Variation in PID autotune results

How much variation is typical between runs of PID autotune in Marlin? And how much does the variation in parameters matter?

I ran bed tuning for 8 cycles at 60C with M303 E-1 S60 C8 and got values of 260.48, 23.66 and 716.78. Then I ran it again within a minute or two and got 198.70, 13.48 and 732.49. The P and I values show quite a large variation. When I ran this again for 80C, I got even more wildly different values: 53.78, 2.64 and 274.05. Is this normal?

---

See my blog at [moosteria.blogspot.com].

Did you let it cool to room temperature?

Quote
imqqmi
Did you let it cool to room temperature?

I didn't, and when I tried this the results were more consistent.

One other anomaly I see is that the time taken to heat up varies a lot from one occasion to another (I mean in normal operation, not when doing autotune). Any insights into why? This is with a MK2 mounted above an aluminium bed and with a sheet of window glass on it.

---

See my blog at [moosteria.blogspot.com].

It may depend on ambient temperature and if its stikl warm from the previous print, but it could be a loose contact on the wire, check the screw in connectors, make sure there's no break in the wires. Do you use insulation under the heated bed? It shaved off a minute of heating time on my heated bed. Is there a fan running pointed at the heated bed?

Thanks for your suggestions. I really appreciate how people are willing to step in and advise on this forum.

There is a fan attached to the extruder, but a very weak one, and I've been doing tests with it 30mm above the bed, so it's not likely to have much effect. I've been running some tests on the time to heat from cool (after leaving it for a long time) to 80C and have seen values of 7 minutes, 10 minutes and 5 minutes. The starting temperatures were within 1 degree of each other. These are in order of when in the day I did them: about 9am, 12pm and 1pm. It's a hot day here today, and I don't think this looks like there is a systematic variation with the ambient temperature. I have no insulation under the heated bed, but was able to slide some pieces of cork in (I was trying to not take it apart) and saw similar inconsistent variation. This leads me to think that some electrical explanation is most likely. The temperature curve shown in Repetier host is quite smooth, and the bed reliably once it is up to temperature, so I am not completely convinced yet. I'll play around a bit more, see if I can see anything.

This line of investigation came from recently wanting to print using PETG. I usually use PLA with a bed temperature of 55 or 60. PETG needs something more like 80 to 90, and it was taking a long time to get there. So I wanted to see what I could do to help it along.

---

See my blog at [moosteria.blogspot.com].

You can set the hotend to say 50mm high and feel with your hand the airflow of the cold end fan. Mine was blowing over the bed, it was stalling a bit blowing air in the opposite direction. I put a 'wind screen' on it to avoid the air to flow over the bed.

You could try connecting a voltmeter over the heated bed pins and see how stable it is over the course of several prints. If the voltage drops, the psu may not have enough power or it's resisted or dissipated somewhere else.

I can highly recommend 24V for the heatedbed, if your hardware supports it, it can cut the heating time in half.