Strange Extruder temp instability

All:
Ran into a weird problem last night. I have finally gotten my Prusa-Mendel V2 tuned in with the newest skeinforge and repetier.
I have a Printrboard Rev B running the Marlin firmware. My feed/flow rate is 30.

Everything was printing well for days then last night in the middle of a print the extrusion stopped with an overtemp fault in the middle of
the first layer.

I restarted the same print and decided to watch it, Looking at the temp graph in repetier I noticed some really wild swings in the
extruder temp. I am talking >10 DegreesC between two consecutive readings. At one point the printer faulted for a second with
a low temp extrusion stopped at another an overtemp fault. Looking at the graphs the faults over and under temps were very
transient. This continued to happen over the first 4-5 layers then seemed to smooth out and stop. That latest print actually
finished but not before 4 more under/over temp faults showed up in the log.

My first thought was a short on the thermistor however the temp readings never dropped as far as I have seen them when there
was a short. I even took my needle nose pliers and tapped, and moved the extruder nozzle down near the tip while watching the
temp graph - no real variance. The printbed temp remained stable.

I fired off another big print job before heading into work today to see if ti finishes.

Anyone have any thoughts about what might be happening here?

I had this exact same thing a while ago... Never solved it, but I have heard of others with this issue.


-Nick

have you averaged out your temp reads? i think it is in configuration.adv, or configuration.h

one of my friends had something like that happen to him, and he said the thermistor lead started to work its way out of the solder joint.

---

[mike-mack.blogspot.com]

Do you have Marlin set to use PID to control the extruder temperature? If so, did you autotune the PID parameters for your hot end? If you didn't you would get this type of behavior. I had the same thing happen to me with a J-head MK-V before I autotuned. It would print fine most of the time, but once in a while the temp will overshoot too far and it would result in a overtemp error.

All:
Well swapping to a different power supply did not help. Based on the suggestion provided I did look at my firmware. PID was enabled with the default P, I, and D values. I ran the M303 command and the values suggested resulted in a very agressive PID. With nothing printing the voltage stayed in a +/- 1.5V of the target. However when I started printing same thing happened. Then I noticed by accident that the baud rate for the USB on Repetier host did not match what was in the firmware. Now as an old school guy who worked serial/tty/rs232 connections generally things do not work at all when baud rates do not match. But I fixed that and now a 4 hour print is running just fine with none of the wild gyrations after more than
3 hours.

I think I will still look to tune the PID so it is not so agressive. Looking at the graphs it goes either full power or no power and so constantly jumps
over then drops under the target.

The baud rate setting should not affect the nozzle temperature regulation. As far as I know, temperature regulation is being done by the firmware on the controller with no interaction with the host aside from the M105 and M109 commands. And those commands just basically change the set point. You can disconnect the host and as long as the controller has power, it will continue to regulate the temperature.

baud rate timing if not correct can result in corrupt data. given that typically 10 bits of data are send 1start bit 8 bit data and 1 stop bit, the error rate can be up to 10% and theoretically should not cause data issues, however if more than 1 byte of data is sent, and start bit is not detected, then there is a real likelihood of data error.

In practice serial is hardwired into controller, but there is only a 2 byte hardware buffer. As long as the second byte is stored in memory before the 1st byte is offloaded the data should not get corrupted.

My guess is what is going on is the adc line is picking up transient noise possibly from the power drain of the serial communication circuitry. Line traces outside may not help either.



I would suggest to take multiple samples of adc and average them, or you could filter noise by using a 0.1uf capacitor across the thermister.

If he's using Marlin with Slic3r, then the print won't start until a somewhat stable extruder temperature at the set point has been reached. If the command was corrupted, then it wouldn't have reached the temperature and it woudn't have been able to start printing. Also, as I understand it, there is error checking in the communication between the host and the controller. The error was an over temp error, not a communication error.

Yeah, the more I thought about it today any comm error of the temp read back from the printer should not have affected anything.
Given the the laptop was set to be slower than the printer buffer overrun on the printer is not likely.

It printed stable all night and all day today. With no temp swings. I may swap back to the old PID values and see what happens.