Dynon AP pitch issue

I'm at my wit's end. When I first flew my airplane last year it left the ground with a Trio EzPilot installed. For whatever reason I was never able to get the Trio to work right. In spite of everything I tried, the thing would just not track smoothly. I was using the SafeAir1 mounting kit with the servo mounted in the last bay of the right wing and a long pushrod connected from the servo to the aileron bellcrank. Anyway, I tried everything -- new servo, new head, wiring, etc. I have detailed this issue on other theads here on VAF, but I discuss it here only for back story, nothing else. I don't blame Trio for this issue, they tried to help, but there's only so much they can do remotely and so many dollars in avgas I can spend before I just gave up. I just sat down and added up the cost for me. It was > $500.00 in avgas diagnosing that Trio problem and I never came up with a solution. That $500.00 figure only includes flights specifically made to try new things with the Trio.

Well, as mentioned I gave up. When Dynon announced their autopilot I thought I would give it a try. I liked their ideas with their AP and I've been a happy customer with their EFIS and EMS in my panel.

I installed their autopilot (two-axis). The roll axis is perfect. It tracks extremely well. The only thing I don't understand is why when you tell it to fly a GPS ground track (i.e you set a track on the Dynon), it overshoots it every time. It doesn't make much sense to me. I also have an issue with the Dynon occasionally barking at me about losing the GPS signal and defaulting to magnetic heading mode. I never lose the HSI and if I go to the AP menu I can re-enable the NAV mode (so it tracks the GPS) with no problem. I've only done preliminary testing with this issue so for me this is lower priority.

The other thing (this is my big problem) is that the Dynon altitude hold is behaving badly. Tweaking the sensitivity settings, the best I can get the altitude hold to perform is +- 20 feet. I engage the altitude hold with the airplane trimmed and it will hold for about one minute or so and then begin an oscillation. The oscillation will grow larger and larger until it maxes at +-20 feet in altitude, but about +- 600fpm in rate. This is with the Dynon sensitivity setting set to the maximum value (25).

I pressure checked my static system and found no leaks.

I vented the Dynon static system to cabin air to ensure that I didn't have any fluctuations in my static system.

I flew the plane from 110kts to 170kts in increments of 10 knots, testing each speed to see if the pitch servo performed better or worse depending upon speed. The speed made little difference.

I pulled the baggage wall and noticed a little play in the mount. In other words, if I engaged the servo, then pushed/pulled on the pushrod until the servo slipped, I would see a little left/right (perpendicular to arm throw) flexing of the mount at the bottom. According to Dynon, the little flange on the bottom of the mount is optional. It's intended to be riveted to the floor. My plane is painted so I wanted to avoid putting drilling through the outter skins and risking chipping paint, etc, so I cut the flange off. In order to stop the flexing I put another fastener in the bottom of the mount. Now the mount is very rigid.

The Dynon support folks are fantastic and are helping me with this issue. Of course I understand that it is a new product and it will suffer growing pains. I fully expect to have some issues, but I'm truly at my wit's end and I'm beginning to believe that it was just never meant for me to have a functional autopilot in my airplane.

I spent all day today on this issue and I finished the day 20 gallons of avgas less, 8 hours without seeing my son and still no answer.

I basically wanted to post here at VAF to see if anyone has any ideas of what could possibly be causing this issue. Also, I know that there are a few Dynon AP beta testers here on VAF so I'm looking to them for ideas.

So, VAF...what say you. Where should I go from here?

Jamie said:

I installed their autopilot (two-axis). The roll axis is perfect. It tracks extremely well. The only thing I don't understand is why when you tell it to fly a GPS course (i.e you set a course on the Dynon), it overshoots it every time.

Click to expand...

Jamie,

On this issue are you talking about it overflying waypoints when you are going from one waypoint to another and then another or are you talking about it zigzagging one side or the other of a straight course line?

Unless you have a GPS connected that supports GPS steering over ARINC, the GPS will overfly waypoints and then turn back to the course line but I am not sure this is what your talking about.

If your talking about it overshooting when joining a course, maybe you can increase the gain a little at a time to compensate.

Brantel said:

Jamie,

On this issue are you talking about it overflying waypoints when you are going from one waypoint to another and then another or are you talking about it zigzagging one side or the other of a straight course line?

Unless you have a GPS connected that supports GPS steering over ARINC, the GPS will overfly waypoints and then turn back to the course line but I am not sure this is what your talking about.

If your talking about it overshooting when joining a course, maybe you can increase the gain a little at a time to compensate.

Click to expand...

Nope, I've very well aware of GPSS, but I mis-spoke (mis-typed?) my original post when I said course instead of track. The dynon overshoots the ground track. So if you're flying a ground track of 090 and you tell it to turn to 010, it will overshoot 010. That's what doesn't make sense to me.

The Dynon doesn't have gain...it has 'sensitivity', vertical speed limit, min ias and max ias, and servo torque % and that's it.

I've tried every sensitivity setting from 1-25.

Jamie said:

... I also have an issue with the Dynon occasionally barking at me about losing the GPS signal and defaulting to magnetic heading mode. I never lose the HSI and if I go to the AP menu I can re-enable the NAV mode (so it tracks the GPS) with no problem. I've only done preliminary testing with this issue so for me this is lower priority.

Click to expand...

Jamie,

I'll answer this as I was one of the beta testers who found this problem.

This has nothing to do with the Dynon, the Garmin handheld GPS's (and maybe some others) stops putting out a course signal every now and then. When this happens the AP has to switch to heading mode and it is good that the Dynon lets you know this happens, IMHO.

