![]() |
RV-9 with Dynon AP servo beta test
Quick note, this is a new thread split off from http://www.vansairforce.com/communit...d.php?t=112766 due to the extreme thread drift, and yet a valuable bunch of info that deserves its own thread. MJS
Two weeks ago I installed the two new Dynon control heads, which are a great addition to the SkyView! However, there is a potential problem with the Auto Pilot that I never experienced with all the beta testing I did on the AP. At the top of the AP control panel is a big button labeled “AP”. After I installed the control head and powered everything up I pressed that “AP” button and the controls started banging around, especially in roll. At the time, I didn’t think much of it. However, two days later when I was preflighting, I noticed the aileron would only travel in one direction. Meaning I could have flown right turns or straight. Obviously not a good situation. After looking in the cockpit to make sure nothing was jamming the control stick, I removed the inspection panel in the right wing, below the roll servo. I found that the AP test function bent the attach plate and allowed the roll servo to go over center. (See pictures below.) I contacted Dynon support and was told that is why you are supposed to install the "Range of Motion Limiting Bracket". Dynon’s installation manual clearly states that they “recommend” the installation of these brackets and you can bet, as soon as I find mine, they will be installed! My recommendations to Dynon were to fix the software so the thing won't beat the controls during the test and to redesign the attach bracket with small tabs that run along two sides to add strength and prevent the bracket from bending. One other thing that might help is to put a large area washer between the aluminum spacer tube and the steel plate to help spread the bending load. The roll servo arm in the "normal" position. Note the bent control attach plate. The roll servo arm in the "reversed" position, jamming the controls. The bent control attach plate. |
Quote:
Is the servo installation exactly as recommended by Dynon? That long bolt that attaches the autopilot push-pull rod to the bellcrank provides a huge amount of leverage to bend the attach bracket. I'd like to see a setup that had the push-pull rod using a very short bolt. In any case, the structure must be strong enough to withstand full servo output, without any damage. You either need a weaker servo, or a vastly beefed up way to connect the servo to the bellcrank. You are very lucky to have caught this now. |
I've seen poor servo geometry lock up the controls. My Rocket did this even without the servo engaged. There are some scary installations out there and sorry Bill, but yours looks like one of them. There are some pretty extreme forces involved here due to leverage and fairly complex kinematics that many people don't understand all that well. I see several areas of Bill's that are just begging for trouble.
It's a lot more than just output torque and shear pins. Be careful with these things! |
Regarding Bill's post and pictures... As someone who barely survived an A/P malfunction years ago, this gets my attention. It was a non-RV, certified A/P in a certified airplane and it still very nearly killed us.
Never - EVER - trust electronics, period. Trust electronics controlled by microprocessors even less. Program-set limits are convenient but absolutely not something you can bet your life on. You absolutely must have mechanical limits to keep hardware and firmware malfunctions from killing you. If it's physically possible for the servo arm to go over-center and jam the controls, you have to assume that it will happen at some time, and build in mechanical means to prevent it from happening. In this case you have to take into account the force a runaway servo can put on the end of that nice long AN3 bolt. It doesn't surprise me that the bolt could provide enough leverage to bend the attachment plate. You'd think the shear screw would have done its job before that happened, but obviously you can't depend on that. I do plan to put an A/P in my plane, but will do so with the assumption that it will at some point become possessed and actively try to bring about my demise. |
Quote:
Quote:
The kicker for me was hitting the AP test button on the ground which simply "exercises" the AP as a pre-flight check. |
Quote:
|
Quote:
While similar, the execution between your pictures and the drawing provided have some subtle, yet important differences. Without seeing the actual dimensions of the two, the spacers are much shorter and larger diameter on the Tru Trak drawing, and the rod end is on the inside of the servo arm. Both contribute heavily toward increasing stiffness of the system. |
Quote:
Look closely at the diagram. The push/pull rod is supposed to attach to the servo arm on the side closest to the servo. This will allow a much shorter standoff to be used on the bellcrank. I think a larger diam. spacer (than what is shown in your photos) is also used to help reduce the bending moment of the bolt. On our trutrak servo installations, the spacer used at the bellcrank attach point is only about 25% as long as what you are using. edit - looks like Michael got a post off before me... |
It has been hinted here but I think it needs to be emphasized. It appears from the photos that this servo installation is capable of creating an over center lock condition. It appears that the length of the actuation rod is too short. The geometry of the set up must preclude over center lock. It appears this one does not
I do not think that there is an electronic mode that can lock the servo. The lock up is electro magnetic and there is nothing that can increase the strength of the magnets to the point that they can not be over ridden. This a wrong and dangerous installation!! |
Quote:
Thanks Gang! |
Bill. The link between the servo and the control attachment must be lengthened so that the servo arm is operating between 10:00 and 2:00. It can not operate between 8:00 and 10:00 as your set up currently does.
