will be fully saturated just trying to maintain the trim of the boat, and even before then, any deflection for trim takes away from the authority available to counter dynamic disturbances. Tail control doesn't have quite as much limitation in this regard. There's not much difference in angle of attack required to climb or descend at a constant angle, so once the pitch attitude is set, the tail control can return to near its trim value as the height changes. So over the long term, the tail can be much more powerful than the flap.

The combination of characteristics means the flap and tail controls can be used in complementary ways. The flap can null out high-frequency disturbances from small waves and the tail can adjust the pitch trim so that over the long term the flap deflection returns to neutral. This maximizes the flap's effectiveness for dynamic ride control, and the steady state flap deflection can correspond to the optimum angle for minimizing foil profile drag.

Most importantly however: I found last year sailing with a mechanical system that I seemed to be hitting a hard wall regarding max speed. Contrary to all the text books the drag appears to be going up far faster than V^2.

Pitch control is very important for hydrofoil peformance, especially if you are using surface piercing foils or if the flap doesn't span the full width of the forward foil. At constant lift, the induced drag should decrease with the square of the speed - this is the whole point of using hydrofoils at high speed. But if the effective span shrinks with speed, the induced drag can actually go up. Pitch control can be important for flying lower and increasing the effective span to reduce induced drag more than the increase in parasite drag due to extra wetted area.

I have an unsubstantiated theory that around 18Kts we are starting to see true cavitation around unfair parts of the structure and that the size of the cavitation bubbles grows rapidly causing the rapid increase in drag.

That's quite possible. At areas where the local flow velocity is increased due to interference, you can get cavitation at 18 kt. You can also start to get cavitation if you have sharp pressure peak, such as at a corner.

How about using something like a capacitive fuel gauge for measuring ride height like aircraft use, http://www.airstuff.com/fuelmon.html#section1 The sensor tube could be fed down the dagger board and is long enough for the entire length of the blade, I think. If the inside of the dagger board filled and drained quickly enough with height changes then height could be sensed without too much delay. This in pair with the accelerometer and gyro seems tunable at first glance.