Jamie said:

The other thing (this is my big problem) is that the Dynon altitude hold is behaving badly....

I pulled the baggage wall and noticed a little play in the mount. In other words, if I engaged the servo, then pushed/pulled on the pushrod until the servo slipped, I would see a little left/right (perpendicular to arm throw) flexing of the mount at the bottom. According to Dynon, the little flange on the bottom of the mount is optional. It's intended to be riveted to the floor. My plane is painted so I wanted to avoid putting drilling through the outter skins and risking chipping paint, etc, so I cut the flange off. In order to stop the flexing I put another fastener in the bottom of the mount. Now the mount is very rigid....

So, VAF...what say you. Where should I go from here?

Click to expand...

Did you get a chance to fly after securing the pitch servo bracket? Your post wasn't clear to me. It has to be very solid to keep the servo from moving and hunting altitude.

Also, how bumpy was it? This has a real impact on how well this or any autopilot performs.

PS. Check your PM's

Jamie said:

Nope, I've very well aware of GPSS, but I mis-spoke (mis-typed?) my original post when I said course instead of track. The dynon overshoots the ground track. So if you're flying a ground track of 090 and you tell it to turn to 010, it will overshoot 010. That's what doesn't make sense to me.

The Dynon doesn't have gain...it has 'sensitivity', vertical speed limit, min ias and max ias, and servo torque % and that's it.

I've tried every sensitivity setting from 1-25.

Click to expand...

Does it always overshoot the same amount in either direction?

Does it ever come back to the desired track? If so, does it do it quickly or does it wonder around and eventually get there?

If it is a software bug, I bet they will fix it in short order. I wonder if any of the beta testers saw this behavior???


On the Alt hold issue...Just looking for reasons it might do this. Have you flown since stiffening the bracket? Is there anything you notice about your airplane when hand flying it that makes it hard for you to hand fly a fixed altitude? Any slight binding or dead zones with the elevator right at the sweet spot? Any slop whatsoever in your control linkages (have someone hold the elevator and check for slop at the stick).

If your sure about all of these things, then maybe Dynon has some more tweaking to do to their algorithms...

I should have clarified -- yep, I've flown since stiffening the bracket.

There is no noticeable drag or slop in the controls. The only noticeable slop is minute -- it's in the bearings which is typical for RVs. The only drag on the elevators is the drag created by the servo itself.

When I engage the servo on the ground it will 'wiggle' the arm a little bit...but it's really not a noticeable amount -- you hear it but you really can't see it. Bill, are you experiencing this?

Jamie said:

...When I engage the servo on the ground it will 'wiggle' the arm a little bit...but it's really not a noticeable amount -- you hear it but you really can't see it. Bill, are you experiencing this?

Click to expand...

Yes, this is normal. The idea is to let you know the AP is working but it should keep you from taking off with the AP turned on.

When you get back to the plane tomorrow, have it run the servo test again just to make sure everything is working as it should before you go fly.

Jamie,
Have you checked your CG recently? A very rearward CG makes for a plane with almost no pitch stability, and that's the only situation we've had in testing where a plane acts like yours does.

Also, are you getting any slip indications or out of trim indications? If you aren't, can you make them happen manually?

Have you been able to fly with the datalogger turned on? Sending us that data would be very helpful.

I know this must be really frustrating for you, but there's got to be something in your install, airframe, or something broken that is unique. We've got the AP working quite well in a bunch of RV's (and other planes). None of them have this issue. We'll do everything we can to figure this out.

Small Idea

I'm not an electronic wizard like most of you, but a friend had one of the inexpensive auto pilots and it would wander side to side of the course or heading. We would trim the plane out on heading (RV-6) with the aileron spring trim (works awesome!) then turn on the AP. It would slowly hunt right and left, basically staying on course, but irritating. I thought that once it was on course and in a neutral position it had to wait for a deviation to correct it. We merely put a very small amount of trim to one side, the plane wanted to go one way and the AP wanted to go the other, it steadied up and flew straight. After all that, I'm wondering if it would work in altitude hold, get to your desired altitude, turn on the alt hold and trim up just a little. Just another reason to burn 100LL.

Let me know

RT

I had a similar pitch problem with a Glasair and STech. AP. A friend of mine, a KC135 pilot flew the airplane and said that it "felt a little tail heavy". He mentioned that when they would setup to refule an airplane, they would transfer large amoults of fuel fwrd. in the KC135, making the plane much more stable. The CG calc. were right where they should have been in the Glasair. In an effort to solve the issue, I removed the battery, a Gill G35 from behind the seat, and installed a G25 on the firewall. Problem solved, the AP worked great.
Dick

Thanks, Dynonsupport for your weekend response!

dynonsupport said:

Jamie,
Have you checked your CG recently? A very rearward CG makes for a plane with almost no pitch stability, and that's the only situation we've had in testing where a plane acts like yours does.

Click to expand...

I did not check my CG after installing the AP, but if anything my CG is on the forward end. I have a Hartzell C/S prop, firewall mounted battery, etc. The only thing mounted aft of the baggage wall is the AP servo, my ELT and the Dynon remote compass (weighs almost nothing as you know). I have hand flown the airplane to and from OSH and several other long trips with 100lbs in the baggage area and the airplane was very pitch stable. An aft CG condition is something that should be immediately apparent to the pilot.

Maybe it's a forward CG issue? I will put some ballast in the baggage area and give it another try.