With current set up, when the servo is at about 9:30 your controls will be locked!!!! It is only by pure luck that this has apparently only happened this one time. |
Quote:
|
Quote:
Dynon Instructions Apparently Dynon installation uses a longer bolt and spacer than TruTrak, and installs the rod-end bushing on the opposite side of the servo arm. I too have the Dynon installation. This is a solid argument for installing the rotation-limiter brackets. I'll be doing so ASAP. |
Time to admit something?
Quote:
Either way, I will try installing that rod per the Tru Trak recommendation this evening and report back. I don?t recall why I put that longer bolt in there, there very well might be an interference problem when putting the push rod on the other side of the servo arm or it could be that I simply messed up. Either way, I will report back after I get a chance to pull it apart. |
Bill glad you read the last post before "fixing" and flying again. This is not about the length of the spacer at the attach point. Yes good practice would dictate re arranging the actuation arm to shorten that spacer.... But the most important point is to get the length of the actuation arm correct so that the geometry of the setup is correct
My reccomendation is to disassemble your current setup. Set your controls at the midpoint of travel. Then set the servo arm at 12:00 or 90 degrees to the actuation arm. Then make an actuation arm the correct length to maintain that geometry. Bolt it all together and check that the servo arm only moves throught 10:00 to 2:00 or so AND NEVER COMES CLOSE TO 9:00 while moving the controls through full range of motion |
Bill. The AP servo can not bend the tab. It was the center locked geometry and you pushing the stick that supplied the force to bend that mount tab. Think of a piston at exactly T D C. You can push it down with lots of force but it will not go down until it rolls past T D C. Your servo rod was "at TDC" and there is enough friction in a powered up servo that you can not get it past TDC. Result - locked control
|
Quote:
|
Quote:
|
Quote:
|
Quote:
I just returned from the airport. It is as I feared, there is not enough room on the roll servo to move the bearing to the top of the arm. This leaves me with three options: 1) Make a push rod that bends up. This could allow flex in the tube. 2) Add a doubler to the existing plate. This will require longer bolts. 3) Make a new plate with tabs bent up to stiffen it. This requires me to buy some 4130 and is probably the route I will go. |
Quote:
http://trutrakap.com/wp-content/uplo...0RightWing.pdf |
Quote:
This is a Dyon servo, not a tru trak. There isn't enough room unless I shave the bolt head and I'm not going to do that. I guess I could shim the servo with washers but that isn't a desirable solution. |
Quote:
However looking at the servo drawings show: TT arm height is .768 Dynon arm height .750 Thats not much difference, the TT installs on top with plenty of clearance so maybe the bearing is differnent? |
It is close, very close.
Maybe a thin washer washer will do the trick. Do they make thin versions of those big washers? Also, I like the idea of the rod dropping away, should the nut come off. |
Yes one thin large OD washer is all you need under the head (I get them from Van's), if you're worried about a new lock nut backing off then install a castelated nut/cotter key (but not really necessary).
