Build -- Active Twin-T notch filter for distortion analysis

[klewis]

Originally Posted by Frex

Hello Klewis,

Do you use symetric or asymetric inputs of the Juli@ ?


FRex.

Hi Frex,

I used the balanced inputs of the Juli@. The setup was HP8903A output (unbalanced), to Pete Millett's sound card adapter > balanced output from adapter > balanced input of the Juli@.

Some of the frequency roll off might be from the sound card adapater asymetric to balanced line drivers. The THAT corp data sheet only rates them to 20kHz.

Ken

[richiem]

Hi folks -- just a quick update on the issue of signal levels into the Twin-T. I've needed a better sound card than the one that is built into my PC, which is bandwidth limited by its maximum 96kHz sampling and by its somewhat higher than good noise floor at -120dB referred to near full-scale input.

I have recently acquired an E-MU 0204 USB "sound card" and find its ADC to be a very good device -- 1V, 1kHz THD is well below 0.005% and the noise floor is around -130dB -- much beter than the on-board sound chip in my system. I'll have more accurate data soon. In any case, when used with the Active Twin-T, good harmonic product measurements can be made on fundamentals of 10kHz or more to well below -130dB using the 20dB post-filter gain of the Twin-T.

I'm still in the early stages of evaluation, but I can definitely say that limiting the input to the Twin-T to something around 1VRMS gives the best overall results. At 3V the amps in the Twin-T seem to be sticking their noses into the results. I hope to be able to say something more definitive about this soon and will post here and on my website.

I have also found some rather strange things with a borrowed Juli@ card in terms of audio bandwidth -- it appears that there is some kind of input filtering going on so that using a higher sampling frequency doesn't result in wider measurement bandwidth. In this regard, the EMU 0204 is clearly superior and gives better results if you want to see distortion products out beyond the 22kHz area. The EMU gives good response to about 90kHz with 192kHz sampling, consistent with Nyquist limit filtering.