Temperature Fluctuations

deckingman

@drmaestro Those temperature transients look too sharp and random to be anything other than noise on the thermistor signal. If you can't easily re-route the cable, then the best way to fix it IMO is to use screened cable for the thermistor and connect one end of the screen to gnd. (At least that has always worked for me).

Edit. On the basis of what @jens55 said below, it's possible that the noise is being generated by the thermistor itself. So the best thing to do would be try another thermistor and if that doesn't work, use screened cable.

jens55

With about 90% certainty, I can say you have a dud thermistor. Another 5% goes to a bad connection someplace and the last 5% is phase of the moon related. Notice there is 0% chance of this being tuning related.
Why am I so certain? Because I happen to have had the exact same issue and struggled with it for quite some time before deciding to put a second thermistor on the build plate. I relegated the original thermistor that is buried in the heater to backup status and use the new thermistor as the thermistor 'in charge'.
I display both values in the temperature chart for kicks and the original one is still doing it's dance.
As a side benefit, the replacement is more accurate as it isn't buried in the heater.
A tuning related issue will show an undulating temperature but what you see here is sudden temperature spikes, instantaneous changes in temperature, that are not physically possible.
You might be able to tune it a bit, as you found out, to stop the safeties from kicking in or you might change the safety mechanism to be less sensitive. This can buy you some time to get a new thermistor in.

pro3d

I am also strugling with this aafter upgrading FW from 2 to 3-RC4

Link to thread

deckingman

@pro3d That's a different issue. Your temperature graph is completely different to the OPs and has none of the spikes that the OP has.

jens55

pro3d, your problem is tuning related (undulating). drmaestro's problem is a bad thermistor.
Re electrical noise ... I have a line voltage 1200W heater and have not had an issue with noise. At the temperatures we are measuring, the impedance of the temperature circuit is really low and unlikely to be affected by noise. Having said that, shielded cable is always good if you have it available (I didn't).

jens55

Dang it .... the perils of one finger typing .... being out-typed by other people

deckingman

@jens55 said in Temperature Fluctuations:

Re electrical noise ... I have a line voltage 1200W heater and have not had an issue with noise. At the temperatures we are measuring, the impedance of the temperature circuit is really low and unlikely to be affected by noise....................

Me too - although my bed is 800 Watt not 12000W. But however unlikely you think it might be, I did have a noisy thermistor signal (tried several E3D type cartridge thermistors which all had the same problem). I cured it by fitting screened cable. So please don't dismiss my suggestion out of hand - there is a reason I mentioned it.

jens55

I sit corrected !

deckingman

@jens55 Thank you and no worries. After seeing your post about the possibility that it could be the thermistor itself, I did edit my post to the OP suggesting that he try a new thermistor first, and if that didn't work, then try the screened cable. Between us both one or other of our suggestions should fix the OP's issue

drmaestro

Thanks for all the answers. I also think it is a thermistor problem. What is really interesting is that it sometimes works. For example that's the actual temperature chart I am getting now:

I am waiting patiently for my thermistor to arrive, then we'll see. I really like the cartidge thermistors that E3D is using (which is my actual thermistor) as they are easy to install but I've heard some of them can be problematic.

drmaestro

Hi,

An update: Changing the thermistor took care of the problem (for now at least). I still have 1 or 2 degree fluctuations but they are not spikes and I didn't run PID tuning with the new thermistor yet, so that's probably to be expected.

Thanks for the contributions...