When I get my fuel tanks burned down low I will drain the remainder of fuel and perform a new W&B. Again, I seriously doubt this is an issue, but I will do a full W&B just to eliminate that possibility.

dynonsupport said:

Jamie,
Also, are you getting any slip indications or out of trim indications? If you aren't, can you make them happen manually?

Click to expand...

I trim the airplane before I engage the autopilot. I don't see any out of trim indications until the oscillations reach their 'peak', meaning that at the top of the oscillations I will see a 'down' trim indication and at the bottom I will see an 'up' indication. With the settings I have dialed in now, I don't see the servo slipping. Should I? I'm using a torque setting of 100%. I've been told that my allowing the servo to slip a little it actually smooths out the ride.

dynonsupport said:

Jamie,
Have you been able to fly with the datalogger turned on? Sending us that data would be very helpful.

Click to expand...

Yes, the datalogger was turned on all day yesterday. Not sure how much it got, but hopefully it will have the data from the last flight. I neglected to bring my laptop to the airport yesterday so I will go out today and download the datalog.

dynonsupport said:

I know this must be really frustrating for you, but there's got to be something in your install, airframe, or something broken that is unique. We've got the AP working quite well in a bunch of RV's (and other planes). None of them have this issue. We'll do everything we can to figure this out.

Click to expand...

Have any other RVs installed the altitude hold and cut the flange off the bottom of the mounting bracket? That's the only thing I've done a little differently.

Yes, it has been very frustrating and please believe me when I say I'm not taking it out on you guys.

One thing I don't understand about the AP, if I'm 10 feet below the target altitude, it will try to correct at about 300fpm and shoot right through the target. The VSI as displayed on my EFIS is extremely precise. In other words, when hand flying you nudge the stick a little forward or backward and you will see the VSI on the EFIS change before either altimeter (EFIS or standby) really sees any change. It's fairly high resolution. So why does the AP seeing the target altitude coming, seeing that it's climbing at 300fpm...why does it shoot through it. It doesn't start pitching forward until it's already passed the target. The same thing happens on the downline. It goes right through the target. It's like the pitch is held constant until the moment it hits the target. By then it's too late.

N661DJ said:

I had a similar pitch problem with a Glasair and STech. AP. A friend of mine, a KC135 pilot flew the airplane and said that it "felt a little tail heavy". He mentioned that when they would setup to refule an airplane, they would transfer large amoults of fuel fwrd. in the KC135, making the plane much more stable. The CG calc. were right where they should have been in the Glasair. In an effort to solve the issue, I removed the battery, a Gill G35 from behind the seat, and installed a G25 on the firewall. Problem solved, the AP worked great.
Dick

Click to expand...

Thanks, Dick for the info. As mentioned in my previous reply to DynonSupport I will try ballasting the airplane fore and aft. It sounds that like me, you were not able to detect any peculiarities when hand flying. Interesting.

Dynon autopilot working

Jamie:

As I said in my other post, I installed the AP74, Dynon servos and connected them to the D100 & D180. My pitch & roll mounts came from TruTrac and were SECURELY mounted during the 18 month build. They have NO play nor movement nor did I drill thru my painted belly to rivet. As my good friend Chet will attest, I tend towards MORE strength and more rivets leaving little or no chance for movement. Zero movement of the bracket is the right amt as far as I'm concerned.

The pitch servo arm attachment was connected at the most distant hole (per specs), 1.50" out. Given its distance from the fulcrum, any movement request issued to the servo will result in greater movement at the control rod versus attachment at 1.25".

The axis of the movement of my pitch servo remains within the radius suggested and is equidistant from both ends of the travel. In other words, assume full movement of the pitch servo arm is N degrees after attachment to the control rod. My arm rotation end travel between final resting point and absolute horizontal equates to 180 - (rotation/2) and is equidistant from both ends of the axis. Movement will not be linear across the full spectrum of movement but both ends of the axis will be alike. IMHO, that's got to be right for the system to work correctly.

Speak to Dynon first, but if you moved the pitch servo arm connection point from 1.50 down to 1.25, each movement of the servo would generate less movement on the rod. Assuming enough travel remained allowing full flap movement without going overcenter (overcenter would result in disaster), that would seem to reduce the servo's impact on the rod but I would not do that w/o Dynon's agreement.

I guess you could also swap servos, re-educate the software to acknowledge the change, and see if the problem travels with the servo. That's a lotta work but the process of elimination might help.

The Dynon manual adds two terms. "Twitch" and "wallow". On flight 1, I found my 9A rocked back and forth as though I was trying to get the tower's attention. I mistakenly thought that was "wallow" so I increased sens from 10 to 20 and the prob got worse. I reduced to 5 (from 20) and that went away.

You indicated that you've tried pitch sens from 1 to 25 so sens is not an issue assuming sens is working right. Surely you detected a difference in performance from 1 to 25. Yes?

Finally, I assume Dynon has some form of checksum which insures that the software upload was successful, but you might confirm that with Mike at tech support. Windows is notorious for bad uploads, but I would hope that Dynon embedded that simple safety factor in their program which ensures that you uploaded with no errors and the copy you downloaded also had no errors.

Last, the GPS option does definitely HUNT. Using GPS, I pointed the AP to a waypoint which had a compass heading of 10 deg from my present hdg which was compass hdg 270. The AP pointed me towards compass hdg 290 and given the mountain I was headed for, I didn't wait for it to correct. I then entered a different waypoint on my Lowrance 2000C. The new waypoint hdg was 60 deg on the compass, the AP redirected and went straight for it like an arrow.

