Education‎ > ‎

Graphing AF Filter Response on the cheap

posted May 5, 2014, 8:51 PM by Charles Boling
Frequency Response Graph from 14-16 kHz
Wouldn't it be nice to be able to be able to test the frequency response of an audio circuit by hitting a button and generating a nice pretty graph like you see in books and magazines?  Wouldn't it be even nicer to do it without much effort, and without spending a dime?  You can!

I've been having trouble lately w/ my DSL connection being dropped when I made phone calls.  Suspecting a bad filter on either my end or the telco's, and having just received a couple of new filters in preparation for a switchover to a new equipment tomorrow (at both ends of the line) that will hopefully get me my first Internet speed boost in 13 years, I decided to do a little testing of my DSL filters, and though this the perfect setup to do it.

Last year I described how to use Audacity to graph frequency response of radios. This time I eliminated the RF and air segments, and kept things strictly wired.  I began by generating a sweep tone.  This time I went from 0 to 22kHz over 22 seconds, making 1 second per kHz, providing a very convenient frequency scale.

To calibrate my input level, I plugged a patch cable right from my sound card's output to its line input jack, and played the sound while adjusting the input gain to suit me.  Then, I used some other adapter cables and clips I had on hand (instead of simply cutting the patch cable in half to make a special-purpose test set) to give me a couple of RJ11 jacks for connecting the DSL filters.  I kept one of the stereo channels directly connected for a reference.

Procedure: Insert a filter, hit the "record" button, get a response graph.  Pretty easy!  You should be able to do this for any passive filter, as well as an active circuit that's properly powered and DC-isolated.

Here you see graphs (click for full-size view) showing sweeps of a complementary filter pair, the output jacks from a DSL splitter.  (Yes, I know I didn't sweep the entire frequency range; I was having software troubles and got sloppy.)

Graphs for a pair of complementary DSL filter outputs

Now here's a question for you:  Don't the two responses (phone jack vs. DSL modem jack) look backwards to you?

Something to think about:  While the old filter (which I did determine is bad -- I think the capacitor blew) is very simple, containing a single capacitor and two inductors, the new filters are glued shut and I don't know what's in them, leaving the possibility for active components like transistors, and a required DC bias (always present in a live phone loop) to power them properly.  It would be interesting to inject DC into the circuit (with capacitors at either end -- esp. the aux input jack --  to keep it out of the sound card) and see it behaves.