Temperature error caused by print fan
I found a Slic3r setting file supposedly good for TPU (via a link on one of the Facebook forums for Prusa printers), but it caused me many problems. One of the worst was that it set the print fan at 100% speed. This apparently is not good for the MK3, especially on the lower layers of flat objects.
The object I was printing was a cellphone case, which has a lot of flat area right on the bed - i.e. there's no place below for air to escape from the print fan, so the cooling air intended for the print ends up going up through the cowling around the extruder and cooling the Hot End itself. I literally got a print shutdown due to "TEMPERATURE ERROR". Using Repetier, I could see that the temperature of the Hot End was running around 240 when it was supposed to be 250. I finally solved the problem by reducing the fan speed to 50%, which is the value in the default Prusa print files for PET. Then the Hot End temperature matched setpoint.
The print isn't finished as I write this, but so far so good. Besides cutting the fan speed, the other key (of course) is limiting the print speed.
Re: Temperature error caused by print fan
"TEMPERATURE ERROR"
I'd be really concerned about your thermistor connections. You should not have any trouble at all maintaining temp at 100% fan up to 290C or so.
I'd do a PID tune with the fan turned ON.
Re: Temperature error caused by print fan
I saw no evidence of broken wires in the temperature curve the other day. So today I ran another test. I had the print fan on at 100% for the whole time. This time, I was not actually printing anything so the thermal load on the Hot End was much lower than would be the case during an actual print. I started with the extruder 20mm above the bed. As you can see from the graph, the temperature rise from ambient to 255 is indicative of a reasonably well tuned PID loop, and once it reached setpoint there was quite a bit of reserve heating capacity.
At T = 22:00 on the graph, I lowered the extruder to 0.2mm above the bed. You can clearly see that the temperature dropped briefly due to the change in airflow pattern. In this case, the temperature did recover to setpoint, but you can also see that the amount of power required to maintain this setpoint increased from roughly 60% to roughly 90% of maximum power. So when the added heat load of melting plastic is taken into consideration, it fully explains my temperature drop the other day.
The bottom line is, keep an eye on the temperature of your hot end if you have your print fan set to anything more than 50%, and be prepared to slow down the fan if necessary to keep the hot end at your setpoint.
Knowing this, I successfully printed a phone case in TPU, including a small bridge, and the 50% fan speed seemed to work out just fine for the bulk of it (the bridge was printed at 80% fan speed - but it spent only a few seconds there).