Later on flight 2, I entered a waypoint which was 190 deg. The AP swung the airplane to 220 and then eventually back to 190. On the moving map display of the Lowrance, the little airplane was visably off course but it eventually straightened out. If I recall correctly, the data on the HSI was correct which says to me that the AP does not gradually adjust as the target comes in (it overshoots) but before anyone criticizes I'd like to test that again on flight 3. If the HSI has the right coordinates for trk and hdg, then the issue would seem to be between the D100 and the AP.

If anyone has any thoughts how to reduce that error, lemme know but I won't be taking any naps (haha) till that's resolved.

Finally, one of Dynon's test pilots lives here in Tucson. He's three of the smartest people I ever met and when I see him next, I will happily discuss your issue and my gps question.

Barry RV9A, Superior O-360, Sens Prop, 90hrs in one year.
Tucson

Jamie said:

One thing I don't understand about the AP, if I'm 10 feet below the target altitude, it will try to correct at about 300fpm and shoot right through the target.

Click to expand...

Jamie,
I don't have much else to say or mention until we see the datalog, and it sounds like you may have one or two unlikely things to check.

As for why the AP is adjusting so hard when only 10 feet off, let's just say that humans make amazing control systems. You have so much information at hand- airspeed, altitude error, pitch, etc. But even more than that, you have knowledge about how much you care about a 10 foot error, you have infinite resolution when moving the stick, and you have hundreds of hours of training in how to control vehicles.

The AP only has a subset of your info. Clearly, something in that subset is making it think it needs to pitch a lot to fix the 10 foot error. You might know that just means a slight pressure on the stick, but when you do that, you also know via the nerves in you fingertips that you are actually applying that pressure, and you have a whole other set of senses that helps you detect the aircraft is responding. The AP can only look at a few things, apply one tick of motion to the servo, then wait to see if the aircraft responds. Clearly it's overestimating how much it needs to do for some reason.

Anyway, long way of saying that an AP doesn't fly the airplane the way you do or with the info you have. If you can write a control system that is as smart as a human and can learn how to fly an airplane over time, then you will be a very rich man. As technology stands today, we haven't figured out how to be as smart as a person. But we're trying hard, and we appear to do pretty good in most installs.

All of this does bring up one other question: How many degrees does the arm on the servo move over the full range of pitch? Barry kind of mentioned this, but if the servo isn't moving much, then every little input we put into it causes a big response. It's fine to move the arm in on the servo end as long as you don't get anywhere near over center.

You should not need to back off from 100% torque. I was just checking to make sure that all the feedback from the servo appeared to be working.

Could reverse sensing be the problem?
On my instal (TT) because of the position I had the servo arm mounted on the servo the AP senors corrections were applied in reverse. When a climb was sensed it would give the apropriate directions to the servo to return to level flight. However because of the way I had the servo mounted the aircraft climed further. It was a viscious circle, the more it climbed the more the AP sensor comanded it to climb.

Snipping a jumper wire on the AP altitude module forced the the directions that were sent to the servo to be reversed, problem solved.

Just a thought.

Possible AP74 GPS explanation???

Folks at Dynon support, please tell me if this explanation of the AP74 GPS function is correct.

Hypothetically, let's say I'm flying towards zero degrees on the compass. I instruct the GPS to GOTO an airport located at 90 deg. At that precise moment, the C program inside the D100 draws a straight line between where I am and where I want to head. But, because I'm still moving north at 150kts, my actual position is further north by the time the airplane makes the right turn towards 90, and so it heads back (at 110 deg for example) towards the original line that the D100 drew. When it arrives at the line the D100 drew, it turns again towards 90.

Another analogy would be driving on the freeway. Assume you're headed due north driving on the freeway. The car goes off the side of the road by 100 feet. Conceivably, the shortest point to the destination would be a straight line 100 feet off the road and the destination, but logically you wouldn't do that. You'd drive at a right angle back to the paved road, and then turn north on the freeway.

Am I correct? Is that the logic you folks used to program the GOTO GPS function?

If so, allowing the AP to execute a steeper bank angle would reduce the oscillation reqd to get on track. I can also foresee assisting the GPS by hand banking 60 deg (the AP is limited to 35) in a tight Class C circumstance one frequently runs into in the LA area and then allow the AP74 to clean up and head for the final direction.

Thanks for a great product and answering questions on the weekend.

Barry
Tucson

Gain

Sure sounds like your gain is too high. You said you're at 25 right. And thats the max "sensitivity" right? Hows it behave at 12 or so?

Dynon Support. I presume that this "sensitivity" is really a loop gain adjustment right?

Jamie,

Did you ever vent the static line to the cabin and try the autopilot?

Jamie did try venting to the cabin.

Reverse sensing is impossible with the Dynon AP unless you lie to it. We don't control polarity with wires like TruTrak does. Once you have the system installed, we ask you to put the stick in the upper left then lower right corner. We detect these positions and figure out what way you have installed the servo. Since we actually know the position of the servo (TruTrak doesn't) there are lots of neat things we can do. We even warn you if it the servo arm moved too far and you're at danger of an over-center situation.

In any case, Jamie's plane does not diverge forever. It's just slow to
respond, but does actually oscillate around the target. A system in reverse would just shoot away and never come back.

I do agree that 25 is way too high for "sensitivity" (which is gain). Most RV's are flying in the 9-15 range. 25 is our max. But Jamie has tried everything from 1 to 25 and feels the performance is best at 25, so clearly something is up.

