A1 unstable

[bertrand35]

Interesting issue from Thomas, he has A1 unstable from time to time. It is so unstable that sometimes voice alarms (relative to A1) are played ...
I copy his message here, in case somebody has an ...

1. The A1 issue is intermittent.
2. Sometimes I can power cycle the Tx or Rx and the A1 problem goes away.
3. It only affects my two DF4R Rx's. My 8-Ch telemetry Rx is Ok.
4. When the A1 voltage becomes unstable the RSSI data remains stable.
5. When the A1 is stable, sometime I can increase the throttle stick to max and A1 becomes unstable. Returning the throttle stick to minimum restores stable A1.
6. When the A1 is stable, sometimes I can add slight up elevator (about 3/4 stick) and A1 becomes unstable.
7. When the A1 is stable, sometimes if I switch the GEAR or RUD switches to +100% the A1 becomes unstable.
8. Sometimes the A1 voltage is stable and works fine. But most often it has the bug.
9. I am using the PPM sum mode on the DF4R. It's PPM frame rate is about 17.7mS.
10. I have tried different frame rate settings in the 9x menu and this does not affect the problem.
11. Changing the 9x's PPM length from 8-CH's to 6-CH's does not fix the problem.

I tested these configurations and they all had the same A1 random data issue:
opentx-stock-frsky-nosplash-en-r2281
opentx-stock-frsky-nosplash-audio-en-r2281
opentx-stock-frsky-nosplash-voice-ttsen-en-r2281

I then loaded this old version:
open9x-stock-frsky-nosplash-audio-voice-haptic-ttsen-en-r1464
And it had the problem too! Which was unexpected since I never noticed the issue before. But on this old version I found that he A1 problem only occurred with stick positions I don't use while I fly the Quadcopter. For example, it required full up elevator before A1 voltage was corrupted. On the newer release the problem occurs in several different switch and stick positions so it is more noticeable.

Both of my D4FR receivers have the problem. The 8-channel Rx is fine. I have three 9x's, but only one has FrSky telemetry. Hopefully you have a hunch on what is going on.

[Kilrah]

I would sniff the serial traffic between the frsky module and radio. I would suspect the issue is outside of the radio firmware... the only thing we do is display what the module gives us, there's no path that changes between a D8R and D4FR.

I believe Thomas' D4FRs are modified to measure an external voltage instead of the receiver's supply, so if the data sent by the module to the radio is wrong I'd check that next. Bad contact?

[bertrand35]

Right, that's what I suppose too, I just read again the code and can't find anything wrong.

[thomas9x]

Thanks for the help. I appreciate the extra muscle on my problem.

I would suspect the issue is outside of the radio firmware... the only thing we do is display what the module gives us, there's no path that changes between a D8R and D4FR.

I agree. After thinking about the problem last night I concluded that the random voltages I see being displayed are actual data from the D4FR, rather than by data corruption in the 9x code. That is because the low voltage beep is heard from the FrSky Tx telemetry board when the "random" data drops too low.

I believe Thomas' D4FRs are modified to measure an external voltage instead of the receiver's supply, so if the data sent by the module to the radio is wrong I'd check that next. Bad contact?

That is correct, I have two D4FR's that do not have the external A2 voltage input. So the original internal attenuator on A1 is replaced by my own attenuator (16V in = 3.3V out). Here's the A1 mod: http://www.rc-cam.com/forum/index.php?/ ... entry26499

Before contacting Bertrand I examined the issue as a hardware problem since that is what it appeared to be. My scope showed that the D4FR's voltages were stable and clean. My tests included running the D4FR on a 4-cell Nicad battery (4.8V) and using a precision bench supply to provide a stable 12.4V to the A1 attenuator. So the issue is not due to noisy voltages or bad connections in the actual Quadcopter installations.

It is a very unusual problem and it behaves like a software bug. For example, this morning I was testing it and the problem would not appear. Without touching anything else, I gently turned off the 9x, then turned it back on. The problem immediately appeared.

Generally speaking, the problem only occurs in certain stick and switch positions. It's intermittent too, so overall a real challenge to troubleshoot.

More news for today: I use the D4FR's in the PPM-SUM mode. This morning I disabled PPM-SUM (removed jumper from D4FR) and the problem does NOT occur. So I am beginning to think the problem is inside my D4FR's. I like their small size so I think I will have to trash them and buy the newer D4FR-II that has the external A2 voltage input. But before I do that it would be great to beat this one to death a little bit more. Any other ideas?

