Just an observation here...
The Dynon autopilot, when flying a coupled approach, pays attention ONLY to the needles (guidance) delivered to it from the certified navigator (of whatever form) through the ARINC-429. If the autopilot flies an LPV approach flawlessly, then the autopilot is functioning fine. The problem necessarily belongs to the receiver and/or the type of nav signal being used by the receiver at that point.
So combining Greg and Carl?s comments, if my needle is accurate on my glide slope and I can follow it smoothly down uncoupled, but my plane is flying above and below that needle when coupled (with healthy airspeed changes), that suggests the problem is not in the 650. Does that sound right?
As you are manually smoothing out the error signal (the needles) that still leaves the possibility the error signal noise from the GTN-650 is causing your problem in the coupled approach mode.
Again recommend testing the autopilot itself. Does the autopilot function properly without the GTN-650 input as in track mode? Altitude hold mode - with manual dialing in of a new altitude to verify autopilot response? Are you using the default autopilot setting as a starting point?
Here a careful monitoring of the GPS signal page on the GTN-650 may help to figure out what what is working and what is not.
Asking the autopilot to follow a glideslope with "healthy airspeed changes", then observing the autopilot fail to do so, is NOT a failure of the autopilot. It's a failure of the pilot to provide proper application of power for a stabilized descent.
If you set the autopilot up to descend at a constant rate (say 800 fpm) and then proceed to jack the throttle all over the place, the autopilot will, of course, fail to maintain the target vertical speed. Any autopilot in any aircraft will exhibit the same behavior. You have to give it a fighting chance...
Dynon autopilot settings for my RV10 - Fixed!
I have gotten my Dynon autopilot on my RV10 tuned really well, with the help of David Brown in Australia! Here is a detailed description of the solution, in case it is useful to someone. The autopilot is now performing outstanding with only a half bar-width deviation on final approach. David and I now have almost identical settings, having come from different directions.
? Slop must be removed. Mine was coming from a cracked capstan. Slop is easy to check if you set the autopilot on ?level? and manipulate the elevator to see how much play you have.
? There are two important trim settings that must be addressed before dealing with autopilot settings (thanks to David Brown for identifying these and getting our two planes on the same page):
1. The trim tab set up should be checked and set according to the procedure posted in Matronics on 2006.
2. The TRIM speed settings must be right to allow the autopilot auto-trim to be fast enough and not fight the autopilot. The settings David gave me (for both pitch and roll) are:
105 Kts ? 100%
130 Kts ? 35%
? Then I re-ran the in-flight tuning (for the millionth time), resulting in sensitivity 13, pitch gain 1.5, altitude gain 0.9. This was fine enroute, but terrible on LPV glide slope (VNAV setting).
? Finally I flew many LPV approaches, testing pitch sensitivities from 13 to 20. For each, I would then test pitch gain until I found the maximum gain that would fly an approach without any oscillations. To do that, I would push the stick to try to induce oscillations at each gain setting. Pitch sensitivity of 17 worked best for me, with a pitch gain of 1.5, and altitude gain has been set at 0.9 from the tuning results. All other numbers are default. David is now at 16, 1.5, 0.9 (only one digit off); he is planning on trying my settings 17, 1.5, 0.9. Nonetheless, we were very far apart and now we are almost identical in our settings.
? The autopilot is a completely different animal as soon as VNAV engages, so flying some VNAV coupled LPV final approaches is a necessary part of in-flight tuning.
o Meeting the in-flight tuning criteria alone results in a sloppy coupled approach
o Setting the autopilot pitch just short of ?twitchy? (in in-flight tuning) results in a rough, oscillating coupled approach.
Here are the settings David and I are using:
Dynon default Rick David
sensitivity 10 14 14
gain 0 0 0.3
sensitivity 10 17 16
gain 2 1.5 1.5
altitude gain 0.6 0.9 0.9
pull rate 1 1 1
VSI gain 1.5 1.5 1.5
g-error gain 1 1 1
g-error limit 0.25 0.25 0.25
Sorry for the formatting problem of the numbers. There are three numbers in each line, separated by spaces. They follow the titles in the first line, so the first number is Dynon default, the second is My (Rick?s) setting, and the third are David?s setting. So for roll try a sensitivity of 14, and gain between 0 and 0.3.
I can?t seem to line the columns up in the forum posts.
Thanks Ron! Much better.
Dynon Skyview A/P
Just started flying the RV-8 with an upgrade to the AFS 5500 with touchscreen & the Skyview network. I have the A/P screen controls and the external A/P control (so I could have auto pitch trim). Using a GTN 650 IFR navigator with the Dynon Airinc and Dynon servos.
Anyone have some ballpark numbers for an RV-8?
|All times are GMT -6. The time now is 11:21 AM.|