Update temperature tests
The pass/fail criteria of the temperature tests needs to be adjusted. Currently the cuts are too tight and random temperature variations can cause the tests to fail.
The adjusted cut values should be high enough to detect a disconnected wirebond.
PTAT adjustment
The average value is very close to the 700 count switch over point. This causes problems in some AMAC's that have a large discontinuity. Would still like to maintain some sort of overheating detection.
New cuts: PTAT is between 500 and 900 counts. Note that the 200-500 count range for compatibility with FEAST is still there.
NTCpb adjustment
The average value is very close to the edge of the AM and has a wide spread. Thus a lot of boards just saturate this reading.
There are two ways to fix this:
- Cut only on the lower bound.
- Adjust the
NTCpbSenseRangesetting to 7. The current setting of 4 is used for consistency with AMAC probing results. It does not matter as much for the NTCpb, which is now connected to an "uncontrolled" NTC. However the NTCx,y do use a fixed resistor.
New cuts: NTCpb uses NTCpbSenseRange=4 and is between AAA and BBB counts.
NTCx,y adjustment
The average value is very close to the 700 count switch over point in the AM. A few readings, especially in NTCy are affected by this and jump to 800 counts.
One can lower the reading by changing the SenseRange switch. However this would prevent comparisons with the AMAC probing data, which use SenseRange=4. The range should be increased instead.
New cuts: NTCx,y is between 600 and 850 counts.
CTAT adjustment
There appears to be a second peak above 500 counts. The reason still needs to be understood.
It is not due to the 700 count switch over point as all values are far from it. Also it causes values to jump to ~750 counts.
The boards in this peak have also other issues (ie: non-linear CAL ramp). Thus they are problematic. No action should be taken here.