PCI 6123 AIGND connection and offset problems

[Cem Ozgur]

Hello,

I am using PCI 6123 with DAQmx. I use this device to make data acquisition from an isolated signal source. I made an interface PCB card which contains an antialiasing filter with a differential opamp (THS4131). This card only filters the input signal (-3dB point at 10 kHz) The gain of this filter is unity up to 3 kHz approximately.

Please see the attached document for the circuit schematic. I take samples with 102.4kS/s/ch. When I short circuit the supply (Vs), I measure approximately a constant 1mVolts DC on LabView. (It varis between 0.8-1.2 mV in fact) I also see a very constant 1mV with a FLUKE 189 multimeter whose probs are connected between the differential outputs of the opamp (THS4131). This value is in fact the offset voltage of the opamp and is within its specified limits. No problem.

However, when I start up the source, I see the 8V rms 50 Hz signal on LabView front panel, but this time with a changing offset voltage. The voltage changes between approximately 10mVs and 0mVs. (The larger the input signal, the larger becomes the offset max value) The frequency of this change is about 0.2 Hz. I again measure the offset voltage of the opamp in DC mode and see that it is 1 mV and constant. 

I assumed that there must be a reference (gnd) problem between my PCB and the DAQ card. I tried several different connection topologies for AIGND, but these did not change the result. The offset kept varying in a slow manner. I will need a very constant offset voltage and I will calibrate it with software before I take measurements, since this will be a precise accurate data acquisition system at the end. However I could not solve this problem and got stuck.

The suggested connection given in "S Series Manual.pdf" is (1) in the circuit_schematic.doc. This does not solve the problem. I tried 6 different connections but could not change anything. Would you please suggest on this issue? Where should I connect AIGND for such an input?

Regards,

Cem Ozgur

[dansch]

Hi Cem,

Thanks for posting! I understand you are having trouble with your readings from a PCI-6123 device. It appears that the first method you tried, of connecting a bias resistor between AI- and AIGND is correct for a floating signal source. Did you notice any differences in the signals you received when you tried some of those different connection methods? I am also wondering if you see this problem with the offset occur if you simply measure the voltage source alone. Also, do you have any way of measuring the output of the amp with an oscilloscope as well, to see if you see the same drift occur with the offset? Thanks, and I look forward to hearing your results.

Regards, 

[Cem Ozgur]

Dear Daniel,

Thank you very much for your reply. Today, I have worked on the subject once again and discovered some points. Let me share them with you.

First of all, as you have mentioned, I measured the signal source (input) offset by the FLUKE when it is not connected to anywhere and saw that the offset is constant for a specified voltage rating but not zero!  For example, if the signal source is 1V,rms, 50Hz, it has an offset of 1.2 miliVolts. Whereas at 5V,rms, 50Hz, it has an offset of 5.8 miliVolts. Thus, you turned out to be right on this. The opamp actually gives the output offset as =  this offset voltage plus its own offset voltage. This problem is understood now. Thank you for your advice.

On the other hand, I had mentioned that in one channel of my PCB card, the offset kept varying between 0-10mVs with a frequency of 0.2 Hz approximately for the full range input voltage signal. We unfortunately could not see this offset voltage on a scope, because we can not subtract the AC part from the total signal to obtain the DC only. (6-7V rms AC+10mV DC)  (The vice versa is possible, we can subtract the DC part from the total signal, to obtain the DC only) In a nutshell, we could not see the offset on the scope.

However, while studying we finally found the problem. It turned out to be something where different from what I expected. The offset swinging problem results from the DC measuring algorithms of the LabView. When a rectangular window method is used, the offset varies in a manner as I mentioned before. When Hanning method is used, the offset stays the same! (with 100-150 microvolts ripple) That problematic channel's method was left unintentionally on "rectangular type". So the problem was on LabView software, not about the interface card nor the DAQ card. A funny tragedy 🙂

To answer your questions:

I tried the different connection methods for AIGND once more, after we have found the problem. For the circuit whose general schematic I have sent before, I tried:

1) AI- is connected to AIGND with a 100k bias resistor,  2) AIGND is connected directly to PCB_GND, 3) AIGND is floated.

Only the output offset voltage seen from LabView  value changed a little bit (150-200microVolts) when (1) is applied. (2) and (3) gave the same results. The offset for example dropped to 1.0 mV from 1.2mVolts when (1) is applied.

However, when I connect the probs of the FLUKE to the opamp output became 1.2mVolts again also for (1). I guess that the loading of the Fluke changes things but I can not be sure. I will appreciate your comments on this. On the other hand, the rms value of the input signal and the opamp output signal and the rms value read from LabView screen all turn out to be same (with an error percentage of less than 0.1% which is reasonable). There seems to be no problem left now.

Finally, I decided to make the AIGND connection as given in (1).

I will be happy to hear any further suggestions and comments. I do not want to unforesee anything.

Regards,

Cem

[dansch]

Hi Cem,

I am glad to hear that things now appear to be functioning correctly using the correct setup. As for the difference in offset when the voltmeter is applied, it is highly possible that it is loading the input and causing the change to occur. However, even with this very slight change, the offset is no longer swinging and so it sounds as if there should be no further problems. Please let me know if you have any further questions or issues that arise. Thanks, and best of luck!