DAC performance at very low frequency (1-20 Hz)

[andreasmaaan]

@maxwell3e10 out of curiosity, are you planning to use the DAC for audio playback or some other purpose?

 

[Veri]

I hope it's not for playback, with such specific needs

 

[Rock Rabbit]

Basically, what everyone else here is trying to do, find the absolute best performance possible, not really limited by the budget (a few $k is fine). So white noise level at or below -160dB and as little harmonic distortion as possible.

24 bits means a digital floor noise at -144 dB without dithering...an absolute figure below any threshold, any pre/amp/hp/ls spoil that so no benefit going below that mathematical limit

 

[Vini darko]

Rotary subwoofer is the only audio device I know of that can preform at these frequencies. I'd love to have one just for fun.

 

[Joachim Herbert]

Noise calculator ahead: Enjoy!

 

[mansr]

24 bits means a digital floor noise at -144 dB without dithering...an absolute figure below any threshold, any pre/amp/hp/ls spoil that so no benefit going below that mathematical limit

I'm guessing he meant a noise level of -160 dB/Hz. At 48 kHz sample rate, that's equivalent to about 20 bits resolution.

 

[JohnYang1997]

Update
1.2M points FFT in 20khz bandwidth

You can parallel left and right channels for 3db better performance.

 

[Vini darko]

I just remembered butt kickers type shakers. They also do this stuff. Another thing I'd like to play with. Especially the amps they come with

 

[Vini darko]

View attachment 102208Update
1.2M points FFT in 20khz bandwidth

You can parallel left and right channels for 3db better performance.

That's looking good m8. D70?

 

[restorer-john]

Nobody will design an audio dac that is optimized for frequencies below 20 Hz, because there is close to no market for this.

The CD format was originally specified from DC to 20KHz. Any decent player will easily produce excellent performance well below 10Hz and many were rated flat from 2Hz. You cannot have ruler flat response to 20Hz without going down much lower.

Here's a 1989 TOTL Sony:

PS. It exceeded all those specs in tests.

 

[JohnYang1997]

The CD format was originally specified from DC to 20KHz. Any decent player will easily produce excellent performance well below 10Hz and many were rated flat from 2Hz. You cannot have ruler flat response to 20Hz without going down much lower.

Here's a 1989 TOTL Sony:
View attachment 102213

PS. It exceeded all those specs in tests.

What? Of course no. The noise is way too high for this. Let alone low frequency distortion. He's likely making something like a low noise laser driver.

 

[pkane]

Made me curious, so I ran the test with Element24 down to below 1Hz (2M FFT, 96kHz sampling rate):

 

[maxwell3e10]

View attachment 102208Update
1.2M points FFT in 20khz bandwidth

You can parallel left and right channels for 3db better performance.

Thanks, this makes sense, there is some upward slope at low frequency. Shifted up by 15 dB this gives roughly -150 dB/Hz at 50 Hz and -140 dB/Hz at low frequency. Perhaps Octo or D90 can do a bit better. Maybe this can start a new trend for impressive but non-audible specs. Technical equipment is often specified by noise density at 1 Hz. So DACs need this spec as well.

 

[JohnYang1997]

Thanks, this makes sense, there is some upward slope at low frequency. Shifted up by 15 dB this gives roughly -150 dB/Hz at 50 Hz and -140 dB/Hz at low frequency. Perhaps Octo or D90 can do a bit better. Maybe this can start a new trend for impressive but non-audible specs. Technical equipment is often specified by noise density at 1 Hz. So DACs need this spec as well.

D90 has rising distortion in low frequency. So maybe Okto. But issue is that it takes a lot of time to get one unless you get second hand. That's why I didn't recommend it. Take a look at X16 as well.

 

[DonH56]

Thanks, this makes sense, there is some upward slope at low frequency. Shifted up by 15 dB this gives roughly -150 dB/Hz at 50 Hz and -140 dB/Hz at low frequency. Perhaps Octo or D90 can do a bit better. Maybe this can start a new trend for impressive but non-audible specs. Technical equipment is often specified by noise density at 1 Hz. So DACs need this spec as well.

There are instrumentation DACs (and ADCs) designed for very low frequency applications that might be better suited to your application. An audio DAC IC vs. the "DAC-in-a-box" you buy is more likely to specify noise density and provide a phase-noise or noise PDF plot. What is the application, and what is the bandwidth of interest?

Self-heating effects and such can contribute at very low frequencies but nonlinearity in the circuits like front-end dynamic range, level matching and stage alignment, etc. is likely to contribute more. Flicker noise, 1/f^n noise, chopper noise (if used for offset correction), clock/PLL phase noise, etc. all contribute to the noise floor. Note -160 dB/Hz is fairly challenging.

 

[maxwell3e10]

There are instrumentation DACs (and ADCs) designed for very low frequency applications that might be better suited to your application. An audio DAC IC vs. the "DAC-in-a-box" you buy is more likely to specify noise density and provide a phase-noise or noise PDF plot. What is the application, and what is the bandwidth of interest?

Well, for instrumentation DACs there are R-2R, like AD5791, but attempts to make a dynamic DAC from them have not been so successful. I am curious about Holo Audio May discrete R-2R DAC, but that's pretty expensive, so it better really stand out in performance.

 

[Jimbob54]

I hope it's not for playback, with such specific needs

Pet audiophile whale?

 

[RayDunzl]

Frequency sweep taken at amplifier outputs while driving speakers at about 5 watts, so this includes DAC, preamp, amplifier, and Focusrite ADC to capture the output:

A whopping 0.1dB drop at 10Hz.

I will file this in my "don't worry about it" folder for my personal situation.

(this is an old measurement in my POS (pile o' sweeps) so it could have gone lower had I done it today)

 

[pkane]

Frequency sweep taken at amplifier outputs while driving speakers at about 5 watts, so this includes DAC, preamp, amplifier, and Focusrite ADC to capture the output:

View attachment 102228

A whopping 0.1dB drop at 10Hz.

I will file this in my "don't worry about it" folder for my personal situation.

(this is an old measurement in my POS (pile o' sweeps) so it could have gone lower had I done it today)

Now you'll just have to run it to below 1Hz, otherwise you'll be worried forever about what might be hiding there

 

[RayDunzl]

Now you'll just have to run it to below 1Hz, otherwise you'll be worried forever about what might be hiding there

DC hides below 1Hz.

---

Experiment:

Wave hand in front of microphone:

4.81Hz at 86.4dB on the spectrum.

Not something to which I would normally choose to listen.