Electric pitch trim runaway - no airframe malfunction selected.

[Hornbeam]

Hi folks.

I have been flying my Aerowinx PSX for several days now, using a Fulcrum yoke, purely manually - no Autopilot, no Autothrust, no Flight Director.

I was having a lot of difficulty trimming in pitch; I trim for the speed, but if I let go of the yoke, and glanced away, sometimes the pitch would veer miles off from where I left it.

I initially put this down to rust on my part, having flown (real) FBW Airbus for my day-job for the last 10 years, so am not used to having to pitch trim. But then I noticed that the Aerowinx pitch trim sometimes moves by itself or continues moving after I have released the pitch trim switch(es).

Looking at the USB list on the Instructor panel, I notice that the pitch switches sometimes latch on as being still pressed, even when they have been released.

Of course, this causes a massive change in pitch and huge deviation from controlled flight - a bit like the Boeing MCAS system must have seemed to those poor unfortunate pilots.

To see if it was the USB yoke or the Aerowinx causing the problem, I plugged the yoke into my old Windows laptop to use the USB yoke test page, and I cannot reproduce this switch latching on that.

But I can replicate the fault on the Aerowinx PSX USB page, so it would appear that Aerowinx is somehow causing the switch latching. (There is no pitch trim runaway malfunction selected on the Instructor panel).

I have taken - but cannot find how to enclose - two photos which if you zoom in, can just make out the pitch nose down switch showing pressed even when you can see that it isn't on the yoke. I have also zoomed into the pitch trim readout, top left, and the second photo shows the situation about 5 seconds later - where the white pitch trim indicator has continued moving way past the correct point, and the USB panel is showing switch pressed.

Any thoughts or fixes? - it is really difficult to fly manually with a semi-permanent pitch trim runaway !

[Jeroen Hoppenbrouwers]

To post images:
https://aerowinx.com/board/index.php/topic,1744.0.html

[Hardy Heinlin]

Hi Hornbeam.

But I can replicate the fault on the Aerowinx PSX USB page ...

So when you're pressing your trim button, the assigned function on the USB page indicates "pushed", and when you release your button, it keeps indicating "pushed"?
Of course, that shouldn't happen. I've never seen this problem before. Have you accidentally assigned any other USB inputs as well to that trim function?


Regards,

|-|ardy

[Hornbeam]

Morning Hardy,

Yes, exactly that.

I use the electric pitch trim rocker switch to trim level, hands off, for the speed I am flying at as normal, but sometimes the last selection electronically latches on, and drives the pitch trim to one limit.

I have selected all the other yoke switch assignments to 'Off', except Rev cycle, Gear cycle, and X and Y. I am using the computer keyboard to select thrust and flaps, and the computer mouse to change the FCU settings.

I have expanded the trim indicator read-out so I can see when the latching occurs and when the pitch trim keeps moving after I have released the switch.

I can reproduce this latching quite often, looking at the Instructor panel USB screen. Sometimes, it even ignores a switch push, but most often it latches to 'pushed'. Once when it had latched I even pulled out the USB lead but the USB page still said 'pushed'

I plugged the yoke into my Windows lap-top to use the Windows USB controller utility to view the yoke movements and switch presses. None of the switches show pushed unless they are physically held pushed, and I cannot reproduce the latching on that, only on the PSX.

I tried to upload my photos, (Thanks Hoppie), but I don't know how to reduce the file size to below 1MB and 500 pixels wide.

Best regards

[Hornbeam]

Right, I have learned how to reduce an image. I don't know if you can see, but the Instructor USB panel is showing 'pushed' for the nose up trim, but you can see the switch is no longer being pressed. Then as a further test, I unplugged the USB connector. And you can see at top left, the pitch trim, (white bar) has run all the way to the maximum nose up limit.

(all other switch functions except  Pitch trim nose down/up, Reverser cycle, Gear cycle, and X and Y are selected "off".

[Jeroen Hoppenbrouwers]

For simple but effective (95% of your needs) graphical image manipulation, I recommend https://www.irfanview.com/

Hoppie

[Hardy Heinlin]

Well, in my opinion, the problem lies outside PSX. The "pushed" signal is simply a signal from the yoke's and iMac's hardware and Ventura's software. There's absolutely no "random" function in the sim's interface. It just gets what it gets.

