Moderation edit:
Got it, thanks. I thought you were seeing 3rd in the yellow curve.
Nice project btw. A few years ago I did a similar project, using the Viktor oscillator and got a noise floor with FFT of better than -140dB.
Jan
I guess you guys mixed up THD versus THD+N. Look, the only MFB LPF with OPA1611 will produce 1.1uV of noise in 20-20k BW i.e. THD+N is impossible to be <-128db @ 3Vrms of the output.
Hello Jan,
I'm not surprised that my projects are rather outdated. I got my electric engineering degree almost 40 years ago but never worked in that field as I got completely different business. Now I'm getting back to it like a DIY-er.
Do you have the circuit for that Victor's oscillator. I've build 3 or 4 different oscillators but I've never seen this one. Thanks
Last edited:
Hello,
I developed and built some ultra low distortion oscillators for a professional HIFI studio in Germany. (HIFI Knopf in Düsseldorf. By the way, this shop has a lot of used audio analyzers from R&S and Audio Precision etc.). Designing an oscillator with a THD below -140 dB at 1kHz is extremely difficult and time consuming. It is even more difficult to measure this exactly. With an output signal of 3V, -140dB THD is 300nV.
In my opinion, amplitude control with a lamp can never achieve such a low THD over a long period of time.
If you want to measure the THD precisely, a passive Twin T notch filter with an attenuation of approx. 60db at 1kHz is sufficient. With a subsequent gain of 40dB with an OPAMP, e.g. LM4562 or OPA 1612 or OPA 1656, you can then measure distortions of a few uV. This can then be done by a QA403, for example.
I developed and built some ultra low distortion oscillators for a professional HIFI studio in Germany. (HIFI Knopf in Düsseldorf. By the way, this shop has a lot of used audio analyzers from R&S and Audio Precision etc.). Designing an oscillator with a THD below -140 dB at 1kHz is extremely difficult and time consuming. It is even more difficult to measure this exactly. With an output signal of 3V, -140dB THD is 300nV.
In my opinion, amplitude control with a lamp can never achieve such a low THD over a long period of time.
If you want to measure the THD precisely, a passive Twin T notch filter with an attenuation of approx. 60db at 1kHz is sufficient. With a subsequent gain of 40dB with an OPAMP, e.g. LM4562 or OPA 1612 or OPA 1656, you can then measure distortions of a few uV. This can then be done by a QA403, for example.
Viktor is a member here as vicnic , one thread is https://www.diyaudio.com/community/...range-oscillator.205304/page-466#post-6427230
Jan
Jan
Hello,
I know the thread low distortion audio range oscillator. I already posted there from page 482 two years ago. In my opinion, it makes no sense to start again with "who builds the best non-commercial 1kHz oscillator".
The QA 403 also has these ESS ADC and DAC's and in its data sheet it says: THD -115dB and THD+N -105dB at 0dbV and 20kHz bandwidth loopback.
Miles away from THD -140dB or THD+N -120dB measurements because there is no analog signal pre-processing here.
I know the thread low distortion audio range oscillator. I already posted there from page 482 two years ago. In my opinion, it makes no sense to start again with "who builds the best non-commercial 1kHz oscillator".
The QA 403 also has these ESS ADC and DAC's and in its data sheet it says: THD -115dB and THD+N -105dB at 0dbV and 20kHz bandwidth loopback.
Miles away from THD -140dB or THD+N -120dB measurements because there is no analog signal pre-processing here.
You're right, it is difficult to build an oscillator having such low distortions BUT it is not impossible.
I strongly disagree that "amplitude control with a lamp can never achieve such a low THD over a long period of time".
Have you tried this particular circuit? If no, please be moderate and respect others efforts.
I tried your approach with passive Twin-T, but it would not work as good as the active notch.
I found as well that using two Twin-T notch filters is not as good as combining Hall network notch and Twin-T notch.
no, this is the best sample of 20pcs tested but THD<-140db I can promise. The AVG THD is about -142-144db at MLCK 24MHz and Fs = 96k. The AVG THD+N is -125.7db, the max I saw was -126.3db.
BTW: I keep other 2 samples I selected from 20pcs i.e. the golden sample I sent to Belgium for review, silver and bronze samples are still in my drawer.
The silver one:
Last edited:
I suspect that this was what was meant:
When we take the Fourier transform of a signal, it is equivalent to (mathematically) multiplying it with a phasor (at the frequency of interest), followed by an integration to remove any residual AC (zero-average) components, giving a steady DC value for the strength of the signal at the frequency of interest. Thus, the bandwidth (and therefore the noise) of the FFT operation becomes very small due to the integration that occurs within the process.
When we take the Fourier transform of a signal, it is equivalent to (mathematically) multiplying it with a phasor (at the frequency of interest), followed by an integration to remove any residual AC (zero-average) components, giving a steady DC value for the strength of the signal at the frequency of interest. Thus, the bandwidth (and therefore the noise) of the FFT operation becomes very small due to the integration that occurs within the process.
So what? Doesn't matter how you measured the noise power within the particular bandwidth, ADC->FFT, or analog method. The problem is the offered MFB LPF is way too noisy for 3Vrms, for a 30V is possible to get -140db, no doubt. Or 3V but with much lower impedance as well, 100nF becomes 1uF, 2000ohm -> 200ohm, and so on.
I'm not an expert. but what I see is that it is complicated system with very high Q and noise calculation is not so simple.
Well, as expected the full circuit simulation shows a bit more noise, hence, the N component of the THD+N limits the number at -126.8db @3Vrms. Practically, we need to add a THD component as well but it's ok, let's imagine that THD is completely ZERO 😉
Yes, I did, at least I tried R/R/R(1000ohm of each, and I replaced 10k APU input resistors for 100k) 1/3 voltage divider to check if the harmonics superposition will change. Furthermore, the particular APU I use is a good one from my test of 18pcs APUs. https://www.diyaudio.com/community/...-outperform-apx555b-for-30-000.386001/page-17
IMO an R divider will not change the harmonics structure. You can check with R divider and LPF which attenuate at the same ratio (but rotate phases harmonics relatively). That ratio can be fine-tuned by changing the playback frequency.
Basically an LPF at about 1kHz will make H2 vs. H1 phase shift different compared to the R divider, then chosing the R values so that LPF attenuates approx same around 1kHz, then fine-tuning the frequency to make the LPF attenuation as close as possible to the (frequency-independent) R divider (because distortion depends on level too). Then you can check what the LPF does to the distortions of the combo DAC + ADC - only DAC distortions will be rotated by the LPF compared to the R-divider path.
Basically an LPF at about 1kHz will make H2 vs. H1 phase shift different compared to the R divider, then chosing the R values so that LPF attenuates approx same around 1kHz, then fine-tuning the frequency to make the LPF attenuation as close as possible to the (frequency-independent) R divider (because distortion depends on level too). Then you can check what the LPF does to the distortions of the combo DAC + ADC - only DAC distortions will be rotated by the LPF compared to the R-divider path.
- Status
- Not open for further replies.