How to correct Bode Analyzer behavior with RC circuits

[drneil]

We are running a simple experiment, measuring the frequency response of a low-pass and high-pass filter using the same resistor (1 k oms) and capacitor (10pf).  Not only do we get erroneous phase results, the gain plot is also incorrect, giving the wrong corner frequency.  We also found differences depending on how we ground the circuit.  We saw similar results with a LRC circuit.  We are running the latest ELVIS software (4.5) and brand new hardware!

[Cavarval]

Hello drneil,

Can you attach an image of your results to see what is exactly your issue?

Also, it would be great to know what exact hardware you are using (ELVIS, ELVIS II / II+), your wiring and circuitry setup, as well as your software configuration. I could use this information to try to reproduce the behavior you are getting so I can try to look for a solution.

[Mark1955]

10pf and 1K ohm has a cutoff frequency of around 16 MHz. Even with the correct hardware your wiring methods will make a difference. I too would be interested in what hardware and wiring setup you are using.

[drneil]

There was a typo.  Our component values are 100k ohms and .01uF.  So that gives a cut-off frequency of ~159 Hz. Correct?  Not what we are seeing in the Bode plot.

[drneil]

I am actually less concerned with the corner frequency than I am about the error in the phase plots and the "bumps" in the gain plots.

[Mark1955]

Here are plots using myDAQ jumpered to a breadboard with the 100K & 0.01uF cap. Interesting little bump for the phase on the high pass filter but the board capacitance using the 100K might be the cause. How does this compare with your results? Also have you tried a 10K with 0.1uF as I suspect the bump I see would not be so pronounced using those values.

[drneil]

Here is what we are seeing.  I agree that the 100k might be the cause, but it would be helpful for NI to simply state that.  I have attached two plots of what we are seeing.  One is similar to yours for the gain, but the phase is much different.  There may also be a grounding issue, as shown by the other plot where we grounded the circuit to the ELVIS board GND which totally destroys the gain plot as well.  While the acquisition channels are differential, it wasn't clear if the function circuit input from the ELVIS function generator needed to be grounded or float.  We are using NI ELVIS II+ and rev 4.5 software.

[Cavarval]

As far as I uderstand, the signal definitively needs to be grounded. As for why you get the shown behavior using the ELVIS ground, I cannot be sure without seeing how is your circuit set up. I suggested uploading a picture of your circuit before; do you have the possibility to do that? (I am particularly interested on the AI ports negative connections).

Also, I agree with Mark that repeating the test with 10 K ohms and 0.1uF would provide useful information.

I really don't beleive the bump is due to the board itself, but I don't want to do any judgement yet.

Just wondering if by any chance you have another board to repeat the experiment (if possible, with your actual set up and the one suggested by Mark).

[drneil]

We tried this on multiple ELVIS systems and boards with the same results. I will take a picture of the circuit and post it.

[drneil]

Here is a photo of the circuit connection without a separate ground (i.e., only using the AI channel +/-) and a capture of the Bode plot.  The gain plot is acceptable except that the actual gain at the corner frequency is not -3dB and it should also be 0 up to that point.  The phase plot is not very good.  I also tried connecting the negative side of the AI channels to ELVIS system ground.  That Bode plot is also attached.  It's much worse.