9x upgrade to ATmega128A3-AU + Reboot With Horiz Trims

[ChrisG]

Hello,
I'm Chris, new to the forum (as a poster ) and I'm trying to mod my 9x with a lot of help from Rainer and this forum. So I started with the FrSky Telemetry mod and it worked great. After that I upgraded to an ATmega128-16AU. Done and working I thought. But I think I got a bad ATmega because it had the problem of rebooting when pushing the horizontal trims. Now I ordered a new ATmega128-16AU.
Today I saw that there is a compatible new ATmega128A3-AU which seems to have the same pinout and instruction set but runs at up to 32 MHz with less power consumption (about -45%).
Has anybody experience with this processor? Can I just replace my 128-16AU with this one? Is there maybe a chance of running the 9x at 32 MHz?
If I had another board I would test it myself but the chance of breaking my board is big because I already soldered the processor a few times.

Thanks for your time and answers!

ChrisG

[jhsa]

Chris, welcome to the forum.. I hope you like beer because most of us here seem to like it..
I'm sure the developers are going to jump in and answer your question shortly. Have fun here..

João

[ChrisG]

Thank you Joao, yes I really have fun at this forum, it already saved me a lot of headache. And of course I like beer, even if it causes a lot of headache...

Chris

[mmilan]

jhsa, is there a wiki for the atmega128 mod? Maybe I am overlooking it. If there is not one how about doing one? You seem to know more about the overall mod than anyone else. If you do, I will buy the beer!
Mike

[jhsa]

I'm terrible at writing. There are a couple threads about the m128.. and they are not so long.. It's been a while since I did it.. Maybe I write something when I replace my 128's with the atmega2561
Just waiting for mbanzi to find out what happened to his tx..

João

[mmilan]

Thanks , I should probably wait and use the 2561 when supported.
Mike

[ChrisG]

Hey mmilan,
There are some boards regarding the m128 mod. It is not a big thing if you are experienced in smd soldering and if you know how to flash your 9x. I could manage to replace the cpu so I'm pretty sure anyone can
This topic is about a new cpu that has not been tested. It is maybe not the best idea replacing with this for the first time.
Try a search in the forum, there is a good explanation about changing the cpu.

Chris

[mmilan]

Thank you Chris, you are right. I have a 128 just sitting in its antistatic bag. It's been calling me "chicken" for a month now.

[MikeB]

There have been several reports of m128s rebooting when using the horizontal trims.
What frimware are you running?
While I have a 128 on a board, it isn't in a Tx, I just don't have enough Tx's to have every combination available for testing!
Can you please confirm the vertical trims do not cause a reboot?
Can you also describe when you get reboots from the horizontal trims, always, sometimes, only when auto repeating etc?

Mike.

[MikeB]

Also, where did you see a ATmega128A3-AU, I don't see one on the Atmel site? It isn't a atXmega128A3-AU you saw is it?

Mike.

[ChrisG]

Hi Mike,
I don't think that it is a software bug, I tested it with er9x and openTx, both showed the same error. It's only on horizontal trims and only when I push it more than 3times. No problem when holding trims, neither when using vertical trims.
I think I got a bad m128 or destroyed it when I soldered it to the board. These little pieces seem to be a bit sensitive to overheating.
I hope to get my new one this week so I can solder it and report whether the problem is solved or not.
If you need more info just ask but I can't test anymore because I was so frustrated that I already removed the old one

Chris

[ChrisG]

Ooooops, I think you are right. I saw it at our local shop and may have overseen the "X". I'm sorry.
Shame on me, I also took a look at the datasheet and didn't get the X. Man, that's a bit confusing
Well, ok, then I think it's not worth a talk because there are more powerful ones.

Chris

[jhsa]

Maybe here?

http://www.reichelt.de/?ACTION=3;ARTICL ... tAodhHYAdw


ATxmega128A3-AU, 1.6 - 3.6V

It's not pin compatible with the atmega128 and can't work at 5V. max is 3,6V,
So, I hope you didn't buy one

João

[MikeB]

Just checking the datasheets of the M128 and the M64. The M128 is shown as typically taking 40mA to the M64s 16mA. Along with everything else, this could mean that the 5V regulator is running closer to its 100mA limit.
Maybe, switching certain trim switches then put spikes on the power and/or ground that then trips the regulator into current limit.
Changing to another chip may get one that takes less current.

Mike.

[jhsa]

so, that means that all of us that changed to the m128 are at risk of having reboots?

joão

[MikeB]

It's possible. The fact it seems to be related to switching the horizontal trims, not holding them, suggests suggested the above to me. It may be the tracking or wiring to the trim switches that helps cause such a problem. Every switch has a 0.1 uF capacitor across it, so when the switch is activated, the 5V charge on the capacitor is shorted out. This is why I suggest it might be a spike on the ground.

Mike.

[Flaps 30]

There have been several reports of m128s rebooting when using the horizontal trims..

Both of my operational transmitters have 128's fitted to them, they were both done at the same time. One of them (with the Emartee voice) does not show any problems with reboots. The other one does reboot with the use of the left or right horizontal trims. The vertical ones are okay.

