MCM-105 was originally tested 12-Jul-99. It failed because "there is a block of 32 dead channels (5th of 8). The 7th and 8th blocks have weaker responses (smaller than normal slope)." The block of dead channels is considered a fatal flaw. The weak response is not a serious problem.

On August 5-6, 1999, it was tested again. I tried varying every DAC value, even the irrelevant ones, to get the 32 dead channels to work. These channels always return 3FF (overflow) as the ADC value. The details of these tests are on pages 51-53 of the logbook, but I will summarize the results here.

Our normal setting for the Vref DAC is 2.97. Our normal setting for the Iref DAC is 7.40. I tried many combinations of Vref and Iref, including Vref changed from 0.2 (min) to 4.8 (max) and Iref changed from 2 (min) to 70 (max) with a number of steps between each extreme. In every case all ADCs in the 5th block were 3FF. The most significant of these combinations was Vref=4.80 (max possible) and Iref=70 (largest, i.e. most negative, slope possible). If the DACs and the PECL clocks work correctly, then this combination covers the maximum possible range of input voltages to the ADCs -- it should result in a non-zero, non-overflow ADC value for any normal input voltage. I also tried raw mode (most tests are in correlated mode) -- in that case both samples returned 3FF.

Aug-19-99, after replacing some surface-mount components and checking for shorts in surface-mount components, retested.

Both the preamp and AMU spy lines (MUX1 and MUX2) worked for the 5th chip. The signals in the MUX's looked normal and responded correctly to the DACs I tested including Vcal, Vcorr, and Vfb. Since the AMU output is normal, the problem is almost certainly with: