Differential measurement across NI Card's own ground referenced output.

[navneet]

Hello guys,

I have a question, our of pure curiosity.

I am using the USB6212 card to apply a 10V sine signal to a set up and then meauring the voltages across two loads. 

I am attching here a simple diagram showing the system. The set up can be modelled as a series combination of RC-R.

Note that I am also measuring the voltage R2 in differential mode. Now what I am after is accurate phase monitoring between the two voltages (across RC and across R2).

My question: Is the schematic correct? I mean is it OK to measure the voltage across R2 in differential mode with the bias resistors attached as such?

This is because one of the lines across R2 is already at AOGND. And AIGND and AOGNd are already connected inside the card, so will this introduce errors?

or does it not affect at all?

thanks guys, will be greatful for a speedy solution.

[johnsold · Accepted solution]

You can do that but differential mode does not gain you very much.  Your signal source is single ended. The voltage across R1-C can be measured differentially. It really does not make sense to measure the voltage across R2 differentially because one end of it is grounded.  The bias resistors, Rb, are not needed in this case because DC paths of sufficiently low impedance exist at all inputs.

What I would do is make two single ended measurements.  Measure AO1 on one channel (AI0).  Measure the R1-R2 junction on the other channel (AI1). Then the input voltage is AI0, the voltage across R1-C is AI0 - AI1, and the current is AI1/R2.  You will need to sample fast enough that the delay between measurements does not contribute excessively to phase error.  This will depend on your signal frequency and allowable errors.  Look at the graphs of settling time on page 2 of the NI USB-620x Specifications for more information about how trade-offs between speed, accuracy, and source resistance on multichannel measurements. 

Lynn

[navneet]

Hello Lynn,

Thanks for the discussion, as always 🙂

The schematic diagram I showed is very basic and there are extra stages of RC which should be modelled in the scheme. But to a first approximation, this is ok. hence I cannot (dont want to) use single endd measurements. Am OK with differential. the only reason I am using the bias resistors is to avoid any ghosting effects. 

I tried the following:

a) removed the bias resistors from R - does not make any difference.

b) removed ALL resistors, still doesnt make any difference, hence the use of bias resistors is not really required, but am still keeping it for any lone gun sample which shows up.

Most of my contact impedance (across RC) are low enough so I guess it doesn't contribute to any ghosting. nevertheless I dont want to go against theory. hence I am keeping the bias resistors for differential measurement across RC - does this contribute to any theoretical errors?

Phase responses look OK as I am only going upto a max of 1 kHz. I guess below this frequency phase errors due to offsetsare very low. i just want to be theoretically sound in using those bias resistors across RC (47 kOhm each).

[Esteban_B]

Hi navneet,

The circuit that you attach seems fine, also the measurements that you are trying to acquire. Is there an specific undesired behavior that you are seeing in your measurements?. I found this article related to ghosting and how to eliminate it, I hope this helps.

Regards,

[navneet]

Thanks Steve.

Infact all was ok, I was just curious about using bias resistors across the reference R.

Thanks again Steve.