Temperature instability triggering thermal runaway protection.

[Refrigerator]

Ever since i've finished my third homemade 3D printer i've been plagued by an issue i've never had before.
For whatever reason my hotend temperature is completely unstable no matter ho much PID tuning i go through.
Yet weirdly enough the instability goes away during the first or second layer and my temperature stays rock solid +-0.5^oC of the set temperature.
But i have to sort the instability out because it feels like more often than not a print will fail due to thermal runaway getting triggered by the instability.
Temperature swings get as bad as +-10^oC, it's so bad i can get extrusion issues.
I'm running my printer off the 12V rail from an ATX power supply through the 12V CPU plug and i think that should be plenty good enough, it certainly was for my first two printers.
The hotend i'm using is an E3D V6 clone. On my previous two printers i used an E3D V5 clone and had zero issues with temps.
And taking a look at the current V6 clone i see no obvious issues. The heating cartridge and thermistor seem both to be tightened in place quite firmly.
The wires going to the cartridge are quite long but so were the ones in my E3D V5 clone so i don't think that's the issue.

I'm quite stumped really  Why would the temperature swing like that and why would it suddenly go perfectly stable? 
Also the temperature instability causes this issue where when i start my print i have to wait ages for the temperature to reach the set point because for whatever reason the temperature just oscillates 5^oC below the setpoint for an eternity and my printer will not start printing until the set temperature is reached. 

[thm_w]

Photos of the hotend?
Do you have a thermocouple you can tape to the hot end block to verify what is seen in the graph is the real temperature?
What are the PID numbers and are you sure they are saving?

It is possible to change the thermal runaway swing to say 11C in Marlin if you wanted, but as you said probably better to figure it out.

[Kleinstein]

Are there things changing when the printer is actually working ? It could be an additional fan or additional heating power and thus less heat form the PID that could make the difference.
I thing we would need a few more details to help. Is the heater controlled by PWM, or voltage controlled ?

The curve does look unstable, like not correct tuned or maybe non working anti-windup.
The distance from the heater to the temperature sensor is a critical point. The closer, the faster the response and the easier to control.
Poor contact at the heater could be a problem and could get less of an issue once more power is needed, as the heat transfer can be nonlinear.

[Refrigerator]

Photos of the hotend?
Do you have a thermocouple you can tape to the hot end block to verify what is seen in the graph is the real temperature?
What are the PID numbers and are you sure they are saving?

It is possible to change the thermal runaway swing to say 11C in Marlin if you wanted, but as you said probably better to figure it out.

I don't think photos are necessary, it's just a regular V6 clone with a fan.
I could try to attach a thermocouple but at 210^oC there's not many things that can hold it in place securely, other than some mechanical means.
And yes, the PID values are saving, rather the save function is disabled and i save by copying the values to my firmware and reflashing so i know they're saved.

Are there things changing when the printer is actually working ? It could be an additional fan or additional heating power and thus less heat form the PID that could make the difference.
I thing we would need a few more details to help. Is the heater controlled by PWM, or voltage controlled ?

The curve does look unstable, like not correct tuned or maybe non working anti-windup.
The distance from the heater to the temperature sensor is a critical point. The closer, the faster the response and the easier to control.
Poor contact at the heater could be a problem and could get less of an issue once more power is needed, as the heat transfer can be nonlinear.

I believe the control method is called "bang-bang" but it's like a low frequency (~4Hz perhaps?)  PWM basically. Perfectly adequate for a heater imo.
I retightened the heater to make sure it's tight but it's still acting the same.
The thermistor and heating element are on the opposite sides of the heatblock with the nozzle in between.

The procedure when i print right now is as follows:
*I manually connect my homemade heatbed to 12V.
*Start print
*As my hotend heats up so does the bed
*When print starts bed is at about 50-60 ish ^oC
And that's it really, the hotend heatsink fan is on always, there's no more things that happen.
So it's not like some fan turns on somewhere, the printer is quite bare-bones right now.