Barry,
You're misinterpreting who does "GOTO" in this situation.

We have no idea where we are going. That's your GPS's job. All we do is listen to what the GPS tells us to do. If it says we're left of course, we turn right, and vice versa.

If you say direct to something that is north of you, but you're headed east, the GPS creates a direct course line from where you are NOW to where you want to go. Direct to isn't a continuous function, it's a request for a line at one point in time.

Ever GPS I have ever seen does this. Direct to means create a direct course from where you are now to where you requested. If you're flying perpendicular to that course, you have to turn around and fly back onto it.

All the Dynon sees from the GPS is the exact same thing that is on the HSI. The course heading, and the course deviation. The AP's job is singular. Make the CDI 0 NM. It doesn't know or care where that takes you, or even if the GPS is moving the CDI constantly. Just do what you can to make it 0. We truly have no idea what your waypoints are or when you create them.

If you're having overshoot issues, try fiddling with the sensitivity some more. When it's dialed in it will fly a perfect curve right onto the course.

Jamie,

You mentioned you originally had a Trio EzPilot installation. I looked at the Trio installation documents and the location of the hole in the bell crank where you connect the Servo linkage is not specified very clearly, they state ?Begin by looking for a point on your elevator bell crank, control pushrod or cable where pushing or pulling the elevator control linkage a distance of 1.5 to 2.4 inches will do the job.? The Dynon installation specifies the hole should be 2.5? from the bell crank pivot axis.

It sounds like your installation is OVER controlling the airplane and this could be caused by the bell crank linkage hole being closer to the axis than the 2.5? Dynon recommends. If that is true you may be able to compensate by using a hole on the Servo arm closer to the Servo shaft but this would result in a loss of torque which may or may not be an issue.

Do you know where the linkage connection hole on your bell crank is located with respect to the bell crank pivot axis?

647jc said:

Jamie,

You mentioned you originally had a Trio EzPilot installation. I looked at the Trio installation documents and the location of the hole in the bell crank where you connect the Servo linkage is not specified very clearly, they state ?Begin by looking for a point on your elevator bell crank, control pushrod or cable where pushing or pulling the elevator control linkage a distance of 1.5 to 2.4 inches will do the job.? The Dynon installation specifies the hole should be 2.5? from the bell crank pivot axis.

It sounds like your installation is OVER controlling the airplane and this could be caused by the bell crank linkage hole being closer to the axis than the 2.5? Dynon recommends. If that is true you may be able to compensate by using a hole on the Servo arm closer to the Servo shaft but this would result in a loss of torque which may or may not be an issue.

Do you know where the linkage connection hole on your bell crank is located with respect to the bell crank pivot axis?

Click to expand...

Hi Joe:

I did not have the Trio avionics altitude autopilot, only the Trio roll autopilot. When I went with the Dynon I elected to install both roll and pitch. My pitch install is per the Dynon installation instructions. The Dynon and Trutrack installations are physically identical and tonight I had the opportunity to inspect a TT install. I could see nothing different physically from my install.

dynonsupport said:

... let's just say that humans make amazing control systems. You have so much information at hand- airspeed, altitude error, pitch, etc. But even more than that, you have knowledge about how much you care about a 10 foot error, you have infinite resolution when moving the stick, and you have hundreds of hours of training in how to control vehicles.

Click to expand...

As a Control systems engineer, I have to disagree. Yes, the human brain is incredible at integrating various kinds of sensory input and determining appropriate control responses. However, there is very little precision, and responses are much slower than an automated control system.

dynonsupport said:

The AP only has a subset of your info. Clearly, something in that subset is making it think it needs to pitch a lot to fix the 10 foot error. You might know that just means a slight pressure on the stick, but when you do that, you also know via the nerves in you fingertips that you are actually applying that pressure, and you have a whole other set of senses that helps you detect the aircraft is responding. The AP can only look at a few things, apply one tick of motion to the servo, then wait to see if the aircraft responds. Clearly it's overestimating how much it needs to do for some reason.

Click to expand...

If the AP does not have enough input to control, then that is a design flaw.

If the controller the AP is using does not have enough resolution to control, that is a design flaw. As you state, placement of that control is critical.

dynonsupport said:

... If you can write a control system that is as smart as a human and can learn how to fly an airplane over time, then you will be a very rich man. As technology stands today, we haven't figured out how to be as smart as a person. But we're trying hard, and we appear to do pretty good in most installs.

Click to expand...

I don't think an AP has to "Learn" to fly over time. This is simple feedback control. If you are familiar with PID control, it is just a matter of tuning the loops to get proper control.

BrentHumphreys said:

I don't think an AP has to "Learn" to fly over time. This is simple feedback control. If you are familiar with PID control, it is just a matter of tuning the loops to get proper control.

Click to expand...

It's that the humans have to "learn" how to tune the thing - that's what takes the time Brent! I have spent countless ours in simulations and flight testing trying to get automatic flight control systems to fly the way we'd like them to. The design only goes so far - after that, there is a bit of an art to it!

Paul

Ironflight said:

It's that the humans have to "learn" how to tune the thing - that's what takes the time Brent! I have spent countless ours in simulations and flight testing trying to get automatic flight control systems to fly the way we'd like them to. The design only goes so far - after that, there is a bit of an art to it!

Paul

Click to expand...

I was going to stay out of this one, but I can't.
You are absolutely, 100% right.
Learning how an aircraft responds to control input and using limited and noisy sensors that contain unavoidable errors and lag is not as simple as it looks.