PS: I always use new nuts and locktite on jesus bolts :D |
Quote:
1. Space the servo off the mount with an aluminum sheet spacer (not washers). Move it about a tenth to .125. 2. Fabricate a new mount tab (the bent piece) out of the same material you made the spacer. Nice and thick and stiff. 3. Increase the diameter of the two spacers that hold the tab to the bellcrank as much as you can without overhanging the edge of the tab. This will give you a nice stable base. 4. Fabricate the final spacer with the same large diameter stuff you made the other two spacers from, and make it as short as possible without creating a weird angle and locking up the rod ends anywhere in the travel. 5. Install the travel limiters! Take it with a grain of salt. |
Quote:
Only one problem, the idea of washers to raise the servo won't work because the critical dimension is between the top of the servo arm and the bottom of the servo body. No joy. As for suggestion #5, as soon as I can find it, it and the one on the pitch servo will be installed! |
This makes me wonder how many other AP servos from all the manufacturers have been installed w/o a limiter.
|
Quote:
Can the bolt with spacer be extended through to the second parallel plate to avoid having a bending load on a relatively thin plate? Clearly a spacer would be needed to avoid crushing. Having said this, I admit to not having a good mental picture of the geometry and if the bolt hole projects in to material or into a lightening hole. In my head, that would allow the bending on the bolt but a shear load between the bell crank plates to handle the force of the servo. Just a thought. |
Quote:
I can't figure out how that would happen...the only test I know of with the Dynon SV is that it sends the servos to full right/full aft (IIRC), and then to full left/full forward, asking you to confirm that's what it's doing (to avoid reversed servo action). But in the case of the ailerons, they hit the stops long before the servo arm gets anywhere *near* an over-center condition on my -7. Are you saying the servos kept running AND generated enough force to bend that part? If so...YIKES. (FWIW, I have the over-center protection thingie from Dynon installed on both my servos...still, they never get anywhere near even those limits). I just couldn't imagine them having that much torque...enough to bend the steel bracket...before they would slip. I know I did a quick test on the ground of the pitch servo during installation, and it did, in fact, slip (with quite a loud bang resonating through the skin) when the elevator hit the stop. (Dynon concurred that's normal, btw...scared me enough to email 'em! :) ). |
Quote:
BTW, this looks to be unique to the -9 and maybe the -4, if the roll servo is mounted in the wing. The reason is the -9 uses RV-4 bell cranks. |
RV-7 Flier,
You read that right. Yes, the servo generated enough torque to bend the attachment tab. Also read my post directly above this one. The -7 and -9 have different attach points for the push rod. In addition, the -9 uses a larger roll servo (SV42) than the short wing RV's due to its longer wing. During testing I found the smaller servo didn't have enough strength to manage the roll forces in the -9. |
So what "test" were you doing on the ground? The sort of standard "engage autopilot and verify roll commanding left and right/verify pilot can overcome servos" sort of test? And if so, you let it run all the way to the stop?
I have to admit, back when flying Pipers and Cessnas, we just did a quick check...engage it, turn the knob left/verify left yoke, turn right, overcome control forces, turn A/P off. Never let if go all the way, as far as I can remember. And on the -7, with the Dynon SV A/P, it's inconvenient to set up and engage/test the A/P during run-up, so I don't do usually do it...hmmmm...maybe I'll start! :) Put those servo limiters in, though...glad I did! |
BTW, I'd submit that if the servo is not slipping when the aileron *hits the stop* (a force against the servo much higher than any it would see in flight, no?), that the slip torque value is set WAY too high.
|
Quote:
|
For the rest of us...
If there are problems in the autopilot installation for a significant beta tester, what issues do the rest us need to worry about? This thread makes it seem as though it's easy to inadvertently install things in a way that's unsafe.
|
Quote:
http://wiki.dynonavionics.com/Servo_...ion_Guide#RV-9 Could it be that the bellcrank bracket simply isn't strong enough to withstand the forces that the stronger SV42 is capable of imparting? |
Quote:
As for running out of space for a bolt head if you flip the rod end- you might try using a Hi-Loc pin as a bolt. Plenty of strength with a Hi-Loc, so no worries there, but they have a very low profile head. I have a bunch of them around and use them in cases like this. If you think that may work and can't find one, provide me the set up (grip length) dimension and I'll see if I have one. If I do I'll throw a couple in the mail to you. |
Quote:
|
Quote:
|
| All times are GMT -6. The time now is 01:57 AM. |