Noise from daq signals sent through an RC circuit

[gimNPC]

Hello,

I am trying to source various AC and DC voltage signals across a device that has resistive and capacitive elements. I am currently doing some testing on very simple circuits - a square wave bias out of an Ao channel from a DAQ 6368 goes across a 1Mohm resistor, then a current to voltage converter sends a voltage signal back into the 6368 Ai channel.

With just a resistor everything works well - accurate voltage/current readings, low noise, etc. But when I put a capacitor in series with the resistor, I get intense noise across a very wide range of frequencies. A 10pF capacitor doubles the noise, and a 100pF capacitor multiplies it by 10x or 15x. The circuit I'm interested in measuring will have about 300pF of capacitance, plus another 300pF from the extra wiring involved.  

Am I doing something obviously wrong, does capacitance inherently screw up amplifiers?

I feel like I should be seeing nice exponential decay or phase shifting from a capacitor, not broadband/HF noise. Take a look at the pictures I've attached with the fft included - they are taken from a circuit under 0 V DC bias, showing the current measurement in units of microamps (I recently posted a similar question one the switch forum, and have now determined that the problem does not come from the switch, but from stray capacitance in the switch).

Thanks!

EDIT: added some pictures of 10hz 100mV square wave pulses, the while line is the current time trace. The 1Mohm resistor one looks great - 100nA for 100mV makes great sense, low noise, etc. But with the 10pF capacitor, the capacitive spikes make sense, but whats up with the increased noise?

[gimNPC]

only lets me attach 3 things per post

[johnsold]

Please post an image of your schematic diagram.  I am not sure exactly how you have everything connected.

Lynn

[gimNPC]

Hello again, Lynn!

I've attached a schematic.

[johnsold]

Assuming that the X-axis scale the data you posted is in seconds, the signal you see in "resistor and capacitor" is 60 Hz with a small 120 Hz component. The spectrum shows multiple harmonics. I am guessing that the the X-axis on the spectrum graph is in samples, not frequency (possibly the scale is 100*frequency).

What do you get if you replace your current to voltage converter with the world's simplest current to votlage converter, a resistor?  If you use a 1000 ohm resistor, it will not disturb the performance of the RC circuit very much and will eliminate any dynamic effects of the converter.  

What is the input impedance of the converter over frequency?  The typical op amp current to voltage converter works well at DC but due to several second order effects may not handle the AC components with the same conversion factor.

The spikes on the edges of the square wave make sense because the capacitor passes the edges.  If you expand the time scale around one of the edges I would expect you to see the exponential charging curve with a time constant of 10 or 100 us, depending on which capacitor you are using.

The current sensor ideally acts like a very low (even zero) value resistance followed by an amplifier. Thus, the series capacitor acts like a high pass filter, passing the power line frequency and the edges of the square waves.  At DC the transfer function (with 1000 ohms and 1000000 ohms) is ~0.001. At infinite frequency the transfer function is 1.

Power line frequency interference is common. It can be difficult to identify the mechanism by which it couples into your system.  The facts that it looks sinusoidal (as opposed to rectified or sawtooth) and has a visible second harmonic in the time domain trace tends to point to something other than the power supply but still with a non-linearity. The spectral broadening of the higher harmonics suggests multiplication which may also be from a non-linearity.

I would not think the DAQ device would have a power line signal like that. A ground loop involving the power supply for the current to voltage converter is possible. 

Lynn

[gimNPC · Accepted solution]

Hi Lynn,

THanks for all the tips - the noise was much higher in frequency, up in the 10's of kHz region. I eventually found that my noise problems were caused by either ground loops or impedence mismatching between my PXI and my bipotentiostat, so I am trying to do the experiment entirely using PXI components.

Best,

Gim