How do I get rid of periodic spikes in USB-9201 output?

[dsimkins]

I've attached a variety of sources to my USB-9201 A/D and its output always seems to be corrupted by periodic spikes.  I've used both LabView Signal Express v2.5 and Matlab to acquire data with the same results.  The spikes are roughly 1/2 volt and sometimes alternate direction.  When the signal input level drops down below 1V they start to go away and aren't really visible at really low input levels (but the quantization is). Their period changes with sample rate but at 250KHz they are 12.3 samples apart.  I'm using only one of the 8 inputs.  Different computers in different locations (across town) give the same result and the time interval between spikes changes with sample rate, so it seems unlikely to be interference (unless it's a much higher freq aliased down).  The input looks clean on a scope, so I'm guessing either the A/D has gone bad or else there is something about the setup that I've missed.  Any suggestions?

[ryan_d]

Hi dsimkins,

Since the voltage spikes are periodic and you’re able to measure a clean signal with an oscilloscope, I think one of two things is happening:

1.       Your signal is correct, but needs to be filtered.

2.       Either the amplifier or ADC is damaged in your USB-9201

Your oscilloscope probably has some sort of filter on its input, which might be why you don’t see the voltage spike in that measurement. The USB-9201, however, does not filter data on its input. It looks like your voltage spikes are coming in at about 20 kHz. Try filtering this frequency by either setting up an external filter with hardware or by adding a filter step in Signal Express. You can add a filter in Signal Express by left-clicking on the Add Step button and selecting Processing » Analog Signals » Filters.

If the above doesn’t work, the amplifier or ADC might be damaged. You can determine if either of these is damaged by testing a known voltage on your analog input lines. Try testing each channel with a known voltage source, like a battery. Alternatively, if you don’t have a battery, short each analog input channel to ground. If you can, perform this test in a Measurement & Automation Explorer (MAX) test panel. If this test reveals that the applied signal does not correspond to the measured signal, it is likely your device is damaged.

Finally, I recommend testing your signal on each analog input channel to see if you can reproduce the voltage spikes. Please post back if you have more questions. Have a great day!