plausible heat profile?

Is this heat profile plausible? It's not at all what I expected.

power temp
31 49
47 74
63 74
79 69
95 111
111 111
127 111
143 111
159 171

I've had similar values on my extruder when I went with the stock firmware and recommended capacitor. When I started looking into it, it looked like that was about as good a result as I could expect with that hardware.

To measure temperature PIC measures the time it takes to charge a capacitor through the thermistor. When the temperature is higher, it takes less time to charge. As a result, the PIC can measure low temperatures with good precision, because the resistance is high, in the neighborhood of 10K. When you're over 100C, you're only seeing resistances of 1K - 400R or less, and the charge times get much faster, to the point where the smallest difference in charge time the PIC can measure corresponds to 20 or 30+ degrees.

The only way to get acceptable resolution in that range is to use a bigger capacitor. That can produce better readings at extruding temperatures, but it makes for much longer readings in the near-room-temperature range. Due to the way readings are done, longer readings fail more often, so you might never get valid readings at low temperatures.

As for solutions, I didn't find any way to get good results with the stock firmware and host software. Obviously people are using it and making stuff, I just haven't figured out how. The two solutions I've found are:

* use the stock firmware, put a fixed resistor in parallel with the thermistor to make its output more linear (suggested by nophead, documented on his blog), increase the size of the capacitor, and modify the host software to compensate for the parallel resistor. On paper, this sounds like it should work, but I got very poor results with it -- I kept getting invalid readings from the firmware. or,

* switch to the alternate firmware Andreas posted, and increase the size of the capacitor. This has been working very well for me -- I get accurate readings through the entire range of temperatures and no invalid readings.

IMHO, the second option is the way to go. Hopefully his firmware changes, or something similar, will make it into the standard firmware. Messing with the temperature readings has set me back a few weeks, and it looks like you're headed towards the same set of problems.

Here's a typical heat profile for me. I've been messing with the profiler dialog to give me something more interesting to watch while the heater does its thing. Ignore the time column, those values are wrong, you can read the actual times off the graph. When I've spot-checked it with my thermocouple, the temperatures match within 3-5 degrees, which I figure is good enough.

Thanks for the advice. Looks like Andreas' firmware is the way to go.

Has anyone tried using a thermocouple + OpAmp? Don