Peculiar multichannel analog read problem with PCI-6014 -- hardware issue

[John_P1 · Accepted solution]

Hi Diane,

50 kOhm would be considered high for an output impedance.  This is almost certainly what is causing the problem.

Essentially the problem is that the ADC has a sample and hold circuit (wikipedia link) to keep the voltage stable while the measurement is being taken.  However, multiple channels are being switched to the same ADC.  So... after a switch takes place, the charge that is stored on the capacitance of the ADC needs to be dissipated across your signal source (remember, the input impedance of the ADC itself is in the 100 GOhm range).  If the source impedance is too high, then this takes too long and you will experience ghosting.  The following image shows an overview of what's happening:

       

Here's a graph from the 6221 specifications that shows the settling time of the inputs for various source impedances (unfortunately I don't have such a graph specific to the 6014, but the overall behavior should be similar):

       

50 kOhm is off the chart, but you can see that the settling time dramatically increases for higher source impedances.  Ghosting is the result of not allowing for enough settling time between consecutive channels on a multiplexed DAQ card.

So, given that you need to achieve higher rates, in this case your best bet would be to use an external op-amp buffer to reduce the impedance of your signal source.

Best Regards,

[JÞB]

Excellent information John.

Diane- sounds like you just made the case to move to a simultaeous sampling DAQ

[John_P1]

Simultaneous is another option (especially if you're purchasing a new DAQ card), but it can sometimes be a bit costly for some applications.  If you don't mind building the circuit then external buffers would be a good option to eliminate ghosting.

Best Regards,

[DianeS]

Hi guys,

Thanks again for your help and input!  John, your explanation and the graphics were really helpful for understanding the problem.  Thank you!

We are redoing the board layout, so adding another op amp buffer is no big deal.  I expect that's the way we'll go.  In the meantime we can kluge something up and see what happens.  (That old board is full of kluges anyway, what's one more?)

I will mark John's post as the solution as soon as we get that kluge in place and are able to increase our per-channel sampling rate.  Thanks, John and Jeff!  As always, your help has been invaluable and is much appreciated.

d