Why are readings inconsistent?

[JeffinCA]

  I have a USB-6341.  I am using it to read two analog voltages (the AC-coupled outputs of an H-Bridge Driver) at 100 KSPS.  The source impedance is an opamp (an LT1638 buffering the signals).

  The readings are inconsistent, varying as much as 5.8% on one of the two signals.

  I know from reading the Calibration section of the manual that I have to let the unit warm up, so I had it on for hours before making the tests.  The auxiliary board where the opamp is located was also on for a long time.

  I wrote a data reduction program that does the following:

• Grab 75,000 samples at 100 KSPS
• Divide the data stream into 600 sets of 125 samples (one cycle of the waveform)
• Get the Minimum and Maximum values in each cycle
• Use those Min and Max values to calculate the average Low and High output levels.

  I then repeat the above sequence 500 times and calculate the average of those averages, as well as the Min and Max values.  This is what I used to determine the error.

  Has anyone else encountered a similar problem?  What was the resolution?

Thanks in advance,

Jeff

[Danny_Funk]

Jeff,

Can you describe the signal you are receiving, and exactly what it is coming from? Also can you provide the most simplified version of the code that does these calculations?

Regards,

Danny F

[AnalogKid2DigitalMan]

What are your results if you connect both DAQ channels up to the same signal source?

Your differences could be due to different charactereisting between the H bridges or op amps.

-AK2DM

[JeffinCA]

  As it turns out, there were two problems.  One is the barely-documented recommendation to let the USB-6341 warm up before making readings (found this in the section dealing with Self-Calibration).  The other was the signal source (an H-Bridge driver whose on-board power supply isn't as rigid as one would surmise, even though it looked rock-solid on our scope).

  I let the DAQ warm up, and replaced the driver with a power supply.  This allowed me to use a 34401A instead of a (non-digital) scope for more precise source measurements.  The resulting sample sets from the DAQ had a maximum deviation of +2-3 mV, considerably less than the 20-100mV I had been seeing.

  So... the system is calibrated and my customer is satisfied with the precision of the results.

  Thanks for the interest in helping with my problem.

Jeff