So is this stab trim switch the only erroneous one, and the other switches on the Fulcrum are working OK with PSX?

If so, is that stab trim switch in any way special, compared to the other ones?

Have you tried to run PSX on your laptop? You could connect your yoke to the laptop and then run PSX on the laptop, and another PSX on the iMac, and sync them with a network cable. The laptop's PSX will then send the flight control actions to the iMac's PSX, so that all levers and switches and effects on the two flight decks will be synchronized.

How to connect a Mac and Windows Computer 2023 Tutorial:
https://youtu.be/GNtIaiE0fWY


|-|ardy

[Jeroen Hoppenbrouwers]

Hardy, didn't you once have a simple Java program (.jar) that only connected to the USB subsystem and then dumped all USB events it saw onto the console? If the problem sits somewhere between the USB Host Controller and the Java library API then this would isolate it.

Hoppie

[Hardy Heinlin]

You mean the file Aerowinx/Developers/USB-Detect.txt which is generated during PSX start?

[Jeroen Hoppenbrouwers]

No, but close. A dedicated, simple, small program that _only_ does the USB dump and then per event a line. This would demonstrate whether "it is PSX" or "it is Java" and help finding a solution.

Hoppie

[Hornbeam]

I don't think the problem is outside PSX, because I cannot get the yoke switches to latch on when plugged into the Windows controller utility, but I can get them to latch when plugged into PSX. It is the PSX USB screen that is showing the latched switch(es), nothing else does.

Why would Apple latch switches anyway, I cannot think of a reason???

Yes, I have tried assigning other yoke switches to pitch trim, and the same thing happens.

The only other yoke switches I have assigned are Reverser cycle and Gear cycle, and they have never latched on.

I don't want to link the Mac to the old "dirty" Windows laptop and share files, because I don't want to risk any "infection" getting into the Mac. But the Windows utility shows all the yoke switch presses and the X Y movements, and I cannot get any of the switches to latch on that, but I can get them to latch quite often with PSX.

I am tearing my hair out here and still don't have a reliable working system. The only game running on my iMAC is PSX, and the only yoke connected to it is the Fulcrum. There is nothing else connected and no network etc, just the Yoke and PSX, (and keyboard and mouse).

There has to be a solution to this..........

I will try turning the pitch trim toggle switch assignments off and assigning some other buttons to the trim function and see what happens..........
...just done this and It happens with other switches too. Going to check with the Windows utility again, stand by.......

OK; I have just run a test where I pressed the pitch down switch 100 times - about once every 1/2 second - and watched what was read:

With the yoke connected to the Windows yoke utility, out of 100 switch presses, the indication always registered and never latched.

With the yoke connected to PSX, out of 100 switch presses, it latched or did not register a button push 22 times.
Pressing another unassigned switch on the yoke 100 times into PSX saw a no register, or a latch 13 times.

[Hardy Heinlin]

A dedicated, simple, small program that _only_ does the USB dump and then per event a line. This would demonstrate whether "it is PSX" or "it is Java" and help finding a solution.

By "PSX" you mean my Java code that reads data from the 3rd party interface Java code?
In my code there's no chance for any random clicks or latching. In that 3rd party interface code I can't change anything. With the USB signals from the Mac I can't do anything at all. Not sure what you mean by "dump" ...

[Hardy Heinlin]

Hornbeam, the only possible explanation, for me, at the moment, is again: Ventura's security interceptions. Perhaps it blocks at random some on-signals and some off-signals from the Mac to the Java interface.

Are you also seeing non-active moments when moving the aileron or elevator axes slowly from one end to the other?

Do you have a small keyboard or a large keyboard with a numerical pad? I'm asking because a temporary solution (temporary!) may be to use the plus and minus keys on the numpad for stab trimming.


One more idea for analysis purposes:
When a buttonpush isn't recognized on the USB page, please don't release it; just wait and see whether the recognition occurs some seconds later.
Same test for release events. When you release the button and it remains "pushed" on the USB page, please wait to check whether there's just a delay in the signal line. Maybe the release event occurs later. If so, it would at least exclude the "random trigger" theory.

[Jeroen Hoppenbrouwers]

Currently I suspect the chain of things between the Mac low level usb handler and the Java API. But it remains always possible that all the PSX code together causes a timing glitch. That is why a minimal program that prints each incoming usb event may be useful.

