DIY Measurements with minimal hardware and open source software

[Lambda]

At this point its just some fooling around!
I'm interested to see if i can uses a cheap old usb audio interface ADC and some basic tools to measure the performance of DAC with (better performance).

1.Noise
Test signal is a 1k sine wave with -60dBFS peak amplitude:

We see peaks that appear like distortion but i cant tell if they com from the dac or the adc.
For now we just count them as noise and going with the SINAD number from Baudline of ~42dB
So Noise(+THD) is 42dB lower as the primay peak Assuming the DAC is linear the peak could be 60bB higher (since this was taken at -60dBFS peak output)
With the assumption the DAC is linear this would give a SNR of >102dB


DAC no output:

Without output the "distortion" disappears Jnoise meter drops from -19,7 to -62dB
Assuming -19,7dB is mainly the -60dB signal and Assuming the DAC is linear this gives a gain of 40,3dB
Subtracting the 40,3dB gain from -62dB noise floor gives 102,3dB SNR

System noise with 100ohm terminator:

But how much of this noise comes from the ADC?
the ADC self noise with the same gain drops to -82.8dB
Since its >20dB lower then the signal we measured before the influence from the self noise was neglectable (error is ~0.5dB)

But all my dB values have no reference!
Lest fix this by measuring DACs output voltage of a 0dBFS 50Hz and 100Hz signal.
The trusty cheap Anang AN8008 says 2.002Vrms i call this 2Vrms and because i'm lazy i call this
0dBFS = 6dBV

Actual noise floor is -96,3dBV

I know i did not limit the bandwidth! and there are a lot of errors her but that's it for today.

the best way to get the right answer on the internet is not to ask a question; it's to post the wrong answer

 

[AnalogSteph]

That's basically the right approach. When it comes to determining DAC dynamic range, a decent preamp and a serviceable ADC plus a moldymeter will go a long way. And your preamp isn't even all that good... equivalent input noise with a 100 ohm source would be -82.8 dBFS - 40.3 dB = -123.1 dBFS or ~ -117 dBV / -115 dBu (a decent mic preamp should do 10 dB better). Should get the job done with about a 100$ DAC performance level but beyond that you'd be well-advised to be using the RMS unsumming calculator for an accurate estimate of DAC noise floor. At SOTA level the error bars would be entirely too big for my liking, but I mean by that point you can probably afford a better preamp, too.

Looks like the DAC under test may have some digital signal getting into the analog stages. It looks worse than it is, considering all the harmonics ultimately are below -120 dBFS and the noise floor is almost 20 dB higher. This is the kind of thing you tend to get with so-so 2-layer circuit board layouts and cheap and crappy bypass capacitors on the reference voltage.

I have used a fairly similar setup myself, with a portable mixer (first a Behringer, now a Mackie 402 VLZ4) running into an Asus Xonar D1, all acquired used. The Xonars happened to be popular HTPC cards around here a decade ago, and PCI slots that'll work with them are a bit thin on the ground in modern PCs, but if you can use one a CS5361 is not a bad midrange ADC. If you can live with using some older hardware, working around driver issues and perhaps carry out some repairs (or even mods) yourself, there are some nice bargains to be had out there. A first-rate ADC from the mid-2000s would still be very good today (there's very little wrong with an AK5394A, really), and it took the (lower) midrange until the MOTU M4 to finally catch up with old EMU 0404 USB, a long-term lab favorite sporting an AK5385A ADC.

 

[Lambda]

@AnalogSteph
Thanks!
I was thinking maybe it would be a good idea to calculate the equivalent input noise of my ADC first...

equivalent input noise with a 100 ohm source would be -82.8 dBFS - 40.3 dB = -123.1 dBFS

Jaaa and Jnoise meter have the oddity to messure a 0dBFS sinwave as 0dB "RMS"
A square wave is actually +3dBFS for it
Baudline in not doing this but baudline is not verry precise and has scalloping loss depending on window function.
should i just subtract 3dB to all Jaaa and Jnoise measurements?

a decent mic preamp should do 10 dB better)

Its an M-audio c 600 and i'm using the instrument in at the front. Gain was about 50%

you'd be well-advised to be using the RMS unsumming calculator for an accurate estimate of DAC noise floor. At SOTA level the error bars would be entirely too big for my liking, but I mean by that point you can probably afford a better preamp, too.

Thanks! i will give ti a try!

This is the kind of thing you tend to get with so-so 2-layer circuit board layouts and cheap and crappy bypass capacitors on the reference voltage.

It is a crappy 20€(now30€) USB dongle DAC with integrated charge pump.

and it took the (lower) midrange until the MOTU M4 to finally catch up

i'm waiting for it to become available.