Bed temperature lower than what thermistor reports

wilriker

I do have the following setup:
3mm aluminium plate with PCB heater at the bottom and a 3mm mirror on top of it. I also insulated the heater on the bottom with 13mm high-temperature-resistant insulation-wool (SilcaWool).

Now I got suspicious If my bed is reaching the temperature I select or not because I have problems with first layer adhesion every once in a while.

So to test it, I just put a DHT22 sensor flat on top of the glass plate and set the heated bed to 60°C and after about 15-20 minutes the DHT22 reading stabilized at around 47°C which is obviously too low.

I did not yet calibrate my thermistor. I just assumed that it would be very similar to Semitec 104-GT2 as this is what Marlin had set for this heated bed of an Anet A8.

But before doing a calibration I wanted to rule out some other thoughts I had that might be causing this discrepancy:

Does glass insulate that much so this difference could be expected?
Is it a problem that my insulation also covers the thermistor? Should I make a whole into the insulation at/around the thermistor? If yes, how much room does the resistor require?
Is there anything obvious that I am missing?

EDIT: If I collect some more value-pairs for thermistor reading and DHT22 reading is there a simple/calculated way of adjusting the corresponding M305 command for this thermistor?

DjDemonD

Yes the glass will insulate a fair amount, but the DH sensor placed on the bed might not be providing an accurate reading. I've tried two ways to improve my measurement of bed surface-temperature. One is to paint an area of the bed with matt black paint/black sharpie marker and use an IR thermometer, or use a k-type thermocouple, with heatsink paste, place it in the middle of the bed and put something heavy on top of it. Then compare surface temps with indicated temps.

As for adjusting it I don't know of a way to offset the temperature in firmware. You can, of course, set a higher indicated temp to achieve the surface temp you want - I don't like this option it seems a bit low tech.

However, if you switch to PT1000s from thermistors (can plug them in direct), you can change the reference resistor value in M305:

M305 P1 X501 R4700

Change the R-value until the indicated temp matches the actual temp. This might change the fit of the line of resistance versus actual temp for the sensor so choose the temperature you most commonly print at to give you the best fit.

dc42

In tests using a thermocouple, I found that the temperature drop across 4mm glass was around 5C at a thermistor temperature of 60C, and 10C at a thermistor temperature of 100C. However, a lot depends on how you measure the surface temperature, because anything that conducts heat away from the temperature sensor will reduce the reading further.

When printing PLA on that 4mm glass bed, I use an indicated temperature of 70C.

wilriker

I only fixed the DHT22 sensor by putting something light enough on top of it to hold it down flat.
I will repeat my tests again with that same sensor Kapton-taped down onto the glass so that no airflow is interfering and I will also tape my BBQ thermometer probe(s) next to it as a comparison - finally I have a real use for this thing.

deckingman

I measured 3 deg difference between the top and bottom of 6mm glass at 50 Deg C if that's any help. Glass isn't that good an insulator - hence the reason most houses in the UK have double glazing

wilriker

@deckingman said in Bed temperature lower than what thermistor reports:

I measured 3 deg difference between the top and bottom of 6mm glass at 50 Deg C if that's any help.

It at least means that either my test setup has flaws (most probably as I did not prevent air flow as part of this test) or that my thermistor configuration is way off. Will see that after retesting.

Thanks everyone for sharing your results with glass!

Glass isn't that good an insulator - hence the reason most houses in the UK have double glazing

Same here in Germany - with a tendency going to tripple glazing for newer more energy efficient houses.

wilriker

I am retesting today and I did not have time this morning to prep my BBQ thermometer probes to the bed but the DHT22 is now covered with insulation and it already shows a much closer picture. Setting the heated bed to 60°C I know have nearly 55°C measured with the DHT22 which is also more consistent with the findings of everyone answered in this thread.

Now I have a follow-up question: it took a lot of extra time for the temperature to pass the glass. Do you add any kind of delay before starting to print? I mean, this is about first-layer-adhesion and it sticks better (roughly speaking) when the surface is hotter. But if the thermistor reports 60°C and the surface has only reached about 40-45°C and you start printing right away that might cause an issue with adhesion.
It would be easy to simply add a G4 S600 to the start code but does anyone of you do something like this?

Pat

Hi.
Do you heat the bed and the hotend at the same time?

Im heating the bed first (with M109, although its deprecated), then the hotend.
That gives the heatbed surface enough time to heat up properly.

wilriker

@pat Depends. If I want to have the printer to start quickly with printing I will command RRF to heat up the hotend when the heated bed is about 10°C below it's target temperature. As the hotend heats much faster they will reach target at about the same time. The default Gcode generated by the slicer will first heat the bed and only after that has reached target temp it will start heating the hotend. But that only gives about 1-2 minutes max time difference.

