h1pemmy is finally up and working.  The first problem relatively easily found was a bad ADC card, which was swapped early on after it was seen to generate 'channel hopping' errors to dmesg.  Said card has been marked as bad (it was probably pulled from somewhere else, but not actually labelled as bad which led to it's later unfortunate re-installation).  The second issue was that after replacing the ADC card the duotone signal was not appearing on CH31 (numbered from 0).  This led to a number of time consuming component swaps to eventually find that the cause was a bad cable from the internal 16 channel Contec card to the IO chassis backplane.  I could verify with the IO chassis powered up, but the front end computer off, that the duotone signal was present on the correct pin of the cable feeding the ADC.  After boot/code initialization, the duotone input was switched off by the relay on the adapter board, and the last channel of the AA chassis would then feed the ADC (verified by injecting a signal through the AA chassis).  I could also verify the relay control voltage with the Duotone0_select testpoint on the backplane.  The code is (in it's present form, in theory) hard coded to never activate this relay, so seeing 5V here is not right.  The cable itself is mis-crimped, such that pins 1/20, 2/21, ... , 18/37 are shorted together, with pin 19 not connected at the male/Contec side of the cable, a fact not immediately obvious in the back of a dark rack, but more so once exposed to the light...  The moral of this tale is to double check the internal 37 pin cable that controls the backplane if the duotone signal seems to go missing between the testpoint on the board and the ADC sometime after boot, there may be more of these floating around.