I know the measurements I've taken are good, and the MAT sensor is within .5C of the K-probe. BUT the coolant sensor is still reading about 4C too low: 13C instead of 17C tonight.
One assumes that Easy Therm hasn't generated an effective curve... so what's the best way to produce one? I'm guessing manual R to ADC conversion + propertherm.pl?
Temperature Calibration
Temperature calibration seems to be reading low. The air sensor is [MSD 2320] with exposed thermistor (100k-50 Ohms), the coolant sensor is a [Land Rover] part.
Some new data, measured by slowly warming a pot of ice until it boiled. This took over an hour, so I'm sure the sensor was keeping pace with the fluid. Measured with a DVM and thermoprobe, and a DVM on the sensor, this curve looks *much* better when graphed in OpenOffice.org. Voltage for ADC calc was measured to be 4.90V at the sensor connector...
| Temperature | Resistance | ADC (Calc) |
| 0 | 5690 | TBD |
| 5 | 4670 | |
| 10 | 3700 | |
| 15 | 3040 | |
| 20 | 2470 | |
| 25 | 2020 | |
| 30 | 1630 | |
| 35 | 1360 | |
| 40 | 1145 | |
| 45 | 940 | |
| 50 | 795 | |
| 55 | 675 | |
| 60 | 580 | |
| 65 | 480 | |
| 70 | 417 | |
| 75 | 358 | |
| 80 | 307 | |
| 85 | 263 | |
| 90 | 234 | |
| 95 | 198 |
The following points were measured, using a digital thermometer with K-thermocouple, and a DVM for resistance:
| Temperature | Coolant | Coolant ADC (Calc) | Air | Air ADC(Calc) | Notes |
| -19.8 | 11170 | 205 | 20930 | 226 | Freezer |
| 1.5 | 5490 | 171 | 9070 | 197 | Fridge |
| 7.1 | 3910 | 151 | 6030 | 176 | Fridge |
| 24.0 | 1950 | 107 | 2743 | 129 | Room |
| 40.6 | 1020 | 70 | 1310 | 83 | Oven |
| 56.0 | 559 | 44 | 696 | 52 | Oven |
| 68.0 | 385 | 32 | 445 | 36 | Oven |
| 89.0 | 208 | 18 | 223 | 19 | Oven |
| 94.0 | 192 | 17 | Kettle |
Previous readings...
| Coolant | Air | ||
| Temp | Resistance | Temp | Resistance |
| -16 | 6650 | ||
| 1 | 5420 | ||
| 3 | 5430 | ||
| 7 | 5370 | ||
| 20 | 2318 | 20 | 3190 |
| 21 | 2023 | ||
| 71 | 480 | ||
| 86 | 242 | ||
| 88 | 297 | ||
| 96 | 203 | ||
The air sensor looks like a common NTC part.
I have a TVR Griffith that seems to use the same sensor. I can confirm your readings from your tests are correct, as I've done the same and got the same answer.
I would be interested to know what your coolant calibration in your config is (or what Easytherm says it should be!)
Thanks
David Blades
The .inc I generated with Easy Therm was reading 4C too low at 18C. It's available [here] for reference anyway. I graphed the points and the curve was perhaps slightly suspect, one point anyway? But it doesn't have ot be out by much to mess up the lower temperature values by a few degrees... so I'll go back and do it again with more points, remove outliers, measure aref, etc. To be fair it's much less of an issue than with the MAT sensor, but I still want to get it right.
WBO2 Voltages
Got it sorted after reviewing several other configs. Nothing significant to say, other than to suggest copying Marcell's WBO2 set up to start with...
| Circuit | Target | Voltage before | Voltage now |
| GND -> Pin 5 (pump) | 3.96 | 4.28 | 3.96 |
| GND -> Pin 1 (nernst) | 4.41 | 4.39 | 4.41 |
| Pin 1 -> Pin 5 | 0.45 | 0.11 | 0.45 |
Got the hardware completed. The issues seem to be getting it set up as described [here].
I took base settings from the above page, and found that 0x63 was closest to 0V for my board (negative 0.109V). Then inserted 100 Ohm resistors into the socket and took readings (nerst_dc_voltage=8D).
Adjusting nernst_dc_target can change nernst value, but pump is still too high, and difference too small. The linked page states "The product of wbo2_pump_pid_ki and wbo2_pump_pid_ilimit should be around 0x4000" but linked values of 0x20 and 0xFF give 0x1FE0 already? Which is more correct, and how can I use this to increase the pump voltage (assuming nernst_dc_target can reduce nerst voltage independently). I think increasing wbo2_pump_pid_ilimit should work to raise the pump voltage? Or should I increase wbo2_pump_pid_kp instead? (after reading PID links)