Review and Measurements of Benchmark DAC3

[bennetng]

For several times I mentioned the importance of the actual audio files.

https://www.audiosciencereview.com/...dio-precision-analyzer-apx555.3442/post-83427

But there is no reply. I don't know if it is the limitation of AP (that it can't save any audio files) or he didn't read my comment. Without audio files there will be no listening tests.

On HA I asked someone to provide the recorded audio files for his RMAA tests, I got the files and was able to trace the source of problems.

https://hydrogenaud.io/index.php/topic,111411.msg918485.html#msg918485

 

[rebbiputzmaker]

There does seem to be disparity between the balanced outputs and SE wrt this device (im sure I remember folks suggesting this could be true of other devices we have tested) so maybe it would be worth looking at this phenomenon in other devices , Speculate on the causes and god forbid conduct some listening tests on a extreme example to see if these discrepancies are audible.

I believe the DAC 3 is using ES9028PRO a 32bit DAC. Does anyone find it strange it is exhibiting such low resolution in SE mode? Maybe less than an ideal implementation in the SE portion of the output stage? Maybe someone can contact the manufacturer and see if they might like to weigh in on the subject.

 

[tobes]

Maybe less than an ideal implementation in the SE portion of the output stage?

Given the balanced results, that would appear to be the case - assuming there is no testing anomaly or production unit deviation.
I wouldn't get caught up in the '32 bit' thing, since the current state of the art analog circuits can only resolve to 20-21bits. Nonetheless many dacs get close to this, including the DAC3 in balanced mode and the dirt cheap D50.
The bigger question is how subjectively important is this particular test? Is it just wanking over specifications?

 

[gvl]

Isn't 32 bit mainly for internal processing, eg volume control?

 

[Blumlein 88]

How different this review and measurement would seem if Benchmark had XLR out only.

Doesn't change how excellent the Topping gear is. But without the unbalanced result muddying the waters this DAC is right up there with the best measured by Amir so far. That is all shown so far that is a surprise, that unbalanced is somewhat compromised vs balanced. Even then it isn't a big difference.

It's probably too late, but I wonder how this would measure with some XLR to RCA cables used on the balanced outputs? Maybe that is the better way to do it.

 

[rebbiputzmaker]

How different this review and measurement would seem if Benchmark had XLR out only.

Doesn't change how excellent the Topping gear is. But without the unbalanced result muddying the waters this DAC is right up there with the best measured by Amir so far. That is all shown so far that is a surprise, that unbalanced is somewhat compromised vs balanced. Even then it isn't a big difference.

It's probably too late, but I wonder how this would measure with some XLR to RCA cables used on the balanced outputs? Maybe that is the better way to do it.

Even if it measures better with a balanced to SE cable, how does this reflect on the manufacturer who claims quality engineering? This is also sold as a consumer DAC, not just a pro-DAC and many home users utilize SE with source components. The Exasound E32 was panned for the same issue, does this DAC get a pass?

 

[svart-hvitt]

How different this review and measurement would seem if Benchmark had XLR out only.

Doesn't change how excellent the Topping gear is. But without the unbalanced result muddying the waters this DAC is right up there with the best measured by Amir so far. That is all shown so far that is a surprise, that unbalanced is somewhat compromised vs balanced. Even then it isn't a big difference.

It's probably too late, but I wonder how this would measure with some XLR to RCA cables used on the balanced outputs? Maybe that is the better way to do it.

I never thought someone would suggest «how this would measure with some XLR to RCA cables used on the balanced outputs» in a Benchmark product. Is it a benchmark (with a small b) product or not?

It seems like @amirm has managed to make some turbulence again, and this time in a product held in the very highest esteem among objectivists.

The response of some ASR members on the non-benchmark results of Benchmark’s unbalanced outputs is interesting as well.

 

[hvbias]

Where’s my bloody tea?! Get along amir , and don’t forget the digestives like last time..

No matter how much you berate amir he never learns , he’s a terrible man servant ..

Chocolate Hob Nobs for me, but I am a low brow fellow.

 

[RayDunzl]

Earlier, in reference to the Bypass/Calibrated Output function, I quoted the wrong manual, as the DAC3 DX (tested) has some different functionality from the DAC3 HGC and L versions (quoted) .

Dac3 DX Manual

Selecting Calibrated output (pg. 19)

I would be curious whether the Unbalanced output was measured using the Main or Aux outputs. (pg.26).

There are two output busses. The Main analog outs would seem to use the same signal as the XLR (three summed DAC channels), the AUX output uses the remaining DAC channel.

The DAC3 HGC and L use four channel summing for all the analog outputs.

