Electric pitch trim runaway - no airframe malfunction selected.

[Hardy Heinlin]

Hi Hornbeam,

there are again a couple of misunderstandings (due to my imprecise language).

#1
I didn't mean to ask where your full-deflection stops are. I meant to ask this: When you are moving a hardware axis continuously (just keep moving in any direction), are there periods in time when the software axis in PSX is not moving? Are there dropouts during the datastream?

By the way, if the aileron in the sim doesn't reach the -90° and +90° full deflection positions, you need to calibrate it (in PSX). If it just reaches 45°, you miss 50% of the available range and precision. Have you calibrated your axes in PSX?


#2
OK, if there's no action after 28 minutes, I guess we can exclude the "delay" theory.


#3
You wrote: "... see what happens if you disconnect the autopilot. The aircraft very quickly pitches up."
So I thought you meant to say that the disconnection causes the aircraft to pitch up. I know that you fly manually. But perhaps you sometimes use the A/P too for some seconds, and when you do that, you might experience A/P disconnections due to stab trim switch actions.

Hornbeam, when I load an inflight situ file, no matter which one, I never get stab trim runaways, neither with nor without A/P. I think we need not discuss the 747 systems. The problem sits in the USB interface which causes random signals. The simulated 747 systems are correct; they just do what the USB signals command. When the USB hardware trim switch is pressed (random or not), it will also press the simulated trim switch in the sim. When the USB gear cycle button is actuated (random or not), it will also actuate the gear lever in the sim.

Could you please just test this? Load the Basic 014 situ and let it run in A/P mode. Don't touch your yoke. Don't touch your desk. Don't make vibrations. Question: Will the A/P disconnect after a while? When a yoke-stab-trim-switch action occurs (random or not), the A/P will disconnect. (Don't make this test in autoland mode; autoland will ignore the switch actions.)


Regards,

|-|ardy

[Will]

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

Have you tried with other controllers? You don't even need another yoke necessarily, just any other USB device with a button. Can you plug in a different piece of hardware and see if the unwanted switch latching is just with your yoke, or with anything you plug in?

[Hornbeam]

Hello both.

#1 No, I think we understand each other; I did what you suggested; moving the yoke slowly from stop to stop in X, then Y, and seeing if the data stream freezes at any point. As far as I can tell, it doesn't, although I might be seeing a freeze at around 40-45 degrees roll demand, but very hard to tell from the tiny yoke readout at top left of screen, and very hard to assimilate from the X and Y values on the PSX USB screen. (The yoke goes all the way to 90 degrees roll).

#3 Let's forget this, it is a red herring. I only mentioned it in case there was something odd going on with the PSX pitch trim function, and the autopilot was compensating with elevator.

I don't use the autopilot or the auto-thrust at all, ever - I am practicing fully manual flying - not even with the flight director; taking off from a runway and doing everything manually. (I am a "real" ATPL pilot with 20 years flying around Europe, USA, India, Canada etc.). But let's forget this particular quirk -although I will do what you ask while I have my cup of tea.
EDIT: I left Basic 014 flying for 25 minutes, and the autopilot did not drop out. When it was approaching PODEM, I disconnected the autopilot and made no switch presses at all. After about 30 seconds, the aircraft pitched up. Anyway, this is not the problem we are trying to fix.

Hi Will, I only have this yoke, no other controller. I do have a cable USB mouse, which I could try.

But I need to practise my manual raw data flying, and am having to do it with the keyboard trim switches instead of the yoke trim switch, which is really awkward.

[United744]

Switch debounce handling drives me crazy. My mouse is supposed to have it in hardware, but obviously doesn't.

Apple have changed the fundamental game controller handling before. It's always possible the Java USB handler is missing an event (synchronous vs. asynchronous) or simple timing issue caused by some subtle internal change.

The Java USB handler is not without issues. On Windows, it can not see one switch on my WH throttle even though it is within the 32-button limit. Consequently, I'd suspect other problems with the programming of the module. A shame Hardy doesn't have source code for it.

[United744]

I'm curious about something: does this latching behavior only occur when trim is assigned in PSX? What happens if you assign one of these buttons to the gear cycle or reverser cycle?

Does it happen if nothing is assigned?

[Hornbeam]

Hi United 744, I did assign two other switches to the pitch trim function, and these would also latch up on occasion.

I only have two other switches assigned; for Reverser cycle and Gear cycle. These sometimes don't register, but they have never latched on. All other functions, I am controlling from the normal computer keyboard. I will now try those two switches to control the pitch trim. All the other switches, HAT etc., are selected 'OFF' on the PSX USB panel.


My investigation time is limited because I really need to practise my manual raw data flying - the whole point of buying this kit and this software. Switch debouncing, (or perhaps the lack of it), is a possibility. Any thoughts, Fulcrum?

