AliExpress cheap low distortion analog sine generator - review and measurements

[pma]

This analog generator has much cleaner ultrasound HF spectrum than any DAC that I have tested and thus allows for proper investigation of class AB amplifiers cross-over HF distortion and dead-time class D distortion, not polluted by DAC HF artifacts.

Below is a wideband measurement of class AB amplifier at 156W/4ohm. Kudos to @IVX concerning E1DA Cosmos ADC behaviour.

 

[SSS]

Thank you for your reply. The generator should be quite immune to its physical placement, this might depend on shielding efficiency. I am also a bit confused by the modulated noise floor in your last plot, it might indicate to frequency instability or the window used.

Tried same setup with another software program and FFT Kaiser. Where a possible modulation could come from, I have no idea since I am not in the FFT math and its software. It is also possible that the oscillator is not perfectly stable. Atached the new plot.

 

[AnalogSteph]

@pma
The frequency being that far off would trigger my OCD. Does anything bad happen if you trim it to be closer to an actual 1 kHz?

 

[pma]

This might be an interesting measurement - left ADC channel measures output from the generator, right ADC channel input is shorted and measures in fact ADC intrinsic noise + generator frequency crosstalk. It can be seen that ADC noise almost overlaps with generator noise, so the noise of the generator is not measured precisely, we would need to have even lower ADC input noise than we get with the Cosmos.

 

[IVX]

For my easy meas setup I could go little bit better regarding mains hum. At this low level I cannot really fast find out where it is injected. There I would need to invesigate with screening cases and probably better power supply aka batteries. Attached the latest FFT. But as I said part of the distorion may come from the QA403 since Viktor claimes somewhat of -160 dB THD or so. Found that even at a distance of half a meter the small power supply transformer coupled into the circuits. Further away I could get the hum into the noise floor. See the second attached file just below this text.View attachment 474251

it seems he claims -160db H3 level, total THD is higher but crazy as well! Archimago measured THD <-150db.

MUSINGS: The hunt for -150dB THD with the Victor's 1kHz Oscillator; A look at Paul K.'s Multitone Loopback Analyzer; And the pinnacle of audio streamer pricing? The Wadax Atlantis Reference Server. (And REW IMD vs. Level Step!)

A blog for audiophiles about more objective topics. Measurements of audio gear. Reasonable, realistic, no snakeoil assessment of sound, and equipment.

archimago.blogspot.com

 

[pma]

Tried same setup with another software program and FFT Kaiser. Where a possible modulation could come from, I have no idea since I am not in the FFT math and its software. It is also possible that the oscillator is not perfectly stable. Atached the new plot.

Good. Maybe THD+N limit of the QA403?

@pma
The frequency being that far off would trigger my OCD. Does anything bad happen if you trim it to be closer to an actual 1 kHz?

It might be possible, but frankly, is there other than fashion reason to set it at 1kHz precisely? The amp under test does not care about it.

 

[pma]

DSO record at Fs = 1.56MHz. DSO input impedance is 1Mohm, so the rms voltage output is a bit higher than with low impedance input of the Cosmos. Generator output impedance is 100 ohm.

 

[amirm]

Very nice. Promoted to home page.

 

[SSS]

it seems he claims -160db H3 level, total THD is higher but crazy as well! Archimago measured THD <-150db.

MUSINGS: The hunt for -150dB THD with the Victor's 1kHz Oscillator; A look at Paul K.'s Multitone Loopback Analyzer; And the pinnacle of audio streamer pricing? The Wadax Atlantis Reference Server. (And REW IMD vs. Level Step!)

A blog for audiophiles about more objective topics. Measurements of audio gear. Reasonable, realistic, no snakeoil assessment of sound, and equipment.

archimago.blogspot.com

Interesting link regarding the Viktor oscillator. One remark, the switching power supply for the oscillator was also used by myself at the first step. But found that there was out-of audio band frequency noise. So I switched to conventional mains operated power supply but as I now found needs to be far apart from the other circuits.

 

[dominikz]

it seems he claims -160db H3 level, total THD is higher but crazy as well! Archimago measured THD <-150db.

MUSINGS: The hunt for -150dB THD with the Victor's 1kHz Oscillator; A look at Paul K.'s Multitone Loopback Analyzer; And the pinnacle of audio streamer pricing? The Wadax Atlantis Reference Server. (And REW IMD vs. Level Step!)

A blog for audiophiles about more objective topics. Measurements of audio gear. Reasonable, realistic, no snakeoil assessment of sound, and equipment.

archimago.blogspot.com

Victor's oscillator is indeed insanely low distortion. I couldn't reliably measure its distortion components at all, not even with a twin-T notch filter:

A bit more information in this post.

To be completely honest this level of precision is total overkill for my use, but it wasn't very expensive and it does come in handy here and there!

 

[pma]

This is a study of 1ET400A PWM modulator noise shaping, please note noise shaping as a function of output voltage. Because the generator is clean at ultrasound frequencies, we may be sure that what we see is coming from the amplifier. Compare to post #21.

