Trouble with Repetier 0.92.2 on V2 unexpected power off

[AndyB]

So I posted this over on the firmware forum at Repetier software, but still haven't found the underlying issue. I'm a hacker by nature, so I always like to try the latest code. I am well aware that I can restore my V2 to 0.91 and just walk away from this issue, but I like to debug and pioneer. That said, here's what's happening:

So, I am running a Rostock Max V2 and on versions 0.92 and 0.92.2, after heating the hot end, as the temperature passes over the set temperature, and starts dropping as it starts to converge on the set temperature, the entire board shuts off. I have to turn off the power to get the board to re-initialize. Once I saw a decoupled in the hot end display (dec), but normally I don't. I've turned on "debugging", to see if I can find out why the board is powering down. On 0.91, I've put 4 days of printing without a hitch, so I know it is either a setting that I entered in the 0.92.2 config, or a bug.

I'm using a 40W cartridge heater, and a M3 screw in thermistor, but otherwise it's a stock V2. I'm using the type 3 heater control manager (as opposed to bang-bang or PID), and the lag is about 12.5 seconds. What should I be doing to isolate this problem? What routine is likely shutting down all of the power to the board? Is there some debugging I can use to either write to the SD card, or to MatterControl console?

Wondering if this is a symptom of too low of a value for decouple test period, which seems to be set at 12...

Upped the value to 120 seconds for decouple test period. Didn't make a difference. I hear some high frequency noises, like two moves of the servos, without seeing any motion, just prior to the shutdown. The shutdown happens sooner if the bed heater is enabled.

I was told by user Repetier to make a change in Extruder.cpp that turned off decoupleTestRequired , which I did.

Made the change and uploaded the new version. It's still happening, all I do is connect and send pre-heats. After the extruder reaches the set temperature, and the bed is approaching the pre-heat, it happens. Attached is the log file, and EEPROM settings.

Modified the Printer::kill function to call UI_STATUS_UPD(UI_TEXT_KILLED); followed by HAL::delayMilliseconds(5000); and watched the screen to see if the kill function was being called. I don't see that happening. Not sure what other functions could kill all the power to the board. Is pin 4 on a RAMBO connected to power input on this board? The pins.h suggests to me that this might be the case, but the schematic doesn't appear to have that type of control.

[Earthbound]

The RAMBo does not have a means to turn off power.

If the symptom you are seeing is loss of power -

LCD backlight will be off...
RAMBo cooling fan will be off...
And these turn on when you cycle the power switch...

Then the issue is a bad power supply.

[AndyB]

I'm running an EVGA 650W supply that I've put 4 days of printing on on version 0.91. If I switch back to that version of code, everything works fine, which suggests it's not the power supply. Every one of the symptoms that you describe is what I see, though. Additionally, the power cuts out, after the hotend and bed reach temperature, which could take several minutes, and there's not a problem with the supply handling this. I also hear two high pitched sounds either in the fans or steppers just prior to the power down. For this reason, I'm 99% positive it is not the supply. Is there something that the firmware could be doing to short something, which would kick in a protection circuit in the power supply? The pins.h file for board 301 (RAMBO) in both 0.91 and 0.92.2 both set ORIG_PS_ON_PIN = 4 and PS_ON_PIN = 4, and the code seems to show that it sets this pin LOW in the Printer::kill function.

[AndyB]

Also, to try out the theory that the supply might be bad, I just tested heating up the extruder only, and although it took longer for the board to shut off, it still did. I can't see how a 40W cartridge heater could be too much for this supply.

[mvansomeren]

if somehow ver. .92 allows the Rambo to pass current more suddenly that could theoretically trip a safety in the PSU and cause it to shut off. I had a CoolMax 700W PSU that was fine with the hot end heated, stepper motors moving but after a few min of the heat bed warming up, a safety tripped in the PSU and caused it to power down. It was drawing nowhere near the rated current of the PSU but I believe it was how suddenly it drew current that caused the issue. Switching to a Corsair 750W solved the problem. I know it's a stretch, but you never know....

[Earthbound]

The power is kicking off for one of two reasons. Either it senses a regulation problem and it's shutting down to protect the load, or one or more components within it are overheating. If your power supply has a thermally controlled fan, the thermal detection is likely placed where the unit heats up under high current demand and might not be detecting the heat rise caused by driving a small PWM 40W heater.

Firmware 0.92 may use a different modulation speed that the power supply doesn't like. It isn't an issue of short circuiting or total load, since the hot end is smaller load than the Onyx bed. and that power supply has plenty of current for both.

