Page 6 of 7

Re: Super Low Distortion Ultra Pure Audio Oscillators Revisited

Posted: Sat Nov 21, 2020 12:44 pm
by JR.
There are probably ways to linearize an output buffer/driver with some local circuit tricks. IIRC there was some work done with audio power amps to linearize the output stages (like feed forward, whatever that means).

JR

Re: Super Low Distortion Ultra Pure Audio Oscillators Revisited

Posted: Sat Nov 21, 2020 1:13 pm
by mediatechnology
This is what the THD looks like when a 1KΩ load is connected to the opposing anti-polarity output.
(Both outputs have 300Ω build-out resistors.)
The output level is +16 dBu (differential).
The notch filter is connected to a single output so the level at that node is +10 dBu.
The peak current in the output is about 14 mA.

Again -100 dBu indicated is actually -140 dBu due to post-filter gain.
Compensating for filter insertion loss the THD-2 is at about -131 dBc.
I suspect, but don't yet know, if the even-order cancels when the FFT is run on the differential output.

Image
Low Distortion Oscillator with Composite Op Amps Notch Out +15 dBu Output Level Into 1K Balanced Load

Still -131 dBc HD at 14 mA load current from a 5532 output is not bad at all.
We're talking THD of 0.3 ppm.

When the 1 K load is connected single-ended the THD-2 barely pokes out of the noise floor.

Re: Super Low Distortion Ultra Pure Audio Oscillators Revisited

Posted: Sat Nov 21, 2020 3:49 pm
by mediatechnology
3) The third source is the output stage when heavily loaded at high levels. Working on it. Unloaded the second order is in the noise floor.
The above appears to be PSRR related.

I need to add bypass for the front end transistors and use shunt regulators for the whole oscillator similar to Vicktor's approach. https://www.proaudiodesignforum.com/ima ... Hz_800.jpg

I currently have 1000 µF of brute force from Vcc to Vee and it attenuates all of the load-related odd-order distortion and reduces even-order by 12 dB.

Supply bypass certainly seems to be the right track to get load-related THD back into the noise.

Re: Super Low Distortion Ultra Pure Audio Oscillators Revisited

Posted: Wed Nov 25, 2020 7:26 am
by mediatechnology
mediatechnology wrote:
Sat Nov 21, 2020 3:49 pm
3) The third source is the output stage when heavily loaded at high levels. Working on it. Unloaded the second order is in the noise floor.
The above appears to be PSRR related.

I need to add bypass for the front end transistors and use shunt regulators for the whole oscillator similar to Vicktor's approach. https://www.proaudiodesignforum.com/ima ... Hz_800.jpg

I currently have 1000 µF of brute force from Vcc to Vee and it attenuates all of the load-related odd-order distortion and reduces even-order by 12 dB.

Supply bypass certainly seems to be the right track to get load-related THD back into the noise.
I didn't like the shunt regulators and ended up using a 78L12/79L12 pair to sub-regulate the 15V rail.
The VCA and THAT340 front-end are powered from the 12V rails; the op amps themselves run on +/-15V.
This made the loaded THD drop considerably.

FFTs are the notch filter output with +40 dB post-filter gain. -100 dBu is -140 dBu.

+4 dBu balanced into 600Ω:

Image

+10 dBu balanced into 600Ω:

Image

At +10 dBu single ended the second harmonic begins to rise out of the noise floor even when unloaded.

I'll post an updated schematic soon.

Re: Super Low Distortion Ultra Pure Audio Oscillators Revisited

Posted: Wed Nov 25, 2020 11:12 am
by mediatechnology
I still have some minor tweaking to do but the Ultra Low Distortion Oscillator looks pretty much like this:

Oscillator schematic:

Image
"ULDO" Ultra Low Distortion Oscillator Schematic

Large format schematic: https://proaudiodesignforum.com/images/ ... ematic.jpg

Notch filter schematic:

Image
"Nacho" 1 kHz Notch Filter

I'll drop a circuit description in this space later today.

