NI 6229 PCI card, strange glitches

[vic_pash]

Hi,

We found the strange simultaneous glitches on the many analog input channels of the NI 6229 PCI card during a measurement procedure.

See details in the attached file.

[John_P1]

What you're seeing is referred to as ghosting. 

The issue is that the 6229 is a multiplexed board--the charge stored on the ADC from the previous channel's acquisition has to discharge through the external source impedance of the current channel being sampled.  In your case, the external impedance is a 100 kOhm resistor.

The specifications page for your device actually has a graph that shows settling time for various impedances.  The graph doesn't go all the way up to 100 kOhm, but you'll notice that for higher source impedances the settling time is dramatically increased (note the logarithmic scale).  The amount of time that you are allowing to settle between channels is the inverse of the Convert Clock Rate.

       

Best Regards,

[vic_pash]

Thanks for your answer.

Do you have other similar DAQ card without ghosting (impedance is 100 kOhm)? Our machine is already done and it’s not an easy task to eliminate the glitches.

Many thanks

Victor

[John_P1]

Hi Victor,

Now that I look at it again, I'm actually a bit confused on what the plots in your attachment respresent and what you are trying to do in the first place.  It looks like you just have a 100 kOhm resistor between AI8 and GND (also between AI24 and GND).  What are you hoping to measure?  It looks like you're acquiring from all 32 AI channels in your VI, are these channels connected to anything?  The diagram only shows 3 channels connected.  You have a plot of Voltage vs. Time but you are running a SW-timed task, where did this plot come from?

I'm still pretty sure that the underlying cause of whatever behavior that you are seeing is ghosting.  Ghosting is going to be a possible side-effect on any multiplexed board if you are measuring across such a high source impedance.  If you allow more time between channels by decreasing the convert clock rate you can reduce/eliminate the effects of ghosting (see the suggestions in my previous link).  Simultaneously sampled boards have a dedicated ADC for each channel and thus would not be susceptible to ghosting.

Best Regards,