[Hardy Heinlin]

Hi Hornbeam.

#1 ... I might be seeing a freeze at around 40-45 degrees roll demand, but very hard to tell from the tiny yoke readout at top left

You can see a larger display of the flight controls when you look at the STATUS page on the lower EICAS.

... When it was approaching PODEM, I disconnected the autopilot and made no switch presses at all. After about 30 seconds, the aircraft pitched up.

This is because you disconnected just the A/P, but not the autothrottle. The 747 automation is not as smart as an Airbus or 777. When the 747, in manual flight, pitches up even a tiny little bit (due to a tiny change in the surrounding air mass or whatever) which also causes the airspeed to decrease a tiny little bit, the autothrottle in SPD mode will try to maintain the command speed and therefore increase the thrust. That thrust increase will cause the nose to go up even more, which will also decrease the airspeed even more, which causes the thrust to increase even more, and so on. -- The same may happen vice versa when the pitch goes down a little bit; the thrust will descrease and the pitch will decrease even more ...


If you don't get any random button clicks in 25 minutes while you are away from your controls, I suspect the problem has, among other things, to do with mechanical vibrations -- and how they are filtered in the interface system.

The gear and reverser switches don't latch because they are momentary action functions. For these functions, PSX only checks for on-signals, not for off-signals. They are single events. In the trim function, on the other hand, PSX checks for on-signals and for off-signals, because the trim function is a continuous state, not a single event. The trim function has a duration with a start signal and an end signal.

In other words, the problem affects all your buttons in the same way, not just the trim switch.


Regards,

|-|ardy

[Hornbeam]

Hi Hardy,

Sorry, I have been away.

I will have a look at the EICAS flight controls page, good idea. As I am purely using PSX for manual flying practice with raw data NAVAIDs I don't have that screen included on my monitor, since I am not flying any system malfunctions. I will bring that screen up and have a look.

Boeing auto-thrust behaviour makes perfect sense, thank you. We can put that to bed now and forget about it. As an avid fan of, and real world Airbus FBW pilot, Airbus really got their FBW automatics suite right, whereas Boeing - playing catch-up - meh. Nowhere near as good. With Airbus FBW, even when you disconnect the autopilot, the FBW is still in operation; making pitch compensations etc. which is a very useful pilot aid, and vastly reduces the flying workload.

I will have a think about what you say re latching. If it is a mechanical issue, it can only be in the sense of a switch bounce when the switches are momentarily pressed I would have thought. I wonder what switch de-bouncing the Fulcrum has? Presumably the computer keyboard has good switch de-bouncing circuits/software, hence why operating the pitch trim via the keyboard never latches on ?

We do see a latched button push remaining even after I have pulled the yoke USB lead out of the computer, presumably because PSX is waiting to see an 'un-switch' event in the USB data stream? But, with the buttons released, surely the yoke USB must be sending repeated 'un-switched' data, so why does the latch remain?? A switch that is not physically pressed, meaning its contacts are separated, cannot bounce on.

[Hardy Heinlin]

Hi Hornbeam.

But, with the buttons released, surely the yoke USB must be sending repeated 'un-switched' data, so why does the latch remain?? A switch that is not physically pressed, meaning its contacts are separated, cannot bounce on.

As we can now exclude the "time delay" theory, I conclude the Java USB interface doesn't get all button events. Some events get blocked at random and will never arrive in PSX.

[Jeroen Hoppenbrouwers]

If it is truly random, Button-Down events should get blocked as often as Button-Release events. Can you please check whether there are cases where you (Hornbeam) need to activate a button TWICE to get anything to work?

Unreleated: Hornbeam, please consider the following.

* Connect your yoke to the Windows PC.
* Copy PSX to the Windows PC, using a USB stick or something simple.
* Run PSX on the Windows PC, zoom in onto something irrelevant that does not move to save CPU.
* Open Instructor, select all buttons and things of your yoke to make it work.
* If you feel like trying it on the PC, try the trim until you know whether it works or has the same issue.
* Start the Server.
* On your Mac, connect your PSX to the Server. Use a simple Ethernet cable in your shared router.
* Nothing needs to be activated on the PC, especially no file sharing. Just run PSX in Server mode.
* Now your PC catches all USB from the yoke and translates it to PSX, using different software.
* See whether this makes a difference. If only to differentiate Windows vs Mac Java.

[Hardy Heinlin]

Can you please check whether there are cases where you (Hornbeam) need to activate a button TWICE to get anything to work?

By "twice" you mean an even number, that turns an event into something "odd"? If you just mean "multiple" attempts, then yes, that was needed, as far as I understand it:

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.

[Jeroen Hoppenbrouwers]

Yes, I meant to see whether both types of events were missed about as often.
And obviously I missed the Hornbeam statement entirely.

