Uncommanded Autopilot Engagement

[AviatorJoe]

About 8 minutes into warming my Rotax, the autopilot came on and then disconnected several times, announcing "Autopilot......Autopilot Disconnect" each time. Then it stopped. I flew for an hour with no issues. It didn't happen on the warmup for the return flight at all.
The very next day, I got the same thing during warmup, but it just happened once and that's it.
I took off and the "REL TO ENG" (release to engage autopilot) came on about 400 feet AGL. No annunciation, though.
I requested a return to field and landed with no issues. During the 10 minutes in the air it was on about three minutes, then went off for 30 seconds, then came back on for 8 seconds. A few seconds after I landed it came on again and stayed on until I turned shut down the plane. Whenever the "REL TO ENG" notice was on the AP status indicated "TRK HOLD and ALT HOLD."

At no time did the servos actually engage, thankfully.

During the taxi, I shut off the Avionics switch and rebooted the Dynon. When it came back up The autopilot pitch and roll status areas on the top left were crossed out and I had an AP DISCONNECT STUCK caution message. So, Dynon had figured out that something was wrong.

I went to the autopilot page and, indeed, TRK+ALT and HSI+ALT were disabled. 180 and LEVEL appeared to be still available, but pushing LEVEL did nothing. Obviously, I WAS level, but there were no indications on the AP pitch and roll status areas. They were still crossed out.


Obviously there's an issue with the wiring on one of my AP Disconnect switches (One on the panel and one on the stick).

This is a question for one of the Dynon folks: Once the AP DISCONNECT STUCK caution message is on, does it have to be reset in software after the problem is fixed or does it continually test? I will need to fly the plane to my dealer and if it's constantly testing, I'll pull my Autopilot servo fuses, just in case.

Thanks!

 

[Rhino]

You should contact Dynon support. This is a customer forum. Although they do check here occasionally, their support contact is intended for stuff like this.

 

[AviatorJoe]

I've seen them on here but, that's true! Thanks.

 

[airguy]

You've got a short in your AP disconnect wiring to ground somewhere.

 

[AviatorJoe]

Yeah, I looked at the AP Status and both pitch and yaw servos show the disconnect switch as "PRESSED"

 

[Rhino]

That would do it.

 

[airguy]

Yeah, I looked at the AP Status and both pitch and yaw servos show the disconnect switch as "PRESSED"

See the attached wiring diagram here. NOTE - this is for the experimental version, not the certified, but the AP disconnect function is, to my knowledge, identical.

You could disconnect one servo at a time and see if you can eliminate the "PRESSED" status on the AP Disconnect to give you an idea of where the trouble is. The only way the Dynon knows that the switch is "PRESSED" is through the datawires between the Skyview and the servos - there is no direct connection to the Skyview from the AP Disconnect circuit. So if you disconnect both/all servos, then nothing is reporting back to Skyview that the switch is pressed, and Skyview will tell you the switch is not pressed (and also that your servos are offline).

If the short to ground is inside the servo or the cable pigtail at the servo, then disconnecting that servo will stop the short. Disconnect just one servo at a time for this so the other servo can report the disconnect status back to Skyview and you can see the "PRESSED" status (or not). If you keep seeing "PRESSED" even after checking all servos, then your problem is in the wiring elsewhere. You have a short to ground somewhere in the disconnect wire that goes from the disconnect switch to the servos, as shown in the drawing.

Note also that on the DB9 connectors at each servo pigtail - pin 2 is tied to aircraft ground, and pin 3 (right NEXT to pin 2) is the AP disconnect switch connection. If you have even one strand of wire that gets loose from either of those connecting pins, and migrates over to touch the other pin, it will give you exactly the symptoms you are seeing.

 

[jakej]

That happened to me once - a person purchased a PTT switch from me & replaced it with another one later - problem was the ‘new’ switch was a NC type (normally closed) instead of an NO ( normally open used for PTT) Both switches look the same but functions are very different, go figure .

 

[JimAZ82]

On my Mooney certified install the 25K resistor called for in the install options created a short in the disconnect switch. You can set the options if you install this resistor or not.

 

[JimAZ82]

And what airguy said. Even one strand that is errant can cause problems

 

[Rhino]

On my Mooney certified install the 25K resistor called for in the install options created a short in the disconnect switch. You can set the options if you install this resistor or not.

What 25k resistor? A resistor, by definition, cannot create a short, unless it's bad.

 

[maartenversteeg]

You mention a 25k ohm resistor, but the AP disconnect description in the SkyView Classic Installation guide mentions a 4.7 - 5.3k ohm resistor for this AP disconnect verification function. Was that the value you were referring to or is it different for the certified version?

 

[Rhino]

It's not different in the certified version. There is no 25k resistor called for in the installation manual. That's why I asked what he was referring to.

 

[JimAZ82]

I may have forgotten the value and am out of the country at the moment

 

[Guenin]

the autopilot came on and then disconnected several times, announcing "Autopilot......Autopilot Disconnect"

The OP mentioned that the AP connect came first, then the disconnect. This pattern doesn't sound like a short in the APD/CWS switch to me, despite the warning on the EFIS. With the AP off, I think I could press my APD/CWS button repeatedly and never get the AP to *connect*, which is what the OP reported as quoted above. Could the fault be elsewhere? Do you have an SV-AP-PANEL (which does indeed let you connect as well as disconnect the AP)? If so, the fault could lie there. I had a key-bounce problem with mine that required a send-back to Dynon for repair.