[Jeroen Hoppenbrouwers]

To connect the pc and the mac no files need to be shared. Just copy PSX using a thumb drive and the rest PSX does by itself. Only an Ethernet required.

[Hardy Heinlin]

But it remains always possible that all the PSX code together causes a timing glitch.

The Java USB interface provides an event queue which PSX polls at 20 Hz. A human finger* can't tremble that fast to overfill that queue in 50 ms. PSX can't miss any press-and-hold status that is longer than 50 ms. And PSX can't miss any released-status. I've never read of any error reports on this subject in the past 12 years. But I agree with you; I suspect a random delay between Ventura and the Java interface.


* Well, maybe electric noise is making 1000 push events in 50 ms?

[Will]

I was thinking something similar, maybe Java and Windows handle de-bouncing differently?

[Hornbeam]

Thank you for your suggestions.

As far as I can tell from the PSX USB panel, there are no latching or non pushing events when I move the X and Y axes throughout their full range.

I tried holding the switch on for 5s several times when it did not register a switch push, and it never recognised that the switch was pushed.

If you load Basic 014 - aircraft straight and level in the cruise - see what happens if you disconnect the autopilot. The aircraft very quickly pitches up. So it is not correctly trimmed and the autopilot must be holding the elevators instead of properly trimming the pitch.

I will use the keyboard pitch trim instead of the yoke pitch trim for my afternoon's flying practice, and let you know how that goes.

[Hardy Heinlin]

As far as I can tell from the PSX USB panel, there are no latching or non pushing events when I move the X and Y axes throughout their full range.

Sorry, I didn't mean latching/pushing; I meant non-motion periods on the axes. When slowly turning the aileron from 9 o'clock to 3 o'clock, the software-aileron in the sim might stop turning at, say, 10 o'clock while the hardware yoke keeps turning further, and at 2 o'clock, for example, the software-aileron might jump from 10 over to 2 o'clock.

I tried holding the switch on for 5s several times when it did not register a switch push, and it never recognised that the switch was pushed.

If the stab trim, in several cases, moved to the max, as you wrote, 5 seconds might be too short for this test. 1 minute, minimum, might be better. Maybe you could do the test when your sim session is over, and then find a missed button-released-event (instead of a pushed-event). And then just leave the sim for several minutes. When you return, that button-released-event might have occurred. It's just a test.

If you load Basic 014 - aircraft straight and level in the cruise - see what happens if you disconnect the autopilot. The aircraft very quickly pitches up. So it is not correctly trimmed and the autopilot must be holding the elevators instead of properly trimming the pitch.

Speaking of autopilot: It should disconnect when a stab trim switch on the yoke is pressed (unless it's in autoland, or when the ALTN stab trim on the aisle stand is used).

You never experienced any unwanted A/P disconnections? That means your stab trim switch doesn't generate random signals during this time. Do you touch your yoke when the A/P is operating?

[Hornbeam]

Hi Hardy

#1 As far as I can tell from the tiny yoke symbol at top left of the screen, the little white yoke symbol does appear to stop in the roll axis, at around 40-45 degrees AoB, but does not appear to stop anywhere on the pitch axis - this is going slowly from full deflection end-to-end in both axes. But to be honest, it is very difficult to tell, and I might be imagining it.

#2 I will try to do what you suggest at some point.

#3 I am not using the autopilot at all, (or any automatics). I am flying fully manually, with raw data ground located Navaids. This is what I bought Aerowinx and the Fulcrum yoke to practice with.

(This is possibly a red herring, but I just happened to notice the Basic 014 out of trim condition while checking a preset situation that someone else had programmed - so not anything I had set up, just in case I am doing something daft).

FYI, Yesterday and today I have been flying with the computer keyboard '-' and '+' keys to pitch trim, instead of any yoke switch, and the pitch trim has not runaway once.

Regards

EDIT to add: #2: I managed to induce a switch latch; confirmed on the PSX USB page, and then left it showing 'pushed' while I had a bite to eat. I left the Sim un-paused and with the motion not stopped, (sitting on the runway with the park brake on). It is still latched; showing 'pushed', after 28 minutes.

While in this latched state, I could see that the X and Y values, as read from the yoke, were varying by ± a few units, so I knew the USB page was active and not frozen.