Might try removing the rear acrylic panel or one of the side melamine panels and improve the airflow to the power supply as an experiment.

[AndyB]

I tested to see if it was the WDT that was resetting the board. An M281 will test this. It seems that the WDT works fine. I recompiled the code without the WDT and the issue persists. It was then suggested that the PWM drive values may be an issue, so I set the MAX and INTEGRAL values all to 255. The problem still persists. I then re-tested heating just the extruder, and then just the bed. It seems that the bed is what causes the issue, as soon as it starts using and PWM value that isn't 100% (255), the power dies. I switched the drive manager for the bed to bang bang (0) and reset everything back to stock 0.92.2 settings, and now the board doesn't reset, but of course the bed temperature varies around the set point (I could live with that). I'm wondering why the 0.91 firmware doesn't have this problem, and the bed was using PID control with PWM, but now the new code exposes this issue. I'll keep experimenting to see if I can figure out exactly what the issue is (certain values of PWM, etc).

[AndyB]

There's some new code that adjusts Heater PWM Speed. I set it to 61Hz, 64 values (HEATER_PWM_SPEED 2) and it seems to have resolved the issue. My guess is the default 15Hz, 256 values makes a very narrow pulse train, and the impulse of switching at that speed was triggering a protection circuit in the power supply. Of course, only time will prove that this is fixed. I hope this will help anyone that tries to upgrade their firmware.

[AndyB]

Happy to report I have the 0.92.2 software running well.

[teoman]

Thank you for sharing.

[AndyB]

Just completed an 11 hour print of the T-Rex lower jaw, and a 24 hour print of the skull. I think that's sufficient to show that 0.92.2 is stable on the V2. Good stuff. I'm uploading my Configuration.h file to help anyone that might be interested. Please note that the tool does not set all of the values correctly, so you may need to do a line by line comparison. Also note that I re-used the temperature coefficients from SeemeCNC's firmware. The temperature presets for ABS are set to values I use for PET.

Configuration.h

Andy

[mvansomeren]

Just out of curiosity, what are the advantages of moving to .92 from .91?

[Nylocke]

Fancy new features, better UI, bug fixes? I also hate speed dependent positioning and the restart controller on job cancel "feature" of SeeMe's .91 firmware. I don't even have mine set to home the machine (most of my cancels are because of z height being off, it's easier to set the height when the nozzle is already near the bed).

[AndyB]

Some new features (which I don't have working yet), bug fixes (most important), that hacker desire to be bleeding edge, and bragging rights!

[mvansomeren]

Fancy new features, better UI, bug fixes? I also hate speed dependent positioning and the restart controller on job cancel "feature" of SeeMe's .91 firmware. I don't even have mine set to home the machine (most of my cancels are because of z height being off, it's easier to set the height when the nozzle is already near the bed).

Some new features (which I don't have working yet), bug fixes (most important), that hacker desire to be bleeding edge, and bragging rights!

So nothing earth shattering? I haven't had any issues with .91 or anything annoying enough to make me want to change. "If it ain't broke..." For now, anyway.

[Nylocke]

You should be fine on .91, I can't stand it because of the UI and the other things I mentioned, but thats just me.

[Jrjones]

There's some new code that adjusts Heater PWM Speed. I set it to 61Hz, 64 values (HEATER_PWM_SPEED 2) and it seems to have resolved the issue. My guess is the default 15Hz, 256 values makes a very narrow pulse train, and the impulse of switching at that speed was triggering a protection circuit in the power supply. Of course, only time will prove that this is fixed. I hope this will help anyone that tries to upgrade their firmware.

Have you tried V0.92.3? I am having the same issues, but changing the HEATER_PWM_SPEED does not help.
As soon as I try turning my bed on, the power supply shuts off. It has no problem with the extruder or the motors.

It stays on if I remove the heated bed leads from the RAMBO. It also works if I swap the +/- leads to the heated bed, but then the LED doesn't work, and only if the extruder stays off.

Have any other suggestions what would cause this? I couldn't really find where to change the PWM speed from 15Hz 256 values to 61Hz, 64 Values.

V0.91 worked fine with the heated bed, never had a shutdown.

[Jrjones]

Setting
#define PDM_FOR_EXTRUDER
to 0 did the trick.

[Eaglezsoar]

Thanks to everyone on this thread for posting valuable information that is usable by every member of this Forum.
This thread will be extremely valuable to anyone upgrading to version 0.92.2 firmware.