Pxi-4071 data acquisition reliability problem

[johngardner58]

Hi,

I am having a reliability problem using my Pxi-4071 in digitizer mode.  

I have a number of tests that use the PXIe-6363 analog output (usually configured for DC) to provide stimulation to our own device which has a number of ADCs.  We use the PXI system for calibration and test.

1.  I select a voltage from a range of voltages.

2. Program the PXI-6363 to drive this voltage

3. Delay about 10ms to settle.  Note there are no discrete capacitors or resistors in the circuit.  Everything is parasitic and would typically be below a nF and less than 10 ohms

3. Configure and Initiate() a data acquisition with the PXI-4071.  Typically I use a sample rate of  1000 S/s  and get about 30 samples (30 ms worth).  Triggering is immediate and I have used default settling time, 0 settling time and it doesn't seem to make a difference.

4. Measure voltage with ADC.  For debug purposes I have sometimes done this twice once before calling Initiate() and once after.  The after is normal.  The time required to measure the ADC is shorter than the acquistion time but regardless the stimulation by the PXIe-6363 is constant

5. Fetch the waveform.

6. Average the waveform and compare to the ADC measured value applying tolerances etc.

Here is the problem:  this works well most of the time.  But say 0.1% of the time (1 out of a 1000 acquisitions),  I get 8-12 samples that are near 0.   This looks like a settling time issue (on the surface) but regardless of the amount of settling time given I still get this behavior.  Not only that but when the ADC measures the voltage prior to calling Initiate(), it always confirms that the driven voltage is already settled at the programmed value.  Nevertheless the acquisition shows near 0 data.

So far our independent ADC always reports the expected voltage both before and during the acquistion (100%).   It is as if the DMM input is disconnected during the acquistion for a period of time, because we have confirmed that the correct voltage is already present prior to the start of the acquisition (calling intitiate).   I get no errors from either the Initiate or the FetchWaveForm calls.  I always get all my samples.  And 99.9% of the time it all works as expected.

The DMM and ADC are connected at the same point and both are referenced to ground and as I said before only parasitic capacitance and resistance (cable).  We do use a switch matrix (PXI-2530b) to make those connections.  We nearly always use 51/2 digit and 10V range for the data acquisition.

[johngardner58]

Additional information:

Sometimes the 0 values are not at the beginning of the acquistion data but at the tail.   It is almost like the Fetch is returning a different section of the acquired data because we have confirmed that the correct voltages are there before during and after the acquistion and yet sometimes the samples are near 0 volts for the beginning samples or the end samples.

BTW we are using C#

[User002]

What values are you expecting during the acquisition? And you say that the voltage is near 0? Are they the same voltage level for each sample, or do they vary slightly?

[johngardner58 · Accepted solution]

Hi,

I figured it out, and was going to repost but got distracted.

The device with the ADC also has a mux and it switches the mux to an internal node.   It only switches during the measurement and is open at other times.   So there is a race condition where the acquisition either starts too early and keeps acquiring after the switch is open.   Unfortunately I don't have the option for triggering.

I had forgotten about the internal mux as I had designed the test years ago and was doing some routine updates to improve the stability of the test.    That is why we start the ADC measurement during the acquisition.

I just added a routine to reject samples below a threshold