• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

linking the subjective with the objective in DAC performance

nsayer

Member
Joined
Oct 17, 2021
Messages
5
Likes
3
I've taken a bit of a self-driven crash course in DACs of late.

It all started with the Belkin Soundform Connect Airplay 2 box. I bought one, but the analog output is easily 15 dB too low (along the way I actually built a stupid op amp 15 dB preamp as a workaround and that works, but that's beside the point). So I bought a $10 DAC from Amazon to take the TOSLINK output instead and that worked. But it got me to thinking whether I could make my own, just to see if I could.

I've gone through a half dozen designs now. At first, I was using the STA120 and CS4334. I got that working, and it was fine, but people waxed lyrical about higher performance DACs, so I decided to see if a higher performance DAC made a difference. I settled on the PCM1793. I also moved from the STA120C to the DIR9001 and more recently the CS8416 receiver chip.

I haven't done any double-blind testing, but there is absolutely no question to my mind that the 1793 sounds very, very different than either the 4334 or the $10 DAC from Amazon (I don't know what's in that DAC because the bastards sandblasted the chips). To my ears, the cheaper DACs sound very flat and the 1793 is far more lively. I am not sure if I have better vocabulary for it than that. The only other subjective comparison I've made is that if I connect the 1793 to my AirPods Max with the analog cable it sounds the same to me as simply streaming the same audio to the APM over its native bluetooth. Again, the cheaper DACs are just flatter to me.

I guess the question I have is, what does "flatter" and "livelier" translate to in objective terms? Because I can't believe that there isn't something objectively measurable in the difference between the 4334 and 1793. The difference is so stark that I just can't hardly believe it's mere placebo.
 

LTig

Master Contributor
Forum Donor
Joined
Feb 27, 2019
Messages
5,760
Likes
9,442
Location
Europe
I guess the question I have is, what does "flatter" and "livelier" translate to in objective terms? Because I can't believe that there isn't something objectively measurable in the difference between the 4334 and 1793. The difference is so stark that I just can't hardly believe it's mere placebo.
The only way to find out is to make a proper level matched blind test.
 

Killingbeans

Major Contributor
Joined
Oct 23, 2018
Messages
4,089
Likes
7,552
Location
Bjerringbro, Denmark.
I guess the question I have is, what does "flatter" and "livelier" translate to in objective terms?

A slight difference in output voltage?

Because I can't believe that there isn't something objectively measurable in the difference between the 4334 and 1793.

What did you measure, at what did you not?
 
OP
N

nsayer

Member
Joined
Oct 17, 2021
Messages
5
Likes
3
I am not very well equipped as of yet (I'm new at this). I have attempted to send a 0 dBFS 1 kHz sine wave through the DAC and use my scope to get an FFT evaluation. The result of that looked not very different from figure 19 in the PCM1793 datasheet. I am not sure that says very much of value, however.

I have a USB to TOSLINK interface coming tomorrow, which I anticipate will be a much better mechanism for getting digital audio into this thing for testing than the Belkin box.

I will measure the relative output level of the 4334 and 1793 with a 0 dbFS tone to see if there is a significant difference that might explain the difference in sound.
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,524
Likes
37,057
If the voltage difference is greater than about 2%, it's enough to make them sound audibly different. Even if there are no other differences.
 
OP
N

nsayer

Member
Joined
Oct 17, 2021
Messages
5
Likes
3
I compared all 3 with a 0 dBFS 1 kHz sine test file and, sure enough, they're all 3 very different.

The 1793 output is roughly 6v P-P. The CS4334 is more like 4v. So I will retry my testing with some sort of pad on the 1793 to see if equal levels still result in subjective differences.

The $10 amazon box was the big surprise. Its output was about 5v, but it was clipped badly, as if it was trying for 6v but didn't make it. It also was heavily DC biased. Absolute rubbish.

EDIT: The picture from the Amazon box
 

Attachments

  • IMG_1492.jpeg
    IMG_1492.jpeg
    265.8 KB · Views: 47
Last edited:

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,524
Likes
37,057
I compared all 3 with a 0 dBFS 1 kHz sine test file and, sure enough, they're all 3 very different.

The 1793 output is roughly 6v P-P. The CS4334 is more like 4v. So I will retry my testing with some sort of pad on the 1793 to see if equal levels still result in subjective differences.

The $10 amazon box was the big surprise. Its output was about 4v, but it was clipped badly, as if it was trying for 6v but didn't make it. It also was heavily DC biased. Absolute rubbish.
So two very different levels and one of them may have been clipping or distorting. You are making progress. Match levels and see what you hear.
 
OP
N

nsayer

Member
Joined
Oct 17, 2021
Messages
5
Likes
3
So two very different levels and one of them may have been clipping or distorting. You are making progress. Match levels and see what you hear.
Well, three very different outputs for comparison. One rubbish, the other two look fine, but are not the same level. So, yeah, I will try to pad the hot one and see if there's still subjective differences.
 
Top Bottom