• State
  • Locked Locked
  • Replies 2 replies
  • Subscribers 33 subscribers
  • Views 787 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

PCM4220 outputs wrong code under certain conditions

MKARGER
Prodigy 55 points

Other Parts Discussed in Thread: PCM4220, THS4521, OPA1632

We have an instrumentation application that uses several hundred PCM4220 ADCs, and we are seeing several of the ADCs output the wrong codes for a given input signal.  If the PCM4220s are powered on and immediately presented with a full-scale 5.6Vpp differential calibration signal, several of the converters will output codes roughly half-scale.  Within the same dual ADC package, one channel may output the correct codes, and the other may output half-scale codes, even though the input signal is identical at both inputs.  Switching out of calibration mode and sampling transducer data doesn't fix the problem, but if we hit the transducer with a large impulse after switching out of calibration mode then the ADCs begin to function normally. HPFDR/L are both hard-wired to GND.  The 4V and 3.3V power supplies are rock-solid.  The firmware may be allowing the calibration signal through before the ADC completes it's initialization reset sequence.  Could this be causing our problem?  The problem seems so random, but certain channels definitely get into this state more often than others.  Any other suggestions?  We're supposed to be going to production this week, so any hints are greatly appreciated!  Our firmware guy is working on a fix to enforce zero input signal on the ADC inputs until after they complete their reset sequence, but I was hoping somebody could confirm my suspicions before he's finished...

  • 0
    Prodigy 55 points

    I have some more information to add.  Enforcing a delay between the time the ADCs come out of reset and the time we begin to send a full-scale calibration signal in order to allow VREFL and VREFR to stabilize helped dramatically but still didn't eliminate the problem.  We're only testing 90 channels at the moment, and we still observe this behavior consistently on a few stubborn channels (the same channels).  Second, I did a much longer oscilloscope capture of the calibration signal at the ADC inputs, and the common mode voltage seems to oscillate just enough to cause the calibration signal to exceed the full-scale 5.6Vpp differential input voltage.  I'm using the VCOML and VCOMR pins to directly set the common mode voltage of a THS4521 fully differential op amp without buffering the pins through a voltage follower, so I'm wondering if the VCOM pin of the THS4521 is loading the VCOM pins of the ADC, causing the VCOM to vary with signal amplitude.

    At any rate, if I reduce the gain of the THS4521 LPF stage, thereby reducing the amplitude of the CAL signal below full scale, the problem seems to go away.  So what I seem to be seeing is that if the input signal to the ADC strays beyond about 6Vpp differential, then the ADC outputs half-scale codes instead of all 1's.  I've never seen that behavior in an ADC before.  Is something I'm doing causing the sigma-delta modulator to become unstable?  Is this a known problem for which there is a workaround (other than the obvious one of restricting the maximum differential input signal)?  The PCM4220 datasheet indicates that the ADC should be stable if you don't exceed VMAX on the analog inputs, which is -0.3V to 4.3V.  Any insight is appreciated.

  • 0 in reply to MKARGER
    Prodigy 50 points

    I have also encountered the same problem. Although I have buffered the VCOML and VCOMR pins to set the common mode voltage of a OPA1632 fully differential op amp, but it does not help. This problem is found in batches of 0AAPN4T and 0AA4VIT, but never found in 19CENYT, so it seems that only parts made before 2011 would suffer such problem.

    I'd really appreciate it if someone could confirm my suspicion.