 

[Blumlein 88]

I never thought someone would suggest «how this would measure with some XLR to RCA cables used on the balanced outputs» in a Benchmark product. Is it a benchmark (with a small b) product or not?

It seems like @amirm has managed to make some turbulence again, and this time in a product held in the very highest esteem among objectivists.

The response of some ASR members on the non-benchmark results of Benchmark’s unbalanced outputs is interesting as well.

It's more that I'm puzzled why Benchmark included compromised outputs. Especially as using adapters wouldn't cause any problems. Maybe it's to have two separated outputs. If so they cheaped out on the spare set I would say.

It's also a good example of why every product is worth measuring.

 

[tobes]

I would be curious whether the Unbalanced output was measured using the Main or Aux outputs. (pg.26).

There are two output busses. The Main analog outs would seem to use the same signal as the XLR (three summed DAC channels), the AUX output uses the remaining DAC channel.

Well picked up!
That would be very interesting to know.

 

[svart-hvitt]

It's more that I'm puzzled why Benchmark included compromised outputs. Especially as using adapters wouldn't cause any problems. Maybe it's to have two separated outputs. If so they cheaped out on the spare set I would say.

It's also a good example of why every product is worth measuring.

Yes, «puzzled» is indeed the right word here.

And «cheap out», if that’s what they did, doesn’t sound good when they use «BENCHMARK» in their brand name.

 

[RayDunzl]

It's more that I'm puzzled why Benchmark included compromised outputs.

Seems to allow separate volume levels between the Main and Aux busses - one could be calibrated while the other is variable, or the opposite, while utilizing the per channel volume control of the ESS chip, in this version of the DAC3.

 

[RayDunzl]

Well picked up!

All I did was RTFM.

 

[RayDunzl]

Yes, «puzzled» is indeed the right word here.

I might be most puzzled by the "jitter" measurement.

 

[amirm]

Earlier, in reference to the Bypass/Calibrated Output function, I quoted the wrong manual, as the DAC3 DX (tested) has some different functionality from the DAC3 HGC and L versions (quoted) .

Dac3 DX Manual

Selecting Calibrated output (pg. 19)

Ah, that explains why it randomly worked for one of the other inputs and not USB. I was holding both buttons down:

I will try again and see what happens.

I would be curious whether the Unbalanced output was measured using the Main or Aux outputs. (pg.26).

There are two output busses. The Main analog outs would seem to use the same signal as the XLR (three summed DAC channels), the AUX output uses the remaining DAC channel.

The DAC3 HGC and L use four channel summing for all the analog outputs.

I have it disconnected now but I am 99% sure I was testing the Main output. I will try the Aux one and see if there is a difference.

Thanks for looking this up.

 

[amirm]

For several times I mentioned the importance of the actual audio files.

https://www.audiosciencereview.com/...dio-precision-analyzer-apx555.3442/post-83427

But there is no reply. I don't know if it is the limitation of AP (that it can't save any audio files) or he didn't read my comment. Without audio files there will be no listening tests.

Sorry, I did not see your response before.

Answer is that Audio Precision does not use files for these tests. The signals are dynamically generated. In this case, a -20 dB 200 Hz tone is used as the starting calibrated level. It then starts playing the 200 Hz at -120 and steps to -60 one at a time. There is no way to save that file but you can create your own if you have proper software.

This type of automation/control is lacking on a lot of other software which runs a non-varying waveform and then takes a snapshot followed by FFT analysis. The AP has closed loop control and synchronization between its generator and analyzer.

The AP can take a file and run that through the test too including changing its volume control.

 

[HifiGuy]

Hi Amir,

on page 20 of the manual is described how you can change the usb mode. Could you please look whether you are in the usb 2.0 mode or not?

thank you.

 

[Blumlein 88]

I might be most puzzled by the "jitter" measurement.

JA has seen those two sidebands roughly where Amir found them on other Benchmark DACs.

 

[Blumlein 88]

Even if it measures better with a balanced to SE cable, how does this reflect on the manufacturer who claims quality engineering? This is also sold as a consumer DAC, not just a pro-DAC and many home users utilize SE with source components. The Exasound E32 was panned for the same issue, does this DAC get a pass?

I'm not sure why my idea equates to giving them a pass. If anything a good result with adapters to RCA would combine with the lesser results on their RCA to paint a lesser picture of the device's design trade-offs. The idea to try it with adapters is to show unbalanced results need not be compromised. Even then the compromising is rather tepid. It's within .5 db all the way to -120. I seem to recall the Exasound was a bigger difference than that if my memory isn't failing me.