[Hornbeam]

Hi Both,

Thanks for the new suggestions. I will try those when I get a moment.

I do not have a server though, just a stand-alone iMac, and an old Windows lap-top. No server, no network.

When I did the 100 switch presses, I just pressed the same trim switch in the same direction 100 times, and counted how many times the indication on the PSX USB page latched on after I had released the switch, OR did not register a push. Then I repeated the test in the other trim switch direction.

I suspect that nobody flies PSX fully manually, the way I need to do - and I never use the autopilot; not even for a moment to stabilise things, so this trim latching issue has probably not been noticed before.

I will report back....

Best regards

[Jeroen Hoppenbrouwers]

The old windows laptop is what you need. Run PSX in server mode. Not a PC server. Just the laptop.

[Hardy Heinlin]

I suspect that nobody flies PSX fully manually, the way I need to do - and I never use the autopilot; not even for a moment to stabilise things, so this trim latching issue has probably not been noticed before.

Nope. I have flown PSX 10,000 hours without autopilot. Many other users also fly manually.

And again: Even when flying on autopilot: Whenever a stab trim switches actuates, the 747 will disconnect the autopilot immediately (if not in autoland). We never experienced such unwanted autopilot disconnections in over 1 million user hours.

Also: It's not just the stab trim switch. This random problem affects your other USB switches as well. We never experienced random gear lever actions or flap lever actions etc.

So the problem you're seeing is not an old undiscovered problem; it's a new problem nobody else has seen so far. Therefore it must be related to your specific hardware and/or software.

Maybe you could borrow another simple joystick and test whether it will generate the same random actions. This way we'll find out whether or not the Fulcrum is a factor here.
 

Regards,

|-|ardy

[United744]

I suspect that nobody flies PSX fully manually, the way I need to do - and I never use the autopilot; not even for a moment to stabilise things, so this trim latching issue has probably not been noticed before.

You suspect incorrectly. I fly PSX by hand most of the time. I have never had an issue with pitch trim.

[Hornbeam]

Ah OK, fair enough, it was just a thought.

Just to add more clarity to our investigations: I have never seen a 'switch pressed' event unless I have actually pressed a switch. I have never had a switch 'on' detected unless I am physically pressing a switch.
I have Gear cycle and Reverser cycle on two other yoke press switches, and these have never actuated 'on' without being pressed by me.
What is happening with the pitch trim is that when pressed on, (trim NU or trim ND), the 'on' event sometimes latches on, or less commonly, does not register the switch press in the first place. (The gear cycle and reverser cycle sometimes do not register an actual switch push, but they have never latched on after being pressed).

I have also just done the flight control full and free check, using the EICAS flight controls page, as you suggested, and there are no gaps, freezes or drop-outs in the flight control movements. So the yoke X and Y are fine.

I don't know where I can get another joystick or yoke, but am thinking about it. I will try to get PSX on the Windows lap-top and see how the yoke behaves with that.

Regards

[Jeroen Hoppenbrouwers]

I will try to get PSX on the Windows lap-top and see how the yoke behaves with that.

Yes, please. We currently think the problem may be somewhere in between the yoke and the software that hands the USB events to PSX. There's enough difference in this between Windows and Mac to have an indication where to look.


Hoppie

[Fulcrumsim]

Hi guys,
Just finished a quick test using Fulcrum hardware on a windows 11 pc. I flew around activating the trim and then trimmed for around 300kts and everything seemed ok. I then tried extended trim activations to try and get a hang up.

Setup was Fulcrum Yoke and throttle with all axis mapped to engines, flaps, spoilers, ailerons and elevator. Switches mapped were elevator trim on the rocker switch on the yoke handle plus A/P disco. Gear cycle and engine run switches.


Below is a video of me activating the trim switch. It is hosted externally so click the link and it will open in a new window.
You can see the trim moving smoothly and then I start activating the switch very quickly (you can just hear the clicking of the switch). There is no bouncing, hanging or locking on at all during this test.

Video demo of trim switch activations and usb interface

I don't know anything really about Macs or Java or how they interface with joysticks. Our products are designed to not require drivers and are always seen as an HID joystick. I thought this test may help with eliminating things. If Hornbeam can try PSX on a windows laptop in a similar way to see if the problem is hardware related then this would be great. Happy to work with you to try and isolate this issue.
Chris

[Hardy Heinlin]

I then tried extended trim activations to try and get a hang up.

Hi Chris,

what do you mean by "extended trim" and "hang up"?

When you press and hold the switch for just a second, the 747 trim will not move. That's normal. You need to hold the switch for at least 1 second until the trim screw in the back of the 747 reaches its normal velocity. It's a huge piece of metal. So, single "clicks" will have no effect. Do you mean that by "hang up"?


|-|ardy