What measurements to look at for comparing DAC/Preamp output stage

[purplezorz]

Hey Guys,

I'm trying to figure out which specifications to look at to try and differentiate between two DAC pre-outs.

I have a SMSL C200 and a Burson Conductor 3R GT.
They are connected to a MiniDSP which is connected to a 2.1 system (Klipsch RP-500m (powered by Emotiva A-100) + SVS PB1000)
The two differences I'm interested in are:

1. /volume - The Burson has much more headroom. I'm sure this will be something like "output voltage", or perhaps the output impedance plays a role.
2. Bass sensitivity - The sub bass frequencies are more prominent with the Burson - or said differently, seems to be more sensitive to them so it feels like they're not missing at lower volumes.

For the latter I'm guessing this has something to do with the output stage electronics, as the DAC stage should probably measure flat (and they're the same ESS ES9038Q2M - though I heard, it's not the chip, but the implementation of the chip).
So I guess the question in that case is, should I be looking at the amp specifications at that point? If I understand correctly, the preamp stage is just the amp at much less power. A logical conclusion may just be the frequency response of the amp stage is straight up just different, so I'd never be able to tell unless they're measured. But hey ho, maybe I'm missing another parameter.

In the mean time I suppose I should level-match the two and measure their frequency response to see if there's any difference in REW and go from there.

TIA

 

[DVDdoug]

I've NEVER heard a difference or defect from a DAC, including the DAC built into my 1st CD player about 40 years ago. (I don't claim to have golden ears but I'm a "picky listener".)

volume - The Burson has much more headroom. I'm sure this will be something like "output voltage", or perhaps the output impedance plays a role

Yes, that would be voltage.

Bass sensitivity - The sub bass frequencies are more prominent with the Burson - or said differently, seems to be more sensitive to them so it feels like they're not missing at lower volumes.

If a DAC isn't defective it will have flat frequency response over the audio range. If "sub bass" is below the audible range (i.e. lower than ~20Hz) there might be differences. It's easy to make a DAC that goes down to DC (zero Hz) but DC (or other very low frequencies isn't "sound" and it can cause problems so frequencies below the audio range are usually filtered-out.

When the volume is turned-up our ears/brain perceive that the bass is turned-up more and when the volume is turned down it sounds like the bass has been turned-down more. (Equal Loudness Curves)

In the mean time I suppose I should level-match the two and measure their frequency response to see if there's any difference in REW and go from there.

Yes. And if possible make electrical measurements rather than acoustic microphone measurements.

...Slight microphone re-positioning or just your body moving around in the room can make a difference in measurements so an acoustic measurement difference isn't always meaningful. I was doing some high-frequency experiments with an SPL meter on a mic stand once and must moving around behind the meter made significant changes in the readings. With constant high frequency test-tones you can also hear drastic changes as you move your head slightly. (These effects are less pronounced at lower frequencies.)

To be valid, listening tests also have to level-matched (that should be done by measuring the voltage), blind, and statistically repeatable. What is a blind ABX test? The main reason for level matching is that if one is louder than the other it's easy to identify "X" and the whole test becomes meaningless. An ABX test only confirms if you are reliably hearing a difference or not. If there is a real audible difference you can investigate more to determine what the difference is or which one you prefer, etc. ABX tests are useless with speakers or headphones because there almost always is a difference in sound, and with headphones, also a difference in how they feel on your head. Blind listening can still be useful but the tests have to be designed differently.

A logical conclusion may just be the frequency response of the amp stage is straight up just different

Most amps also have flat frequency response.

maybe I'm missing another parameter.

With electronics, there is ONLY frequency response, noise, and distortion. (See Audiophoolery) Usually frequency response and distortion are better than human hearing unless you overdrive an amplifier into clipping/distortion. Sometimes there is audible noise (hum hiss or whine) in the background. ...That's assuming no gross problems like no sound out of one side, or one side louder than the other, or audio cutting-out, etc.

With speakers & headphones, frequency response is the main concern and with speakers in a room, the off-axis response and effects of room acoustics-reflections.

 

[purplezorz]

I've NEVER heard a difference or defect from a DAC, including the DAC built into my 1st CD player about 40 years ago. (I don't claim to have golden ears but I'm a "picky listener".)


Yes, that would be voltage.


If a DAC isn't defective it will have flat frequency response over the audio range. If "sub bass" is below the audible range (i.e. lower than ~20Hz) there might be differences. It's easy to make a DAC that goes down to DC (zero Hz) but DC (or other very low frequencies isn't "sound" and it can cause problems so frequencies below the audio range are usually filtered-out.

When the volume is turned-up our ears/brain perceive that the bass is turned-up more and when the volume is turned down it sounds like the bass has been turned-down more. (Equal Loudness Curves)

Yes. And if possible make electrical measurements rather than acoustic microphone measurements.

...Slight microphone re-positioning or just your body moving around in the room can make a difference in measurements so an acoustic measurement difference isn't always meaningful. I was doing some high-frequency experiments with an SPL meter on a mic stand once and must moving around behind the meter made significant changes in the readings. With constant high frequency test-tones you can also hear drastic changes as you move your head slightly. (These effects are less pronounced at lower frequencies.)

