Passive filter to improve signal generator THD performance

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

Tubetec

Well-known member
Joined
Nov 18, 2015
Messages
5,948
So with the help of various tweeks ,Ive managed to get THD and noise from my budget Feeltech arb gen down below -95db or 0.002 %  , that only occurs at favourable  frequencies where processing clock speed coincides mathmatically , in practise as you sweep frequency distortion varies as well , the basic spec of the machine from the factory is 0.5% thd, stepping the waveform between known sweet spot frequencies allows me to get a consistantly lower distortion  across the range .

Could a passive filter tuned for specific test frequencies like 1khz and 10khz get me an extra  reduction in THD on the output of my generator ? I could  live with some loss in signal level ,which of course would be expected .

I cant figure out what combination of L,C and R would be  best .  I have  several  tapped inductors  in suitable values for audio lying around .

 
Yes you can selectively pass the fundamental and attenuate the distortion products, but it would help to understand what the distortion products are that you are trying to filter out.

THD+N includes "noise" that can be significant when measuring a couple zeros past the decimal point. Wide band noise is generally more noise from more HF and/or LF bandwidth.  A simple passive LPF can scrub off a bunch of HF noise, likewise a HPF can reduce hum and 1/f noise. Such crude filters need to be spaced well away from the fundamental, but you can tweak level with filters in place for modest attenuation.

The actual harmonic distortion generally occurs at higher incremental multiples of the fundamental, so a 1kHz fundamental can have distortion spurs at 2k, 3k, etc. A simple passive pole can be more effective at scrubbing off higher harmonics than the low order ones.

Ideally a very narrow bandpass  (opposite of notch filter) could be useful, but these are not trivial passive. Active you could put a notch filter in a negative feedback path to attenuate all but the fundamental.

It seems like these days you could find digital samples of clean sine waves to just playback involving less effort.

JR 
 
Cheers John,

Yeah I can certainly get test tones and crank them out of my D/A convertors , and get equal or even slightly better THD figures than the Arb at its sweet spots  , still the versatillity of the standalone unit is handy sometimes ,the abillity to drive low loads as well as outputting into the Mhz range is handy .

The limit of the best A/D interface for audio I have is 0.001% thd&N @ -0.5dbfs (digital) , to make the most of that ideally I want a source signal an order of magnitude better again .

I have of course taken into account the other  excellent active options you mention to solve the problem ,

Interesting also that the notch filter can  be used on the recovery side in FFT /Spec to better display the upper harmonics .
 
Tubetec said:
Cheers John,

Yeah I can certainly get test tones and crank them out of my D/A convertors , and get equal or even slightly better THD figures than the Arb at its sweet spots  , still the versatillity of the standalone unit is handy sometimes ,the abillity to drive low loads as well as outputting into the Mhz range is handy .

The limit of the best A/D interface for audio I have is 0.001% thd&N @ -0.5dbfs (digital) , to make the most of that ideally I want a source signal an order of magnitude better again .

I have of course taken into account the other  excellent active options you mention to solve the problem ,

Interesting also that the notch filter can  be used on the recovery side in FFT /Spec to better display the upper harmonics .
Basic THD+N analyzers use a notch filter to remove the fundamental then measure what is left over.  Back when I could not afford expensive test equipment, I would feed the product output (after notch filter) from my inexpensive Heathkit distortion analyzer, into an old spectrum analyzer (singer) I bought used. The spectrum analyzer had only 50dB of dynamic range (and 30kHz bandwidth) so not very impressive by itself, but reading the product output I could actually see the distortion caused by the Heathkit audio path. I ended up operating the heathkit 10dB below it's nominal zero level to get a clean measurement floor.

This gave me pretty good visibility into marginal audio performance for half a century ago... today is so much easier (and cheaper).

JR

PS: Just for perspective I paid something like $700 for that used spectrum analyzer back in the 70s, today we can get smart phone spectrum analyzer plug-ins for almost nothing that out perform that old soldier. 
 
I started out with software called Yoshimasa Audio tools(Ral) years and years ago ,
Its a basic Scope ,sig gen and a few other tools all run off your computers audio card , Its a miniscule program , but it still works even today ,it still does just what it says on the tin . 

Audio precision is way out of my league ,with even model one systems fetching many thousands despite being what ,30 years old .

I wouldnt mind a copy of the AP burst generator utillity ,if anyone has it around .
 
Or, if you head over to this thread:

https://www.diyaudio.com/forums/equipment-and-tools/205304-low-distortion-audio-range-oscillator.html

you can see how to get a single frequency oscillator with distortion about 150dB down for about 40 Euros from a chap called Viktor.  The thread itself is also very interesting about oscillator design.  You may have to get a logon to get Viktor's email address and send him a PM.

More details at

https://viccc42.wixsite.com/uld-audio

Cheers

Mike
 
Thanks Mad,

I had taken account of Victor's amazing work alright , it would be nice to have a couple of Victors boards as a benchmark .
If I could get maybe an extra 20 to 30 db reduction of THD+N that comes out of my ARB  that would bring me to around the 125db below mark which I think would be plenty good for anything I do , I can see if someone needed to test cutting edge AD/DA they might get use out of  the extra zero Victors board allows.

I guess as much as getting extra resolution from my test kit , I also want to try and get a feel for how inductor  based filters  work in the audio band  . Thanks for the link John , looks like a good read .
 

Latest posts

Back
Top