My test showed that it takes significantly longer (10-15 minutes) for the surface to reach its final temperature.

Pat

10-15 minutes is pretty long.

G4 S600 after heating should work.

wilriker

@pat I know it's pretty long and I already think about switching away from glass to something rather thin that will be attached to the aluminium plate like GeckoTec or something similar to get the heat faster to the surface - but then I like my glass plate.

Anyway this question was how about other people do or do not take care of the fact that the temperature obviously needs a fair amount of time to pass through the glass to the print surface after the bed thermistor reports temperature reached.

deckingman

@wilriker said in Bed temperature lower than what thermistor reports:

.................
Now I have a follow-up question: it took a lot of extra time for the temperature to pass the glass. Do you add any kind of delay before starting to print? I mean, this is about first-layer-adhesion and it sticks better (roughly speaking) when the surface is hotter. But if the thermistor reports 60°C and the surface has only reached about 40-45°C and you start printing right away that might cause an issue with adhesion.
It would be easy to simply add a G4 S600 to the start code but does anyone of you do something like this?

I repositioned the bed temperature sensor on my printer but then I do use 10mm thick aluminium tooling plate so it takes a long time for the heat to find it's way through to the upper surface. The trouble with using the thermistor that was built into the silicon heater, is that it senses the temperature at the boundary between the heat pad and the underside of the aluminium. So when you start to heat the bed, the temperature where the thermistor is fitted, rises quite quickly. The control system senses this and turns the heater off, even though the upper surface is still cold(ish). Then the temperature (of the heat pad) drops, the heater comes on but quickly goes off again which means that it takes forever to get the top surface up to temperature. So by fitting a temperature sensor into the edge of the aluminium, close to the upper surface, the control system is now regulating temperature based on the upper surface, not the underside. This means that the heater is on for much longer initially and 10mm thick aluminium has quite a high thermal mass. My start gcode heats the bed first, then heats the nozzle to 140 degC then homes all the axes. This extra minute or so after the bed has heated is enough to get the top of the glass up to temperature so I don't add any other delay. Of course, if you only have thin aluminium under the glass, it may not be possible to fit a thermistor in this manner.

wilriker

@deckingman said in Bed temperature lower than what thermistor reports:

Of course, if you only have thin aluminium under the glass, it may not be possible to fit a thermistor in this manner.

Yeah, currently this is still a 3mm aluminium plate with PCB heater directly attached to it. Trying to fit a thermistor somewhere other than its current position will probably end up in me drilling a hole into the PCB layer rendering it useless.

I have plans to switch to a 5mm aluminium plate plus either silicone or kapton heater (currently I prefer the latter) when I change from 12V to 24V.

Pat

@wilriker
Why using a plate between heatbed and print surface at all?

Im using 5mm plywood with a pcb heatbed on top and either 3mm pei-coated aluminium or 4mm glass. Takes around 12 minutes to heat up (110°C). Difference between the thermistor and the surface is 5-10°C.

wilriker

@pat said in Bed temperature lower than what thermistor reports:

@wilriker
Why using a plate between heatbed and print surface at all?

Basically just because that is what I have. I have a cheap MK-something heated bed (they do not even care to define correctly what it is, might be MK2b or MK3) which means PCB heater directly attached to an aluminium plate - but as I said it's a cheap one (as part of an Anet A8) and therefore surface quality and evenness of the aluminium plate are rather doubtable. That's why I put a mirror tile onto it from day 1.

Anyway, AFAIK it is either way recommended to have an aluminium plate also in the sense of heat spreader on top of the PCB heater as they tend to heat unevenly and the aluminium plate will compensate for this. Otherwise I could just flip the bed over and have the PCB sandwiched between an aluminium support-structure and the glass plate. Maybe I will try that also.

deckingman

@pat said in Bed temperature lower than what thermistor reports:

@wilriker
Why using a plate between heatbed and print surface at all?

Im using 5mm plywood with a pcb heatbed on top and either 3mm pei-coated aluminium or 4mm glass. Takes around 12 minutes to heat up (110°C). Difference between the thermistor and the surface is 5-10°C.

Er because aluminium tooling plate is flat - unlike plywood. So heater stuck to the underside of the flat aluminium (which also acts as a heat spreader so there are no hot spots) then you can either print directly on to that, or do as some of us prefer and use a removable glass (or some other material) build surface. No need for any flatness compensation.