More on the hotbed: it's about a 72W heater and the temperature kind of just settles at about 60^oC. Letting the bed heat up fully or not had little to no difference to thermal stability.

What confuses me most is why does the temperature just randomly go perfectly stable? 
For now i've tweaked some settings in marlin to widen the limit on thermal runaway to just ignore the instability until it settles.
But it's by no means a permanent fix, but as the old saying goes "there's nothing more permanent than a temporary solution".

[thm_w]

I could try to attach a thermocouple but at 210oC there's not many things that can hold it in place securely, other than some mechanical means.

Kapton tape, aluminum foil, etc.

[Kleinstein]

A bang-bang (= on / off with hysteresis) heater controller is not really PID. It is more like the most basic version and not much to tune. The further the sensor is from the heater the lower the frequency for the natural bang-bang "oscillation". So the frequency would be a first hint. The hysteresis for the bang bang part would be about the only parameter to adjust.  Contineously staying below the set point does not make much sense for classical bang-bang control, even if the system is unstable, unless the heater is too weak and always on.
So there may be some slow software type PWM and still a PID controller.

If this is PIC with a slow PWM control, a integral term that is way to small (e.g. mixed up seconds and minutes or worse) could cause such trouble and very slow settling.

The sensor on the opposite side is good to get a reasonable accurate reading, but it is also slow and thus more difficult to control.
Is there some heat transfer material at the sensor ? I know this can be somewhat tricky at the high temperature.

[Refrigerator]

Code: [Select]

#define PID_FUNCTIONAL_RANGE 15 // If the temperature difference between the target temperature and the actual temperature
                                  // is more than PID_FUNCTIONAL_RANGE then the PID will be shut off and the heater will be set to min/max.This setting was previously 10 IIRC and changing it to 15 seems to have helped with reaching that stable state.

[mazurov]

I've had a similar issue with my extruder recently. Temperature swings, oscillations, etc. Turned out the heater was faulty, you may want to check yours.

[Refrigerator]

I've had a similar issue with my extruder recently. Temperature swings, oscillations, etc. Turned out the heater was faulty, you may want to check yours.

That could also be true and i could swap the heater from my old V5 clone into this V6 if it fits.
Currently the printer works fine as is and ithink i'd be entering the area of deminishing returns.
Not to mention that changing the heater might open a new can of worms.
My printer doesn't have a part cooling fan yet but i had some ugly overhangs to deal with so i pointed a small blower fan at the hotend and caused my printer to shut down.
So the heater might be struggling a little, or my fan was just blowing too much air on the block.

[thm_w]

Use a silicone sock if not already. Basically no reason all printers should not have one on.
Could you tell when you pointed the fan if the heater FET came on 100%? Or whats it at for regular printing speeds.

[Refrigerator]

Use a silicone sock if not already. Basically no reason all printers should not have one on.
Could you tell when you pointed the fan if the heater FET came on 100%? Or whats it at for regular printing speeds.

I was too focused on pointing the fan right to pay attention to the duty cycle.

But i've since noticed that the heater is very sensitive to airflow so i might actually go and swap the heater cartridge to be sure.
Changing the settings around did help with stability but it's by no means perfect yet.

[Refrigerator]

Swapped the heater from my old V5 and noticed almost no difference. I guess this is as good as it gets 

[thm_w]

silicone sock?

[Refrigerator]

silicone sock?

I'll try to make one, i think it would be fun to do some molding with a 3d printed mold, much like Stephan from CNCkitchen did.
I don't have any two component silicone, but i know a secret to make RTV act like two component silicone and harden in like 30 minutes. 

[Refrigerator]

Also since i ripped my old V5 hotend apart i had a spare thermistor so i glued it to the bottom of my heated bed and enabled heated bed PWM.
And after an autotune the temperature at 50^oC is stable within less than +-0.5^oC.