- Thomas

[ReSt]

I don't know anything about the receivers and their analog inputs.
But, if the processor/A-D converter of the receiver has more than one analog input and that / one of the neighbouring inputs was floating, it could be possible, that a floating, unused analog input would influence the used input.

Reinhard

[Flaps 30]

FWIW - I have a D4FR modified in the same way. With OpenTx A1 set on 13.2 Volts. I do see the odd drop by around 0.3 Volts now and again, when the throttle is set fully open while operating the elevator (channel 2) ... The same happens on Er9x... So it does look on the face of it to be something associated with the D4FR... I would have expected this to be seen on an unmodified D4FR. Sadly I haven't one around... The D6 RX I have doesn't show this problem..

I cannot detect any pattern to what is going on or any way of getting it to repeat the error reliably.

[thomas9x]

But, if the processor/A-D converter of the receiver has more than one analog input and that / one of the neighbouring inputs was floating, it could be possible, that a floating, unused analog input would influence the used input.

Good point, that is a issue that can occur on some microcontrollers, usual in companion with an ADC channel that is not given enough acquisition time before reading the analog input. So I just opened up my D4FR again, identified the floating STM8S005 analog inputs, and tried grounding them. Unfortunately no joy.

FWIW - I have a D4FR modified in the same way. With OpenTx A1 set on 13.2 Volts. I do see the odd drop by around 0.3 Volts now and again, when the throttle is set fully open while operating the elevator (channel 2)

Ditto. I see an occasional 0.5V drop on A1 when the crazy random problem is hibernating. BTW, are you using the D4FR in the PPM-Sum mode? I think that is the mode that makes this issue come alive.

So it does look on the face of it to be something associated with the D4FR... I would have expected this to be seen on an unmodified D4FR.

I agree, this looks like a D4FR firmware bug. My supplier is out of D4FR-II's, so it looks like the big hammer fix will have to wait until it is back in stock.

I have something new for the Open9x voice developers. When the 9x is overwhelmed with too many voltage alerts, it does not speak the fractional voltages. Only the whole number voltage is spoken. For example, instead of hammering me with 10.3V, 9.5V, 11.5V, ..., it says 10, 9, 11, and so on.

- Thomas

[Flaps 30]

Not sure what mode I am using it in to be honest. Just running it like any other RX.

[thomas9x]

PPM-SUM mode requires the jumper across CH3 and CH4 and only one servo cable is connected to the Rx. You probably aren't using it this way, which may explain why you don't see the nasty random A1 voltage issue that is torturing me.

- Thomas

[Flaps 30]

Okay.. That makes sense. No I am not using it in PPM-SUM mode. But I am seeing the dips on A1 as I mentioned earlier.

[bertrand35]

@thomas35
The voice fifo can only have 8 files on stock board!

[thomas9x]

The voice fifo can only have 8 files on stock board!

I recognize how that would create the truncated voice announced voltage. But even with the limited buffer size, it would be preferred to have the 9x say "twelve point five volts", rather than say simply "twelve". The truncated information does not say anything useful and would be best to just skip it.

Here is my volt voice setup:
CUSTOM SWITCHES:
CS3 |d|>x A1 0.39v ---
CUSTOM FUNCTIONS:
CS3 Play Val A1 10

As you can see, I tried to prevent the constantly changing voltage from a buffer over-run by using the repeat time field on the CUSTOM FUNCTIONS settings. But the Play Val Tmr2 logic ignores it. So how about using the repeat time field to skip voltage messages until the time period occurs? Plus, wait until the buffer has sufficient space for the full message.This would reduce the burden on the voice queue and help allow a complete voltage message to be spoken.

Just things to think about. I fully understand the effort required to make code changes and I am the last guy that would expect special handling. So let's say it's just something for the wish list.

- Thomas

[thomas9x]

I placed an order for some D4FR-II's (hobbyking had them in stock). I expect they will solve the random voltage problem.

I appreciate all the help from everyone. My humble apologies for detouring Bertrand's time on the FrSky Rx's bug.

- Thomas

[thomas9x]

SOLVED: The D4R-II receivers arrived this week and they solved the A1 voltage problem. So the voltage bug only affects the original D4FR, especially when it is operated in the PPM-SUM mode.

BTW, the new receiver's design has other benefits. It has both internal A1 and external A2 inputs. I also noticed a slight RF sensitivity improvement (based on RSSI), perhaps due to the longer coax on the antenna's feedline (which allows for better aerial placement in my model). The price is the same as the old D4FR.

A happy ending to a very weird problem. Many thanks to all that helped me sort it out.

- Thomas