tecnico Posted August 5, 2012 Share Posted August 5, 2012 Hi all, I have a very strange problem with OPT12(A) of Turbo PMAC PCI. I'm installing 3 of these boards as a replacement for the older Turbo PMAC PC (ISA BUS). The old boards were acquiring analog signals through ACC-36P, but now we rely on OPT12(A) to get those analog inputs (even if they accept 0-5V instead of 0-10V). I followed the instructions on the manuals and I got to this configuration for the automatic A/D processing ring (0-5V setting): I5060=8 ;line 13 I5061=$78808 I5062=$78808 I5063=$78808 I5064=$78808 I5065=$78808 I5066=$78808 I5067=$78808 I5068=$78808 I5081=0 I5082=$1001 I5083=$2002 I5084=$3003 I5085=$4004 I5086=$5005 I5087=$6006 I5088=$7007 After succesful tests in my office I went on the field... The first machine runs fine, but the others give me random numbers (most of the times 4095) on the converted analog inputs. Since the software is the same I'm a bit puzzled....it looks like the processor can't access the ADC or that the conversion can't be triggered... Any hint? Andrea Link to comment Share on other sites More sharing options...
tecnico Posted August 6, 2012 Author Share Posted August 6, 2012 Update: after further checking it looks like a ground problem. The tests in my lab were made providing analog power supply from the bus, while on the plant analog power is external and isolated (E85, E87 and E88 are OFF). According to the manual this shouldn't be a problem (it clearly states that ANAIN GND is not isolated from digital) but testing the board I found it only tied to AGND; moreover on JANA there is a +/-12V output that comes from the bus but the power supply for the ADC is again tied to analog supply voltage.... Testing for voltage on the field I found that in the machine that works fine AGND is tied to DGND but in the "bad" one I read about 1.2V difference. At the end it seems that the only option is to connect AGND and DGND, but I would like a confirmation from those who actually have the schematics and have designed the product. Link to comment Share on other sites More sharing options...
Sina.Sattari Posted August 6, 2012 Share Posted August 6, 2012 Andrea, I have checked the schematics and you're correct as far as the analog circuitry is being tied to AGND. All the analog input circuitry is being supplied by the analog input AGND and +15V. However, the manuals note on lack of isolation on analog inputs is because of data lines from ADC to DSPGate. This means when you have the OPT12 or OPT12A, effectively you have to connect the GND to AGND for your data lines to transfer the data properly. Regards, Link to comment Share on other sites More sharing options...
tecnico Posted August 7, 2012 Author Share Posted August 7, 2012 Hello, this explains why, to make the ADC work in a setup with no 60 poles cables connected (just for tests) I had to close jumpers E85 and E87 (e88 not needed since the ADC take -12V directli from the bus). I wonder why this choice to tie the ADC to AGND forcing the customer to connect GND to AGND thus defeating the isolation.... BTW the inputs look pretty noisier than the ones on the "old" ACC-36. Can this be related to the fact that at the moment the link between AGND and GND is made externally (on the 60 poles terminal blocks)? Would it be better just to close E87? Or might it be an effect of the defeated insulation? Out of tree boards I managed to have 2 working (even if a bit noisy), the last one (the one with OPT12A) only replies with 4095 (or 2047 if I call for bipolar conversion). Even putting it in the same environment of the good ones doesn't give results, so I'm afraid it's faulty... Link to comment Share on other sites More sharing options...
Recommended Posts