Whats the best way to get an averaged measurements while scanning?

[JonJay]

I currently have a PXI 4071 DMM and a PXI 2575 Mux. I've played around with the examples of how to do scanning of channels which all works out fine. However the scanning only takes one measurement every time it steps down the scan list. What I need is for every scan step to take X number of measurements and average them before moving on to the next scan step. I've played around with the Sample Count and Trigger count but cannot seem to get this done correctly. I've gotten it to the point where it runs through the scan list multiple times and I can average from that but I don't want to tax the relays that much. Can anyone point me in the correct direction for this?

[Now_With_Underscores]

Hey Duke,

If your goal is to increase accuracy, simply increase the aperture time.  The 4071's FlexDMM architecture averages multiple points for each measurement; increasing the aperture time increases the number of points we average, even though on the software side we're only taking 'one' measurement. 

If you're trying to sample your signal piecewise over a longer time period, let us know; we could achieve multiple samples per scan list entry using counters.

[JonJay]

To be more specific about what I'm trying to do:

I am working on building/programming a new test station for our facility. I'm trying to replicate as much of the functionality of our older test systems as possible. The old test systems are using an out of date DAQ card that can only use traditional DAQ on it. Because of whatever reasons the previous programmer decided that when reading voltages he'd take 5 samples and average them. (Something about cutting out noise and timing.) Because of this I wanted to replicate this on the new system.

For the most part our single measurement acquisition on the new system is fine but I just stumbled upon a product that has a PWM output that, before, he read using the average and tested them verses a "average" limit. This caused issues on the new test system due to the single measurement reading either high or low and not an average voltage.

So now I'm trying to determine how easy it is to take average values using a scan list or if I should simply reconstruct the test so I measure the values properly.

[Now_With_Underscores]

Hey Duke,

I can say with great confidence that the 407x DMM is much more accurate than any DAQ product NI sells; if you're seeing unexpected accuracy errors after acquiring one data point with the 407x DMM, I think the best method is to figure out the discrepancy, not average it out.  I recommend you verify the DUT signal with a scope (you can use the DMM in digitizer mode) to see if the signal is still settling as you take your first measurement.  If this is the case, each successive measurement becomes much closer to the value you expect,  so averaging 5 samples allows the system to settle at the later points, and we get close to the actual value.  If this is the case, then I recommend you increase the settling time property (configuration»Advanced»Settle Time) enough that your DUT signal has settled and then take a single measurement (with increased aperture time, if desired, as mentioned previously).  Note that the value you obtain will differ from the old system if the above conditions exist... the 407x measurements are more accurate.

Obviously, if you scope the signal and it's steady during your first measurement, then my previous post would solve the problem while still taking only one measurement.  For my own curiosity, what type of measurement are you performing?

Keep us up to date on your progress and don't hesitate to post of with any additional questions.

[JonJay]

The measurements are standard voltage measurements from I/O modules. Nothing special about them nor any large amount of noise. I believe the former programmer was averaging the measurement due to a voltage divider he installed to halve the voltage into the old 0-10v range. Because of this he wanted to get a more "accurate" reading. I'll stick to taking one measurement and just increase the settle time if for some reason it is required.

Thanks for your input.

[Brandon_G.]

Hi Duke,

Please let me know if you have any more questions or run into any more problems.

Thanks!

Brandon G.