Once you are done you will have a lot of "If ... then ... else" stuff in your system, it's definitely not just a simple PID controller. If you're going to implement this using a simple PID feedback system I can guarantee you that you will not be happy with your systems response.

This requires a LOT of fine tuning and a clever idea here and there to get it perfect.

Dynon has my full sympathies on this one and I know that their system will continue to develop as user input starts pouring in and it will be as close to perfection as you can ever get it very soon. Give 'em a chance.

Rainier
CEO MGL Avionics

Jamie,
I had the same pitch problem on my RV-6 with a totally unrelated autopilot, an S-TEC 30, and the problem was resolved eventually. I tried everything; I even had S-TEC give me an adjustable gain by providing a rheostat on the pitch computer where I was able to adjust the loop gain in flight. But the problem turned out to be in my static system. I vented the autopilot altitude-hold transducer to the cabin (basically just pulled the static line off) and the problem went away. I think I just had too many items feeding off the static system and it was causing some instability within the autopilot pitch computer in the altitude-hold mode. I know you said you had already vented your altitude-hold transducer to the cabin, but did you re-adjust the gain after you tried this?

lets try and give a logical look at this. you had problems with another AP in your aircraft, you put in the Dynon and you still have problems. What if, you have something in your aircraft putting out something that's effecting the system. Have you tried turning everything off except the dynon, AP, and gps. I mean even turn off the field for the alt. OF course if it works fine, just start turning things back on.

I fully expect Dynon will work through this issue with me until the very end. I feel that a) They realize that I will probably not be the last person to see this issue and b) it's just the right thing to do customer service wise.

Again, I'm not upset with them in the least, just generally frustrated. In fact, I'm more and more impressed with them and their responsiveness and genuine desire to work out this issue. Heck, they even responded to my VAF posts on the weekends.

I have decided that I will meet my frustration with a head-on, methodical approach. One of the neat things about the 5.0 firmware release is they added internal logging to the EFIS (previously you had to bring a laptop, etc. along and capture the data). This data logging will allow me to capture some fairly precise information (1 sec. increments) regarding the flight characteristic of my airplane.

Anyway, I will go up and fly some pitch stability (phugoid) tests and get some good hard data from the Dynon regarding the performance of my aiplane. Then I will engage the Dynon AP and log performance at each sensitivity setting. Hopefully this will give the Dynon guys something to chew on.

In my day job I have done quite a bit of automation work with industrial-type machinery, so I can definately sympathize with them. Once I had a bug in a piece of code that was randomly shooting holes in the side of Coca Cola syrup boxes at a fountain plant. It took me two weeks to find the problem -- so again I can really sympathize. Having said that, I don't think the install base is large enough yet to place the blame squarely on the airplane (yet).

Here is the test card that I plan to fly. If anyone has any other tests that may be pertinent for diagnosing this problem I'm all ears.

Test Card

allbee said:

lets try and give a logical look at this. you had problems with another AP in your aircraft, you put in the Dynon and you still have problems. What if, you have something in your aircraft putting out something that's effecting the system. Have you tried turning everything off except the dynon, AP, and gps. I mean even turn off the field for the alt. OF course if it works fine, just start turning things back on.

Click to expand...

The Trio was only a roll autopilot. The Dynon works perfectly in roll, but not in pitch. This leads me to believe the problems are related. Unless of course we're talking about voltage fluctuations or something. That's the only scenario I can come up with, but the Dynon pitch servo is behaving smoothly, just incorrectly which would probably lead me away from the electrical problem theory.

During the course of testing the Trio I did turn everything off except the ALT field. I have a Plane Power alternator and they say it's fine to turn the ALT field on and off in flight, so maybe I'll give that a try.

Jamie,

Here is something else you can try. If you can get to the static lines under the panel vent everything connected to it except the altitude hold to the cabin for a test flight. Then connect only the altitude hold to the static system which will simulate adding a stand alone static port for the altitude hold which is known to cure this type of problem. Then go fly and try various gain settings and see if this helps. I'll bet you a coke that this makes your problem better.

Bryan Wood said:

Jamie,

Here is something else you can try. If you can get to the static lines under the panel vent everything connected to it except the altitude hold to the cabin for a test flight. Then connect only the altitude hold to the static system which will simulate adding a stand alone static port for the altitude hold which is known to cure this type of problem. Then go fly and try various gain settings and see if this helps. I'll bet you a coke that this makes your problem better.

Click to expand...

Bryan:

I have vented the Dynon EFIS (which drives the AP) to the cabin and it behaved the same. That's why I pretty much excluded the static system as an issue.

You seem to be suggesting connecting just the EFIS to the static system and nothing else, right? If this works, wouldn't it imply that there is a leak in one of the other static-connected instruments?

Jamie said:

Bryan:

I have vented the Dynon EFIS (which drives the AP) to the cabin and it behaved the same. That's why I pretty much excluded the static system as an issue.

You seem to be suggesting connecting just the EFIS to the static system and nothing else, right? If this works, wouldn't it imply that there is a leak in one of the other static-connected instruments?

Click to expand...

No, your thinking to much. I'm suggesting you go to the extreme correction to prove or disprove the static system and all associated components (which were still hooked together with the autopilot when you vented to the cablin, weren't they?) as your cause. If this helps then you can start thinking about why, such as diaphrams in your instruments inducing surges, or leaks, or whatever. If it doesn't work you can look to your electrical system or the autopilot itself. My money is on the static system or related components. Another method that you might try is to put an accumulator in the static line for testing only. Basically a sealed cannister with enough volume that it will dampen out surges or sudden changes giving you a nice smooth pressure at the autopilot transducer. If this works it will likely make the autopilot slow to respond to altitude changes so it wouldn't be a good idea to leave it in there long term, but might help to prove or disprove your static system as your problem.

