Low-distortion Audio-range Oscillator

JensH, did you read my post yet? #9,756
I have attached the calib.file for REW/Arta with APU frequency response correction. BTW, APx555b does the same thing as you probably know.
1528516083789.png
 
Last edited:
  • Like
Reactions: 1 user
That's one of those "impossible" chip makers - for about 10 years starting in the late 70s or so, there were analog synthesizer specific chips made by SSM and CEM, used in most all the famous analog polysynths of the time. Then the all-digital Yamaha DX-7 came out, which sucked the sales from analog synths.

But more recently, a miracle occurred:
https://electricdruid.net/analog-renaissance/
Be aware that page hasn't been updated in recent years, so there are likely more available from those manufacturers than listed, for example:
https://www.alfarzpp.lv/eng/sc/application.php

It appears the best performance (specifically distortion-wise, but likely overall as well) VCAs are from THAT corp, previously the semiconductor division of dbx.
http://www.thatcorp.com/THAT_IC_Selection_Guide.shtml#VCAs

Most of these makers don't appear to be carried by the major distributors, but I at least I found THAT parts at Mouser.
 
yeah, I prefer to use ADAU1701 with a potentiometer to set the frequency if really need such gen. Actually, $50 ES9038Pro will outperform AP as a low distortion generator even without LPF.

You know, that I have JKGEN :) Which is able o generate various signals on it's I2S/SPDIF/Analogue outputs.
Maybe in the next revision it will make sense to replace the AK4490 with ES9038 to reduce the THD on the analogue output, but even a tunable LPF is not suitable for non-sinusoidal signals. The same problem with the ADC notch - only for THD with a sinusoidal signal. Not even IMD is possible.

By the way, when I played with notch+ mdac (not successful due to THD and SNR), I used a slightly different technique than AP:
AP uses the second channel for the direct signal, then digitally filters it and mixes it with the second (notched) signal in the microcontroller.

I used only one channel of a "good ADC" for notch+amplifier and a simple 12-bit microcontroller ADC for the direct signal - only to determine its amplitude and frequency. Then I generate a pure 32bit sine wave (DDS) with this amplitude and frequency and add it to the notch channel.

I can not say that it is better in terms of signal, but it requires significantly less resources of the Microcontroller.
 
yeah, I prefer to use ADAU1701 with a potentiometer to set the frequency if really need such gen ;) Actually, $50 ES9038Pro will outperform AP as a low distortion generator even without LPF.
yes, basically you can hook up any DAC to an ADAU1701.

Or otherwise just use an usb audio interface.
There are plenty that will do around 0.0005% THD+N
And/or can even be cleaned up a lot with some additional circuitry
 
Member
Joined 2004
Paid Member
Every DAC I have checked has serious out of band grunge. I use a 4 pole passive filter to deal with it. Opamps just are not adequate filters for 25, 50, 100 MHz stuff. Victor's oscillators are good to -150 dB. I can't imagine anything that needs a lower distortion source and I would want to see several verification methods for validating those really low numbers. You are in the range where the passive parts can be the limitation.
 
  • Like
Reactions: 1 user
why do you guys don't like the idea to filter DAC by LPF and get -150-160db stable and low noise? Almost any DAC is ok for that because you can use REW to compensate its harmonics, 2nd order LPF will improve 3kHz for -15db and wipe out all higher ones. No frequency adjustment is required, the level sweep is available, 1kHz THD+N <-130db i.e. APx555b is worse for 6db.
HI,IVX, Can the LPF schematic diagram be shared?