random DC offset on PCI 4472

[xred]

Hello,

One of my colleagues is experiencing some "random" DC offset while using PCI 4472 cards. By "random", I mean that this offset is different each time we reset the cards but, while reading the samples between two resets, the offset is constant. The value of this offset affects the 8 LSB's of the card (so only the 16 MSB's are "useful" and not affected by this random offset). The experience my colleague performed was the measuring of stable DC voltage sources as well as measing the ground.

Additionnal information:

• happens with both DAQmx and DAQ;
• happens with samples read both in Doubles and in INT32;
• we have not characterize the "randomness" of this offset but we suspect that it is not fully random: the offset seems to "hesitate" between several discrete values.

Any advice is welcome.

Thank you.

[xred]

Hello,

As a complement, here is an image illustrating our offset problem.

As you can see, the signal is really stable between two successive resets of the card, even for long periods of time (1 hour). 

But why does its level change so much when we stop, reset and start again the acquisition?

Thank you.

[Michiel_H]

Hi xred

Could you check the revision of the 4472 cards you are using? There was firmware bug in early versions of the 4472 cards causing an offset error. If you have a revision of the card prior to revision F, you are likely dealing with this issue. I've tried with a revision C PXI version of the card and I can see the same behaviour where the offset switches between device resets.

http://digital.ni.com/public.nsf/allkb/914D96F95777CBDF86256C05007774C0?OpenDocument

[xred]

Hi Michiel,

Where can I check the "part number" or the revision of the 4472 cards?

In NI MAX, I can find the "serial number" of the cards: 0x13C35C7 and 0x13C35D2.

These serial numbers do not follow the same format as the "part number" described in the URL you sent. So I guess these are not "part numbers".

Thank you,

Kind regards.

[Michiel_H]

Hi xred

The part number is printed on the PCI-card itself. I've checkd our db using the S/N you provided, and you are actually having a revision H of the board, so the offset behaviour described is not applicable here.

I will do some further investigation and keep you updated.

Message Edited by Gliderpilot on 09-29-2009 09:18 AM

[xred]

Dear Michiel,

Indeed, the "part number" on both cards is: 187611H-01.

We bought these cards end 2008.

Regards.

[Michiel_H]

Hi xred

I've done some further research on the offset. The behaviour we are seeing is actually normal and inherent to a DSA card's structure. When you reset the device, the card actually performs several functions. One of these is performing a self-calibration of the ADC's (this is a build-in function of the ADC itself, not to confuse with device calibration). This self-calibration might induce a small offset in the ADC itself. The offset however should not vary as the calibration is a function of the ADC itself.

Although this might seem inconvenient, keep in mind that the PCI-4472 is a DSA board. Because these boards are meant to acquire dynamic signals, in which most of the time a DC offset is not important, they have several specific features build in, like the highly accurate ADC's.  These features allow the card to achieve the very high 24-bit accuracy and other favourable specifications for measuring a dynamic signal. Unfortunately, these specific features have some drawbacks, such as the small offset you might see on your card.

This is also the reason why the specifications of your card contain an item 'DC-coupled offset', in case you do measure a DC signal with the card. In case of the PCI-4472, it is spec'ed at +/-3mV max(see p.2 of the cards's specifications: http://www.ni.com/pdf/manuals/373861d.pdf). So as you can see, the card is actually performing well within its specifications.

So to conclude: there is, unfortunately, no straight forward solution to this offset behaviour when measuring a DC voltage. Simply because the card is not ment to do this type of measurements. If you do want to measure a DC voltage, you will most likely be better of with an M-series card. Or even better if you want a very high accuracy, a Digital Multi-meter (DMM).