Bed temperature offset and stabilization

I have a 1/4" aluminum bed, with a magnetic sticker sheet and steel PEI sheet on top. The bed is heated by a 750W silicone heater driven by mains electricity. I have already run a PID autotune.

When I set my bed to a setpoint of 70C, this is the temperature stabilization trend. Time = 0 represents when the printer beeps, indicating the bed is fully heated. The temperature is measured with an IR gun in the center.

Time Temperature
0 40
1 52
2 57
4 59
10 61

First of all, there is a 10C offset in programmed temperature and actual temperature. I can compensate for this in the slicer/Gcode, although I would prefer if there were a calibration offset in Marlin (I haven't found on, which is really odd to me). This isn't the biggest problem.

Second, it takes 1 to 2 minutes AFTER the printer beeps, for the bed to fully heat up, and it takes a full 10 minutes for the bed to stabilize temperature.

I suspect both the temperature offset and the delayed stabilization is a result of using the thermistor in the silicon heater, which reads the temperature on the inside of the silicone heater, and not the bed temperature, much less the temperature of the steel sheet. However, this cofiguration seems very popular, especially among Voron-type printers that use a fixed bed...often they use even thicker aluminum beds, like 1/2" thick. Is this a common problem?

One, I could use an external thermistor embedded in the aluminum bed. If I do this, what kind of thermistor do I need to buy?

Two, I could manually adjust the PID values somehow to compensate...such as by reducing the P value to a very low value, maybe it would heat more slowly, so that when it does reach the temperature, the surface temperature will be closer to the thermistor temperature. But I don't know if that would improve the stabilization time, or just take longer to heat up, then I'd still have to wait for it to stabilize.

Three, I could just program in a delay after bed heating, of 1 or two minutes. Marlin has a parameter TEMP_RESIDENCY_TIME which I believe is for this purpose.

Is there something wrong with my hardware setup or is is this typical?

What you describe is not only common, it's unavoidable physics.

You are heating a 1/4 inch thick aluminum plate from the bottom while it radiates heat into the environment from the top. Because there is no such thing as a perfect thermal superconductor ( a heat pipe or thermal plate comes close), there will always be a temperature difference between the heat source and the print surface.

The PEI on the printing surface only makes the "problem" worse, as it is not a good thermal conductor itself, and its contact with the aluminum is likely less than ideal.

Then there is the IR gun, which is NOT a reliable thermometer. Its reading will vary with the emissivity of the surface it is measuring.Generally speaking, a dark surface will show a higher temperature than a light colored surface, though "dark" and "light" mean very different things when it comes to infrared and visible light.

Next is the thermistor, which is nowhere near the surface of interest (the printing surface), and can only report as accurately as the calibration table Marlin uses to interpolate its resistance to the displayed temperature. Thermistors do not rely on an intrinsic physical property like a thermocouple or platinum resistance sensor, but have a "calibration" curve that each manufacturer attempts to measure and supply with each device. How good that calibration is is subject to many factors, the primary one being the price. You get what you pay for.

You can be darn sure that, with all of the above, the temperature you see is no-where near the "truth".

Does any of this matter? Not really. The time required to get to a certain temperature can only be reduced by increasing the power (wattage) of the heater, and/or increasing the thermal conductivity of the plate (OFHC copper, pure silver, or a heat pipe style device).

The delta between what the sensor says and the actual print surface temperature will always be there. You can:
1) live with it
2) compensate for it mentally, although it will vary with the ambient temperature
3) modify the thermistor table in Marlin so that the thermistor lies in a fashion that makes the print surface be where you want it
4) accept the fact that, in most cases, the actual temperature is not that critical, but its stability is

As a completely anal exercise, I have fabricated a vapor filled copper heat pipe printing bed, with a 400 watt silicone heater, and surface reading PT-1000 platinum sensor on the printing surface. The surface heated to 80C within a minute, with a consistency of better than 1/16 degree over the 12inch by 12 inch surface.

It was a very fun complete and utter waste of time.

I confirmed the temperature with a bimetallic hot-plate thermometer as well...I trust it, but it responds very slowly so I had to use the IR gun to measure the stabilization.

If the thermistor were in the aluminum, instead of the heater, the PID control should compensate for at least some of the offset and "lag" between the heater and the top of the aluminum.

For now I will just add 90 seconds extra stabilization time, but this doesn't seem a common thing to do, at least not within the Voron community. Even the default stabilization time of 10 seconds in the Marlin configuration files seems to indicate many people are not letting their bed properly heat up before printing. Since the first layer is obviously the most important one for bed adhesion, the temperature of the bed during the first layer is what's important, and it seems incredible if everyone is printing a first layer 10+ degrees cooler than their nominal bed temperature without knowing it.

Quote
PCLoadPLA
I confirmed the temperature with a bimetallic hot-plate thermometer as well...I trust it, but it responds very slowly so I had to use the IR gun to measure the stabilization.

If the thermistor were in the aluminum, instead of the heater, the PID control should compensate for at least some of the offset and "lag" between the heater and the top of the aluminum.

For now I will just add 90 seconds extra stabilization time, but this doesn't seem a common thing to do, at least not within the Voron community. Even the default stabilization time of 10 seconds in the Marlin configuration files seems to indicate many people are not letting their bed properly heat up before printing. Since the first layer is obviously the most important one for bed adhesion, the temperature of the bed during the first layer is what's important, and it seems incredible if everyone is printing a first layer 10+ degrees cooler than their nominal bed temperature without knowing it.

Cool. Good luck. Carry on.

Quote
PCLoadPLA
it seems incredible if everyone is printing a first layer 10+ degrees cooler than their nominal bed temperature without knowing it.

Typical, not incredible. For some reason everyone wants to make their printer go as fast as possible and can't wait for the tugboat to finish before they start the next one. You think they'll wait a few minutes to let the bed heat up properly? Hah!

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]