WebDAQ/100

[hpmaxim]

I bought a WebDAQ/100 several years ago off eBay and hadn't really played with it until now.  I am trying to use it to monitor several microphones (don't know what type) simultaneously.  I noticed some undesirable behavior and was hoping someone could clue in me in as to what's going on.

At the moment, I am running 2 microphones @ 5kSa/s in differential mode.  Eventually, I will go to 6 microphones, and might increase the sample rate.  The sampled signals looked pretty noisy and I could not detect any signal out of the microphones.  I cranked up the gain to 11 (actually, more like 100), and noticed that the noise seemed to decrease.  I am assuming that the signal was actually below the noise floor of the digitizer and the gain prior to the digitizer made things better.  Here are some observations:

1) There is a DC bias, typically between 5 and 40mV.  The gain doesn't seem to impact this, which implies to me that the bias is in the digitizer.  The bias seems kind of random though.

2) The gain seems to decrease the range of the digitizer.  With 100x gain, I am limited to a +/- 100mV signal.  I don't understand this.  The digitizer is after the gain stage so it should still be a +/-10V limit.  OTOH, the firmware may be dividing down the value by gain in software and displaying the divided down voltage.  If that's the case, then the aforementioned DC bias, is *NOT* from the digitizer but is actual bias on the signal lines.

3) The DC bias issue actually becomes a serious problem at high gains.  At 400x, the limits are now +/- 25mV.  With such a low limit, the digitizer gets maxed out and I get no signal much of the time.

4) There seems to be some crosstalk between the lines even though they should be differential.  I'm not sure if the cables on the microphone are shielded or not (they are part of an assembly).  If the lines are coaxial...  If they are, it seems that could make things worse,since the outer conductor could get impacted by crosstalk but the inner conductor may be shielded.  This would be fine if the outer conductor was tied to ground, but if its not, it could be worse than two parallel wires in the cable.

5) I tried a few things to reduce the DC offset.  I tried tying one line to ground and also tried running it in single-ended mode (with the other line tied to ground).  In both cases, this made the DC offset much worse.  In fact, at 100x gain, it was continuously railed at 100mV.  I have no idea why this is unless the digitizer doesn't like to be near ground.  It's hard for me to understand how "floating" lines could be preferable to tying one to a DC voltage.  However, if that's the case it makes me wonder if I want a series capacitor to remove DC bias.

Anyone have any experience with WebDAQ/100s or have any other thoughts about this DC bias issue?