Another thing you might try is to get the plane just slightly out of trim so that without using the altitude hold you feel very slight pressure on the stick to hold altitude. When you get this to where it is almost not noticable, but you will certainly climb or decend when you let go of the stick try the altitude hold again and see what it does. (This assumes you don't have auto trim available with your autopilot. If you do and can disable it do.) Often the servo likes a slightly out of trim condition better to keep from osciallation on both sides of null. If this helps your problem is with the autopilot.

Making Progress

I am making progress resolving this issue.

Yesterday morning I went out and flew some flight tests aimed at determining the stability of the airplane. If the airplane is unstable, then obviously the autopilot may have trouble.

I performed a couple of phugoid tests. As expected the airplane dampened out nicely with just a couple of oscillations. My airplane is definately pitch stable. I also trimmed the airplane to see how long it would hold altitude/heading with no intervention. I gave up after 10 minutes.

Today I had an opportunity to make a 100nm trip (200nm round trip) and had lots of time to play with the autopilot. I decided to try putting slight load on the servo by trimming the nose down. The Dynon AP will indicate on the EFIS if the servo needs up or down trim. I trimmed nose down just until the nose up trim indicator would hold steady. Doing this, I was able to get the autopilot to hold altitude very well. With the airplane slightly out of pitch trim the airplane handled as it was supposed to -- too high of a sensitivity and you could feel the airplane overworking to stay at altitude. I backed the sensitivity down a little and it smoothed out.

I know of at least one other autopilot manufacturer with auto trim that will actually trim the airplane out of trim (they call it bias) to deal with slop in the control system. Every control system will have a little play in them because of the bearings.

I know that other autopilot manufacturers also introduce a 'deadband' setting. I understand this setting exists to deal with minor slop in the control system. The Dynon AP does not appear to have this type of setting.

My next course of action is to go back into the back and double check the control system for any unnecessary slop (seems like from the bellcrank to the elevators is all that's important). Again, I could have a bearing that has a little too much play in it, etc. but my RV appears to have the same amount of slop (very minimal) as other RVs.

Very interesting, Jamie.....

....and I've been following this thread since I'm about to install my TT servos. The bias helping stability is interesting. I also have a very slight 'play' in my elevator system from the stick to the elevators but not from the rear bellcrank to the elevators so that 'should' be ok.

Regards,

torque and Sensitivity Settings

Jamie,

Like Pierre, I'm following along as well. My Dynon servos and AP-74 are in, but I'm having issues with the 5.0 update (working with Dynon on it) so I'm playing catch up ...thanks for leading the way!.

I was just wondering what torque and sensitivity settings you are finding work best...pretty close to default, or did you have to make large changes? (I know every plane may be different, just wonderin'!)

Interesting note on the pitch trim findings...good to keep in the bag o' tricks!

Thanks, and sounds like you're dialing it in nicely!

Cheers,
Bob

firmware 5.0

Bob
I'm having issues with the 5.0 firmware also. What's the problem you're having? My problem is, that after updating the D180 to 5.0 as per directions, then trying to update the AP74 and servo's I can no longer connect to the D180. My laptop has always had some issues with the Dynon Support program unless it was acually connected to the D180. This time I tried 2 different laptops and 2 dif. USB cables. I get a " chip communication" type error. The D180 was overnighted to Dynon on Wed.
Incidently everything appears ready to go except 1 servo says "upgrade needed".
Regards
Jon

Jon,

My issue is probably self-induced. I had trouble getting the driver installed for the USB to Serial cable, then when that was fixed, I had trouble getting the product support program to recognize the D-100. I had both my D-100 and D-10A, as well as the AP-74 and servos powered up (based on my interpretation of the query "are all DSAB components powered up and connected").

So I tried turning off the avionics master to shut down the D-10A, and hit the "ack" button on the D-100 to keep it on. I intended to turn the av master back on to have the D-100 running on ship's power during the update (instead of the D-100 backup batt). At that point, the support program recognized the D-100, and I got excited and jumped the gun and started the update on the backup battery (by mistake). When I realized what I had done, I was afraid to switch power sources, and let the 5.0 update run on the D-100 backup batt. The backup batt held throughout, and all went well till the very end, when the program rebooted the D-100, and I received the error msg on the yellow screen. Couldn't get the program to recognize the D-100 after that. So I'm calling Dynon on Monday, and may have to send my D-100 in as well. I'll ask them for a clarification on the servo update when I call as well, and let you know what they tell me!

So close...

Cheers,
Bob

Everyone,
We've had some issues with people updating EFIS units to 5.0 while running only on the internal battery. While this seemed to work OK in the past, it's causing issues now. We're investigating why.

For now:

PLEASE MAKE SURE YOUR EFIS IS RUNNING ON MASTER POWER WHEN UPDATING

dynonsupport said:

Everyone,
We've had some issues with people updating EFIS units to 5.0 while running only on the internal battery. While this seemed to work OK in the past, it's causing issues now. We're investigating why.

For now:

PLEASE MAKE SURE YOUR EFIS IS RUNNING ON MASTER POWER WHEN UPDATING

Click to expand...