 

[IVX]

Interesting link regarding the Viktor oscillator. One remark, the switching power supply for the oscillator was also used by myself at the first step. But found that there was out-of audio band frequency noise. So I switched to conventional mains operated power supply but as I now found needs to be far apart from the other circuits.

Archimago measured Victor's osc with Cosmos APU notch, and that notch utilizes three SMPS.

 

[SSS]

Archimago measured Victor's osc with Cosmos APU notch, and that notch utilizes three SMPS.

If bandwitdh covers the regular audio band (20 - 20000 Hz) then the noise of the SMPS may not be detected. I talk about noise at 80 kHz and much higher which by the way is also produced from DACs, also my QA403 does it and I use a BNC inline filter 100 kHz and 200 kHz according to the bandwidth to measure. I detected this when I also use traditional analog distortion and level measurement instruments which some can measure up to 1 MHz. Also my scope (500 MHz) shows spurious signals up to 100 MHz. With audio band limited measurements this all my not harm.

 

[Dawsoo]

I was the first pioneer, I bought an empty pcb in June and assembled it myself with quality material. It is much better than my own the Victor oscillator.

https://www.aliexpress.com/item/1005003099039608.html?

https://hififorum.sk/phpBB3/viewtopic.php?f=20&t=955&start=40

Freely on the table other configuration. Power supply +-12V Jung superreg. The limit only is in my cosmos E1DA.

 

[pma]

The limit only is in my cosmos E1DA.

It is the limit in noise performance. It limits the measured noise accuracy. I have calculated the correction on the Cosmos intrinsic noise as follows:

Vnadc = 8.746 uV ..... this is the intrinsic noise of my Cosmos ADC at 10V input range, with input shorted
Vnmeas = 9.817 uV ..... this is the noise read by REW from the THD+N measurement of the generator connected to 10V Cosmos input

Vnmeas is obviously affected by Cosmos noise. We can calculate the generator noise by well known formula

Vngen = sqrt(Vnmeas**2 - Vnadc**2) = 4.4588 uV. This is the generator noise without additional ADC Cosmos noise.
Now, as generator output voltage is Vout = 7V, the real SNR = 123.9 dB, as calculated from 7V and 4.4588uV.

Note: the parameters may differ a bit with every piece, depending on op-amps, parts and final setting. However, not so much and considering the price, this is unbeaten.

 

[JohnPM]

Please can you explain details how RTA in REW handles slow frequency drifts of analog oscillators?

The fundamental is determined from a centroid-like calculation of powers of bins around the peak, taking into account the spreading span of the window in use. Harmonic powers use multiples of that fundamental and take into account the spreading span of the window in use.

 

[capslock]

I was the first pioneer, I bought an empty pcb in June and assembled it myself with quality material. It is much better than my own the Victor oscillator.

https://www.aliexpress.com/item/1005003099039608.html?

https://hififorum.sk/phpBB3/viewtopic.php?f=20&t=955&start=40

Freely on the table other configuration. Power supply +-12V Jung superreg. The limit only is in my cosmos E1DA.

Much better in noise or harmonics? Current or older Victor's oscillator? Your measurement shows -134 dB HD3 and HD5. The current Victor's ocscillator has been reported to be a lot better regarding harmonics.

I read (with autotranslation) a few pages of the thread you linked. There is a lot of talk about your line source speakers but I didn't find what quality components you used. Different op-amps? Special resistors and capacitors?

 

[phofman]

The fundamental is determined from a centroid-like calculation of powers of bins around the peak, taking into account the spreading span of the window in use. Harmonic powers use multiples of that fundamental and take into account the spreading span of the window in use.

John, thanks a lot. Please what would happen if the fundamental slightly shifts within the window of samples (as long FFTs take seconds/tens of seconds of samples)?

 

[pma]

what would happen if the fundamental slightly shifts within the window of samples (as long FFTs take seconds/tens of seconds of samples)?

What exactly do you mean as "slightly". Could you define how "slightly" (exactly in Hz) in case of the measurement below.

 

[Dawsoo]

Much better in noise or harmonics? Current or older Victor's oscillator? Your measurement shows -134 dB HD3 and HD5. The current Victor's ocscillator has been reported to be a lot better regarding harmonics.

I read (with autotranslation) a few pages of the thread you linked. There is a lot of talk about your line source speakers but I didn't find what quality components you used. Different op-amps? Special resistors and capacitors?

Hi, I used normal metallized resistors selected under 0.5%, cermet trimmers Bourns, MKP and NP0 capacitors, operational amplifiers thanks to the sockets I tried many types and chose the best of which I had a sufficient amount in the reduction of soic8 to dip8. JFET original 2SK30 from Toshiba, OPA definitely were not the prescribed TL071 and NJM/NE5532. I have three boards, so I still want to make generators for 100Hz and 10kHz. Victor generator (1st gen) has a very poorly designed power supply so it makes noise and distorts more.

The limit is in my ADC E1DA cosmos, no notch filter or scaler/buffer is used.