PID acts different w/ different pit probes?

Started two Boston butts with my Weber kettle last weekend with the following PID settings:
P=15 I=0.001 D=20
and posted it to the live HM thread. Bryan suggested to set I=0.08 which brought the pit temperature closer to set point (~midnight).

I went to bed and when I looked at the graph at ~4am, I recognized that the thermocouple had stopped working at 2am :-(
(Meanwhile I have fixed it, it was a cold soldering joint at the TC amp). To save the butts I went up and reconfigured the HM to use the first thermistor probe as pit probe and thermistor 2 and 3 for the butts (Thank you Bryan, for the possibility to do this!). I did leave the PID settings untouched at this time (4am).
As you can see, the amplitude is greater than before. Does the PID work different with the different pit probes?
Later I played around with the setpoint and with the PID settings and rebooted the AVR, but as I said before I'm curious w/ the greater amplitude starting at ~4am.
Any thoughts or comments on this?

PS: I used the minion ring method. The fire runs along the side of the kettle resulting in having a different distance of the pit probe from the heat at every stage of the cook.

I'm willing to bet that the changing position of the fire resulting from the minion ring in relation to the position of the probe has a greater impact on this variation than changing of a probe at 4 am. ie: If the fire is directly below the probe, then adjustments from the blower would noticed quicker by the probe. If the fire is not under the probe, the temperature adjustments would be slower to be noticed.

I think your P = 15 is the main problem. I've used different probes and there was no difference.

I think the probe can make a difference if they are of significantly different construction. For instance, I have a TC Pit probe that is a very thin needle, it reacts to temperature changes wikked fast. I have another TC that is a rod as thick as a pencil, it reacts much slower to temperature changes. You can see the difference really easy by grabbing the probe with you hand and watching how fast or slow the temp raises, and how long it takes to settle back to room temp when let go. I would think the difference in reaction would effect how the HM reacts and could have some effect on the ideal PID settings.

I was thinking into this direction:
Maybe there is a difference in how the inputs (TC and thermistor) are treated in the PID algorithm?
Or maybe there is a problem with changing pit probe from one input to another mid cook w/o restarting the HM (an offset not cleared, probeType not reset, ...)
Another observation during the cook: the logfile entries showing the PID suggestions (Ziegler ...) stopped after changing the pit probe.

PS: The switch from "Probe 0" to "Probe 1" as pit probe was done by disabling "Probe 0"

The HM will continue to react and adjust to input data it receives so I would expect it to eventually settle out even if the pit probe characteristics change, I wouldn't expect that a restart would be needed.
I'm guessing the function that calculates the suggested PID data is hard coded to look at Probe 0 and so it stopped calculating when you had to disable probe 0.
Looking at the graph it seems the pit sat without a pit probe for a couple hours before you switched over to Probe1, so maybe the difference in behavior after your changed pit probe had more to do with the changes that happened to your fire while the pit probe was dead rather than the difference in pit probe type?

RalphTrimble said:

I'm guessing the function that calculates the suggested PID data is hard coded to look at Probe 0 and so it stopped calculating when you had to disable probe 0.

Click to expand...

That would mean, PID calculation and PID suggestion use different sources.

RalphTrimble said:

Looking at the graph it seems the pit sat without a pit probe for a couple hours before you switched over to Probe1, so maybe the difference in behavior after your changed pit probe had more to do with the changes that happened to your fire while the pit probe was dead rather than the difference in pit probe type?

Click to expand...

Yes, ~2hrs from when the cold solder joint stopped my TC pit probe until I noticed and switched over (hopefully my neighbors didn't spot me, running in my pajamas at 4am around the patio ;-) ). My servo closed position was set to keep the vent a small amount open, that saved my fire.

Thomas K said:

That would mean, PID calculation and PID suggestion use different sources.

Click to expand...

Not if Probe 0 is not disabled...

I am not SURE this is the case, but the idea to have the pit probe flow down stream if Probe 0 ia disabled was mine, a secondary idea to my initial suggestion that we somehow be able to just select which probe functions as the pit probe. Bryan rejected that idea, which seemed pretty simple to me, but he explained there was a lot more too it code wise than it seems, so perhaps all the functions do not follow the pit probe when Probe 0 is disabled... Like, for instance, the Pit Probe display will show OFF rather than display the value of Probe 1 or whichever probe is controlling the pit....

Like I said, I don't KNOW this to be fact, but do know the above to be true and am speculating this may be a similar situation, where the funtion that calculates the suggested PID values was just hard coded to look at Probe 0 rather than actually follow the "Piit Probe".