Low distortion 1kHz sine generator for THD measurements.

  • Like
Reactions: NickKUK
NickKUK, I recommend you take a look at the DAC+LPF idea. Simple, reliable, no any adjustments required, at least a level control by SW(not by hands as analog OSC), unbeatable THD+N is guaranteed(Victor OSC may have -150db of THD but THD+N -118-120db in a lucky day), 2nd ord LPF with a decent DAC -145db H3(3-4ord lower -150db with pretty much any DAC).
 
If you have a DAC with USB interface, you can simply run the Soundcard Scope program from PC to generate a low distortion 1Khz tone from it.....
You can, but you usually don't get down near -140 dBc THD. You may break -120 dBc – which is good – but if you want to make sure you're measuring only the DUT and not the DUT+signal generator you'll want something better.

Of course you'll want an analyzer to match. That's possible to build if you can make a low-distortion notch filter to knock down the fundamental tone. That's somewhat straightforward to make work at one frequency, but hard to make work across many frequencies. I spent a few years off/on fiddling with filters to extend the measurement capability of the Audio Precision APx525 only to finally cave in and buy the APx555 instead. The APx555 gets me down around -148 dBc in loopback measurements (so measuring both the source and the analyzer).

Tom
 
If you have a DAC with USB interface, you can simply run the Soundcard Scope program from PC to generate a low distortion 1Khz tone from it.....
Of course using any kinds of DAC, together with DDS and other algorithms is very convenient and flexible to generate various shape signals with a good quality.
But no one DAC is able to provide so low distortion, which are required for the modern THD/IMD measurement purposes.

One improvement method, as mentioned above, is to use a Low Pass (or Band Pass) filter to reduce the harmonics and leave only the fundamental frequency.
But it also has its drawbacks - it is good for obtaining the sine of one frequency, but what to do with sine of the different frequencies?
What to do IMD, multi-tone, etc?

Of course, an analog oscillator like Victor's has the same problem as any other fixed frequency oscillator.
(btw, I own 2 Viktor's - 1kHZ and 10kHz).
 
yeah, Band Pass is Ok too but with some adjustment of the central freq. LPF is nice because +7..10% of the testing freg covers all C0G caps tolerance, hence no trimming is required. I've got the THD+N -123-124db@1kHz with dual ES9038Q2M, and with LPF -131-132db, H3 -145db. REW and others offer the harmonics compensation technic if I can see residual harmonics well lower -160db, I don't understand why I may need an analog OSC at all.
 
I have some spare surface mount PCBs for this design. Also for an associated notch filter. So far it's better than I can measure on the QA 401 I have. Eagerly awaiting a chance to get a new version of the QA once the get parts. If there is interest I can post more information.