Sometimes it will lose its voice and beeper leaving the haptic and the main system working. Other times it will reboot. Some shots of what it can do is below. Note that the model memory is restored to normal after the TX is powered down and restarted.

Main display after error occured

Model memory

Normal

The above is the worst case. Obviously the TX isn't going to be used until this is cleared. As to why the other TX isn't showing any signs of this problem is just as worrying as the one with the 'fault'.

The TX normally uses Opentx (as shown) so just to see if the same would happen with Er9x, that was loaded onto it.. Same result as to the trims causing a reboot.

[jhsa]

Any chance of someone having this problem trying to bypass the 5V regulator and use another one with more power?

João

[Flaps 30]

Any chance of someone having this problem trying to bypass the 5V regulator and use another one with more power?

I doubt that the Dilithium Crystals could take any more power.

On a serious note. I do see that all the trims come from one side of the processor, with the earthing of the capacitors being done by the same PC strip. In a way it doesn't make sense that only the rudder and elevator trim switches are affected. I would have thought this might also occur with all the other switches (including all toggle switches and push buttons) if earth spikes was an issue. But I do think that this is a possibility as it only happens if you are clicking the switches. All four (aileron and rudder) trim switches show up this problem..

[mbanzi]

Just checking the datasheets of the M128 and the M64. The M128 is shown as typically taking 40mA to the M64s 16mA.

Mike, which value are you looking at in the datasheet? I'm trying to determine the m2561's consumption, as I also had the trim issue.

[ReSt]

Obviously I'm one of the happy persons that do not have this problem.
I just tried, repeatedly clicking both horizontal trim switches (single and simultaniously) as fast as I could, and it didn't show up any problems.

My configuration is:
NO telemetry mod, voice module with m328 ( on an arduino nano v3.0 with USB), beeper not(yet) connected to the voice module
Nearly all compile options activated: PCB=STD128, EXT=STD, PPM_UNIT=PERCENT_PREC1, TRANSLATIONS=EN, UNITS=METRIC
{HELI, TEMPLATES, AUTOSWITCH, AUTOSOURCE, DBLKEYS, AUDIO, VOICE, SPLASH, GRAPHICS, BOLD, BATTGRAPH, PPM_CENTER_ADJUSTABLE, PPM_LIMITS_SYMETRICAL, FRSPY_HUB, WS_HOW_HIGH, GAUGES, GPS, VARIO, FLIGHT_MODES, CURVES, XCURVES, GVARS = all of them = YES}
all others =NO or default.

Reinhard

btw, by testing the switches, I discovered, that one of the trim switches didn't work and found a poorly soldered cpu pin

[MikeB]

Mike, which value are you looking at in the datasheet? I'm trying to determine the m2561's consumption, as I also had the trim issue.

Typical characteristics, find the graph of power for "Active Supply Current vs. Frequency (1 - 16 MHz)".

Looks like 20mA for the '2561.

Mike.

[ChrisG]

OK, those graphs are telling me the m128 is taking 32ma at 16 MHz and even 16ma at idle state. That's a lot I think.
The m128A seems to be a better choice. It's taking 18ma at active state and about 7ma at idle state. (http://www.atmel.com/Images/doc8151.pdf page 338)
Do you see any possible complications? At Atmel I found this sheet: http://www.atmel.com/Images/doc8166.pdf
If this looks good to you I would give it a try

Chris

[ChrisG]

If I would want to replace the 5V regulator with a low dropout one with higher output which one would you recommend? I found a LM1117S-5.0 which is available as SOT-223 and can output up to max 1A ( or at least 800ma). Will this be ok or are there other bottlenecks?

[mhotar]

I have fitted M128A in T9XR and had the same reset issue with UP/DOWN/LEFT/RIGHT buttons (replaced with microswitches). And there is more powerfull regulator than 9X has.
Hot-fixed with 1K resistors in series to each switch.
Yesterday I placed M2561 into 9X radio and the horizontal-trim-issue IS there too.

[jhsa]

Check the datasheet.. I think (but not sure) the LM1117 can only handle 12V max at it's input. So 3s lipo would be a no go I know the AMS1117 is like that..

João

[jhsa]

I have fitted M128A in T9XR and had the same reset issue with UP/DOWN/LEFT/RIGHT buttons (replaced with microswitches). And there is more powerfull regulator than 9X has.
Hot-fixed with 1K resistors in series to each switch.
Yesterday I placed M2561 into 9X radio and the horizontal-trim-issue IS there too.

Hmm, would placing a resistor in series with the trims switches, or increase the value of the existing ones fix the problem?

João

[mhotar]

I think, but I am not absolutelly sure, but I already changed existing 200 ohm resistor with 1k with no sucess. I will do more investigation later today.
This looks like cap-discharging issue, I never seen reset on button release, everytime on pressing stage.

[jhsa]

so what would the fix be then? remove the caps?

[ChrisG]

The Datasheet says max input of 20V. But I'm not that good in reading datasheets . It also shows an output of 1A at 6,5V<= Vin <= 12V but I don't know it this is the max limit or if this just describes the optimum in the test environment.
Also I'm using LiFe which should not give more than 12V.
But for all others we should find out the max Vin so they don't burn their TX.

Chris