How to measure the frequency response of a non-textbook filter circuit

[fizzbang]

I'm trying to characterize the approximate performance of a black-box (potted) DC power filter which includes a common-mode filtering component, presumably it'll be similar to one of these.  Problem is all the instructions for using the Network Analyzer assume the existence of a common ground point to connect Ch1-/Ch2-/W1- to, however in this case there's two locations, the filter input (before the choke) and the filter output (after the choke).  How do you connect Ch1/Ch2/W1 to measure the frequency response of such a circuit?

(I'm ignoring source and load characteristics and other complications for now, I just want to get a general overview of what the filter is doing).

[attila]

Hi @fizzbang

The Analog Discovery 1,2,3 oscilloscope inputs are suitable for such measurements since these are differential.

To measure the entire filter: channel 2 measured relative to 1.

To measure just the second part influence: channel 2 measured relative to 1.
In this case channel 1 absolute measurement will represent the first part only if the load does not cause W1 to drop.

[fizzbang]

Thanks!  Forgot to mention that I'm using the BNC adapter with grabber cables which makes the GND status a bit unclear since there's GND, SGND, and WGND, and Ch1 and Ch2 share the one SGND.  This produced rather a odd plot:

but then bypassing the BNC adapter produced something fairly similar:

I'll have to grab a known-good filter to see whether this is operator error or the one I'm testing is the problem, it's of unknown age so I don't know what state the electrolytics are in.

[attila]

Hi @fizzbang

The Channel 1 curve shows low impedance at low frequencies exceeding the AWG output drive capability, dropping the output voltage significantly.
The phase can only have a range of 360*, adjust the offset for this.