Re: Super Low Distortion Ultra Pure Audio Oscillators Revisited

Posted: Thu Nov 26, 2020 8:49 am
by terkio
I understand the objective is ultra low THD+N.

How does this analog solution compares with digital synthesis.
_Synthesis using a high performance sound card, focusrite, evaluation boards.
_Synthesis re inventing all from scratch: High speed clock, Ram holding Sinus data, high resolution DAC, appropriate smoothing/filtering.

Digital is a very simple way.... All is in the details.
No problem with amplitude stability and frequency stability which is not easy with an analog solution.

Re: Super Low Distortion Ultra Pure Audio Oscillators Revisited

Posted: Thu Nov 26, 2020 9:42 am
by JR.
terkio wrote:
Thu Nov 26, 2020 8:49 am
I understand the objective is ultra low THD+N.

How does this analog solution compares with digital synthesis.
_Synthesis using a high performance sound card, focusrite, evaluation boards.
_Synthesis re inventing all from scratch: High speed clock, Ram holding Sinus data, high resolution DAC, appropriate smoothing/filtering.

Digital is a very simple way.... All is in the details.
No problem with amplitude stability and frequency stability which is not easy with an analog solution.
apples and oranges....

JR

Re: Super Low Distortion Ultra Pure Audio Oscillators Revisited

Posted: Thu Nov 26, 2020 9:51 am
by terkio
About digital synthesis, I found this: https://www.analog.com/en/technical-art ... rator.html
Jumping at the conclusion.
They claim: The –120 dBc harmonic distortion figure could be met with a careful design....

Re: Super Low Distortion Ultra Pure Audio Oscillators Revisited

Posted: Thu Nov 26, 2020 10:57 am
by JR.
terkio wrote:
Thu Nov 26, 2020 9:51 am
About digital synthesis, I found this: https://www.analog.com/en/technical-art ... rator.html
Jumping at the conclusion.
They claim: The –120 dBc harmonic distortion figure could be met with a careful design....
I have been synthesizing sine waves digitally for 15 years in my drum tuners. I did not have enough bits to be very low distortion, Resotune used an 8 bit processor, the Resotune II used 16 bit processor but both used the processor's internal PWM output to generate sines. While not very clean, good enough for drums and the frequency stability was useful as I also make phase shift measurements directly.

I did some bench work using a DSP micro with built in 16b DAC that would be adequate for basic bench testing (like a digital TS-1). FWIW I didn't use the that DAC to only make one high quality sine, instead I made 8 different frequency sine waves simultaneously, to quickly find drumhead resonances.

====

I designed a pretty interesting voltage controlled analog sine wave oscillator for the old TS-1, again not very low distortion but clean enough to hear distortion in faulty loudspeakers, or misbehaving signal paths. The distortion trade off there was to deliver an acceptably quick amplitude settling time for an oscillator that sweeps over a thousand to one rage with a single knob. I think I compromised on a couple tenths of a percent THD.

JR

Re: Super Low Distortion Ultra Pure Audio Oscillators Revisited

Posted: Thu Nov 26, 2020 11:49 am
by mediatechnology
I don't have a commerically-made DAC that can even come close to this low a level of THD.

The THD and purity of this oscillator, or other analog solutions is at least an order of magnitude - maybe two orders of magnitude - better.
It can be used to check A/D distortion and it's notch filter used to check the D/A.

I looked at the DAC output of my Focusrite 2i2 using the notch filter and compared to the oscillator. The D/A output is full of trash and harmonics when viewed with 40 dB of gain.

The limit I'm aware of for a DAC-based solution is also about -120 dBc.
This is -140 dBc or better at lower levels.

You can tell me a DAC_based solution is better when it clearly isn't.
Build a DAC-based oscillator with a distortion level of -140 dBc with the same level of spectral purity and get back with me when you do.