NI 6210 Analog Input Drift

[yoda8579]

Hello,

I am using a 6210 to measure the voltage output from an SGLux Transimpedance Amplifier connected to a Thorlabs FS1010 photodiode (Ai0 in differential mode 0V-4V). The amplifier board is powered by 24VDC from a BK Precision 1685B switching mode power supply.

Transimpedance Board:

http://in.element14.com/sglux/tw-mf2cab/amplifier-board-photodiode-2-ch/dp/1209955

When I first start the acquisition, everything works fine (i.e. I get a stable output from the board of around 700mV). After about 1 hour, I start to see spikes down at ~0V and this gradually worsens until the entire signal is down around 0V (see images below). Checking the outputs with my Fluke 179 voltmeter reveals that the measurement is actually ~700mV but that the DAQ is reporting the wrong information. Hot swapping to a different analog input and I get the same signal. If I disconnect the DAQ from the USB port and then restore power, (takes about 20s), the signal is recovered and stable again for 60min... Then the cycle repeats.

Any ideas???

Initial Waveform

60min Waveform:

90min Waveform:

[yoda8579]

I forgot to mention that the data in the plots is 100mS in duration.

[johnsold · Accepted solution]

So, the interfering signal is at 60 Hz. If you are in the U.S., then you likely have power line interference and perhaps a ground loop.

Do you see the same drift if you generate a signal with a resistor voltage divider? Connect 4700 ohms from the +5 V line of the DAQ device to the AI pin. Connect 1000 ohms form the AI pin to AI Gnd. That will give you ~0.9 V. Watch that for a while to see if it drifts.

Lynn

[Henrik_Volkers · Accepted solution]

You wrote you measure differential..  do you have a bias path?  I assume the output impedance of the amp is low enough to add 2 1M resitors to AGND at the diff input.....

BUT a 1h drift sound more like setpoint drift in the amplifiers due to temperature or again a amplifier bias /ref point is drifting outside it's CMRR range ? ...

[yoda8579]

Thank you both for the insight! I can't believe I didn't notice the cyclic nature of the noise at ~60Hz!

I have added 10Kohm bias resistors to the + and - of the signal to AIGND and I'm going on 2 hours without any issues. I'll continue to test a bit longer and then consider the matter closed!

Thanks again for the help!

[johnsold]

Without bias resistors input capacitances can charge, changing the DC component with respect to AIGND. This produces the drift. When the total signal (DC bias + 60 Hz + desired signal) gets very close to saturation or cutoff, strange things can happen such as the apparent increase in the 60 Hz component. 

Very likely the bias resistors will solve the problem.

Lynn