DS,

Thanks for the reply! Can you speak to the issue (discussed here and in another thread) of updating the AP servos to 5.0. Seems that some are having trouble, in that when they do the EFIS/AP Servo update, it tells them that the EFIS is updated, but then the update program does not sequence to the AP Servos. Is there a step, or trick, to have the update program move to the servos? (That is under the assumption that the servo status says "Update Req").

Thanks much!

Cheers,
Bob

I believe that issue is related, but I can't say 100%. I'll try and get a more informative update tomorrow when I can ask the software team.

Today I spent a couple of hours doing misc. maintenance on the airplane, then pulled the baggage wall in order to check for slop in the control system. I taped the elevators into trail and climbed into the fuselage. I am unable to move the servo pushrod or servo arm. In other words, there is absolutely no slop whatsoever between the servo and the elevators.

I am completely at a loss here, but I'm not giving up. At least I have a 'solution', which is to fly the airplane out of trim. However, I am super anal about things working the way they're supposed to so this is driving me freakin' nuts.

Anyone else have any ideas?

Jamie

I don't think the issue is slop in the elevator pushrod/servo path. Its the slop in the trim linkage that is the problem. I think that what you are seeing is a non-linearity induced stable limit cycle due to a little (or a lot) of deadband in your trim system. This slop in the trim allows the tab to flip one way or the other, this in turns puts a force on the elevator, which moves the plane in pitch, the autopilot responds and the oscillation begins when the tab flips the other way.

By trimming the elevator one way or the other you've taken the deadband out of the trim tab linkage and this has the effect of removing the non-linearity that is perturbing the auto-pilot. This allows everything to settle out since the trim tab isn't moving across the deadband.

All of the RV's have some compliance in the trim system, but whether yours is too much or not also depends on your planes exact CG, decalage of the stab etc...

Its just about got to be a mechanical linkage issue with the trim tab.
...
I hope.

good luck

Thanks, David.

That's a good thing to check. I did a cursory check of the trim a while back and couldn't find any play in it, but I'll do a very thorough check of it again just to be sure.

A friend of mine that used to run an avionics shop told me that they would often have problems with altitude hold units in certain airplanes that were subject to oil canning on the elevators. I have absolutely no oil canning on my elevators but I do have some on the bottom of the fuselage (as documented in this thread).

At this point I'm curious if this oil-canning could be contributing to the problem.

Unfortunately

its a little hard to know what the trim "should" be like. Mine on the '9A has a fair amount (1/2" at the trailing edge) of combined "slop" (manual trim). It flies fine (admittedly with no autopilot). My 8 has almost no slop (also manual trim, go figure). How your airplane will behave is a function of its rigging, exact CG, etc... but I would tend to guess that less slop is better.
Good Luck

I went out last night to check the trim tab and there's absolutely no play in it whatsoever. I guess I can check that off the list.

BTW: I have electric trim.

Looking for other Dynon AP users

If anyone else has the Dynon AP installed, can please move your stick in the following positions and tell me the position number as displayed in the EFIS's AP status screen? Thanks!

full pitch up:
neutral:
full pitch down:

I'm trying to figure out exactly what is unique about my installation. Supposedly I'm the ONLY person having this issue so far and I just can't see how that's the case.

In fact, if you have a Dynon AP and the pitch is working fine, could you please just chime in and let us know!

As far as I know I'm one of the very few that isn't a better tester.

Here are my numbers:

full pitch down: 57
full pitch up: -147
pitch neutral: -18

Since we know that there are 800 steps in the motor, it means that my AP arm is swinging ~92 degrees stop to stop.

Dynon AP RV6

We just installed ours and have a couple of test flights completed.
D100 only, no AP74 installed.
The both pitch and roll seem to be working well.
I will try to get the data you requested in the next couple of days.

Another Dynon AP, RV-6

Jaime,

Just received my D100 back from Dynon after a 5.0 glitch (self-caused, I believe). Did the DSAB config, and servo and AP-74 updates today, and will calibrate and (hopefully) test-fly tomorrow. Will report back on pitch effectiveness, and look for those numbers.

It may become obvious when I run the cal, but just in case, where is the menu page that you're referring to (is it on the DSAB page, scolled to servo status, or in another menu?)

Thanks, and backatcha as soon as I can git 'er done!

Cheers,
Bob

rvmills said:

It may become obvious when I run the cal, but just in case, where is the menu page that you're referring to (is it on the DSAB page, scolled to servo status, or in another menu?)

Click to expand...

Thanks, Bob. The menu is: SETUP->AP->STATUS. I'm working from memory here, so it may say something like "SERVO STATUS" instead of STATUS, but it should be obvious at that point.

Thanks, Guys.

Alschief said:

We just installed ours and have a couple of test flights completed.
D100 only, no AP74 installed.
The both pitch and roll seem to be working well.
I will try to get the data you requested in the next couple of days.

Click to expand...

Have you tried it up at cruise yet (i.e. 155kts, 8500ft) on a long trip? Mine will seem like it's holding fine for a minute or so and then the oscillations will begin.

Are you computer savvy (this goes for anyone reading). If you could turn on the datalogging and download and e-mail me your datalog that would be a HUGE help. I would like to compare the dynamic performance of other installations to mine.

Unfortunately Dynon does not log the servo positions in the log file. It is hard to fathom why they do not do this. The only thing I can think of is they do not want to publish how they are flying the airplane. If they publish the servo positions in the log file one would know almost exactly what the AP is doing to fly the airplane level (at least within the 1 second resolution of the log file).