To be valid, listening tests also have to level-matched (that should be done by measuring the voltage), blind, and statistically repeatable. What is a blind ABX test? The main reason for level matching is that if one is louder than the other it's easy to identify "X" and the whole test becomes meaningless. An ABX test only confirms if you are reliably hearing a difference or not. If there is a real audible difference you can investigate more to determine what the difference is or which one you prefer, etc. ABX tests are useless with speakers or headphones because there almost always is a difference in sound, and with headphones, also a difference in how they feel on your head. Blind listening can still be useful but the tests have to be designed differently.


Most amps also have flat frequency response.


With electronics, there is ONLY frequency response, noise, and distortion. (See Audiophoolery) Usually frequency response and distortion are better than human hearing unless you overdrive an amplifier into clipping/distortion. Sometimes there is audible noise (hum hiss or whine) in the background. ...That's assuming no gross problems like no sound out of one side, or one side louder than the other, or audio cutting-out, etc.

With speakers & headphones, frequency response is the main concern and with speakers in a room, the off-axis response and effects of room acoustics-reflections.

Thanks for this!
When I say sub-bass, in this case I mean 40-17 (my sub rolls off at 17).
Unfortunately, I don't have the means to make electrical measurements, however, I made some measurements with my UMIK-1 after level matching within REW and I leave the room when the measurements take place.
Both at moderate and high volumes, the graphs are within a db with each other - thus the perceived difference in sub-bass must be fictitious, or related to volume mismatch.
Whereas the output level should just be the voltage spec.
I didn't go the whole hog and do three separate measurements as well as three different axis points, but yes, as you mentioned slight variations in the listening position makes a big difference, turning peaks into nulls and vice versa.

I'll attach the graphs just for fun (Ignore the overall uneven FR, the MiniDSP is using a correction from my old room).

Cheers!

 

[RalphB]

How important is it to match the DAC to the headphone? I would think there would always be optimal configurations.

 

[Sokel]

THD+N vs level and frequency and Multitone distortion mainly.
Look the ballpark at the levels you predict you'll use it as it's there that you want it to be ok.

 

[RalphB]

When transducers have best case distortions of point something, does DAC and amp parameters become irrelevant?

 

[RalphB]

I've NEVER heard a difference or defect from a DAC, including the DAC built into my 1st CD player about 40 years ago. (I don't claim to have golden ears but I'm a "picky listener".)


Yes, that would be voltage.


If a DAC isn't defective it will have flat frequency response over the audio range. If "sub bass" is below the audible range (i.e. lower than ~20Hz) there might be differences. It's easy to make a DAC that goes down to DC (zero Hz) but DC (or other very low frequencies isn't "sound" and it can cause problems so frequencies below the audio range are usually filtered-out.

When the volume is turned-up our ears/brain perceive that the bass is turned-up more and when the volume is turned down it sounds like the bass has been turned-down more. (Equal Loudness Curves)

Yes. And if possible make electrical measurements rather than acoustic microphone measurements.

...Slight microphone re-positioning or just your body moving around in the room can make a difference in measurements so an acoustic measurement difference isn't always meaningful. I was doing some high-frequency experiments with an SPL meter on a mic stand once and must moving around behind the meter made significant changes in the readings. With constant high frequency test-tones you can also hear drastic changes as you move your head slightly. (These effects are less pronounced at lower frequencies.)

To be valid, listening tests also have to level-matched (that should be done by measuring the voltage), blind, and statistically repeatable. What is a blind ABX test? The main reason for level matching is that if one is louder than the other it's easy to identify "X" and the whole test becomes meaningless. An ABX test only confirms if you are reliably hearing a difference or not. If there is a real audible difference you can investigate more to determine what the difference is or which one you prefer, etc. ABX tests are useless with speakers or headphones because there almost always is a difference in sound, and with headphones, also a difference in how they feel on your head. Blind listening can still be useful but the tests have to be designed differently.


Most amps also have flat frequency response.


With electronics, there is ONLY frequency response, noise, and distortion. (See Audiophoolery) Usually frequency response and distortion are better than human hearing unless you overdrive an amplifier into clipping/distortion. Sometimes there is audible noise (hum hiss or whine) in the background. ...That's assuming no gross problems like no sound out of one side, or one side louder than the other, or audio cutting-out, etc.

With speakers & headphones, frequency response is the main concern and with speakers in a room, the off-axis response and effects of room acoustics-reflections.

Electronics have improved so much even a blind squirrel can design decent equipment today.

 

[dualazmak]

Hello OP @purplezorz and ASR friends onboard on this thread,

Although quite belated, I just came across with this thread of yours.
My points described below maybe out of the scope of your interests, but let me share just for your possible reference.

I too have been much interested in actual objective measurements of Fq-SPL response in various stages in my PC-DSP-based multichannel multi-SP-driver multi-amplifier fully active stereo audio setup i.e. throughout all of;

1. most upstream digital domain before going into DACs,
2. line-level analog outputs of DACs,
3. high-level SP outputs of amplifiers before protection capacitors,
4. high-level SP outputs of amplifiers after protection capacitors,
5. room air sound of each-of/all-of the SP drivers using calibrated measurement microphone,

all should be measured by consistent and reliable/reproducible (with essentially no statistic fluctuation/deviation) FFT spectrum analysis.

I applied, therefore, the method of "FFT averaging of rich recorded/accumulated DSP-processed sound data of strictly QC-ed flat white noise" all the way through above 1. to 5.

Only if you would be interested, please visit my summary post #404 on my project thread, and also please visit my post #1,009 sharing recent re-measurements/re-confirmation of above stages of 3. 4. and 5.