Review and Measurements of Benchmark DAC3

[amirm]

Was the jitter test done with spdif? I have a feeling those spikes on the dac3 are from the src4392.

Please see: https://www.audiosciencereview.com/...easurements-of-benchmark-dac3.3545/post-85548

Same spikes remain on Toslink input as with USB.

 

[HifiGuy]

Hi Amir,

could you please do following tests with activated ht bypass:

• 1 khz sine wave
• jitter and noise spectrum
• 1 khz tone noise and distortion spectrum
• linearity

on the unbalanced output. Thank you!

 

[amirm]

Hi Amir,

could you please do following tests with activated ht bypass:

• 1 khz sine wave
• jitter and noise spectrum
• 1 khz tone noise and distortion spectrum
• linearity

on the unbalanced output. Thank you!

Measured Jitter and made no difference. I also can't get it to go in calibrated mode in USB. But in Toslink it worked (may be pilot error).

Here is the Dashboard showing 1 kHz tone response:

Looks like SINAD is improved by a few dB.

Linearity actually got worse with lower level (as one would predict);

 

[amirm]

I have a Lynx Hilo to still measure today folks. So please scream if you need something badly but otherwise I need to do all of that today besides fixing my broken sprinkler .

 

[HifiGuy]

Hi Amir,

thank you it seems that you are tested at usb audio 1.1 mode that why ht bypass couldn't activated on usb. Good luck on measuring the Lynx Hilo.

 

[mindbomb]

Please see: https://www.audiosciencereview.com/...easurements-of-benchmark-dac3.3545/post-85548

Same spikes remain on Toslink input as with USB.

It might be that the usb input also goes through the sample rate converter. It has somewhat high intrinsic jitter apparently.

 

[svart-hvitt]

Here is the comparison of unbalanced output with USB vs Toslink. Sadly the same spikes remain:

View attachment 13390

Thanks for the Toslink view!

Still, as suggested previously somewhere, I wonder if Toslink is the way to go due to its aesthetic advantages combined with very good jitter performance in well-engineered boxes.

 

[mindbomb]

I have a Lynx Hilo to still measure today folks. So please scream if you need something badly but otherwise I need to do all of that today besides fixing my broken sprinkler .

I just had a thought. Is there any chance the linearity on unbalanced is better with toslink?

 

[Dialectic]

Hi Amir,

could you please do following tests with activated ht bypass:

• 1 khz sine wave
• jitter and noise spectrum
• 1 khz tone noise and distortion spectrum
• linearity

on the unbalanced output. Thank you!

Lots of new taskmasters here.

 

[mindbomb]

@amirm
there is a typo in the linearity graph. -92db is 15 bits, not 14 bits.

 

[Blumlein 88]

@amirm
there is a typo in the linearity graph. -92db is 15 bits, not 14 bits.

That is not a typo. It is inherent bias against Benchmark since they never called him back about discount pricing and from being a paid shill of Topping. Don't let him fool you.

 

[xxie]

That is not a typo. It is inherent bias against Benchmark since they never called him back about discount pricing and from being a paid shill of Topping. Don't let him fool you.

Wow...

The DAC3 results honestly strike me a bit. I would expect a far better performance from just what I read about Benchmark. Thanks Amir doing us a huge service.

 

[mindbomb]

Wow...

The DAC3 results honestly strike me a bit. I would expect a far better performance from just what I read about Benchmark. Thanks Amir doing us a huge service.

I feel kinda the same way, in that I was expecting perfection. Most of the results were excellent, just jitter could have been better, and unbalanced linearity was poor.

 

[rebbiputzmaker]

That is not a typo. It is inherent bias against Benchmark since they never called him back about discount pricing and from being a paid shill of Topping. Don't let him fool you.

Either way, the Benchmark is not that good. Sad showing I must say for a firm that claims such high engineering standards.

 

[derp1n]

Only 14-bit linearity on unbalanced is very hard to believe.

 

[dallasjustice]

Only 14-bit linearity on unbalanced is very hard to believe.

When you turn the volume down, you lose dynamic range. It’s what you would expect.

 

[tobes]

Either way, the Benchmark is not that good. Sad showing I must say for a firm that claims such high engineering standards.

Don't know if I'd call it a 'sad showing', since measurements from the balanced outputs are excellent. However I agree that the unbalanced measurements are not what I'd expect from Benchmark (I own the AHB2 amp) and quite puzzling given given their pursuit of perfection in measurements.
Without undermining the results, given that Amir was unable to enter calibration mode from the usb input I wonder if this sample may have been faultly in some way? It would be interesting to test another sample and see if the results can be corroborated.
I can't see anything in JA's Stereophile measurements that would indicate the above issue - but maybe he didn't do a similar test from the unbalanced outputs?
JA's comments on resolution are more similar to Amir's balanced result:
"When the DAC3 decoded dithered 16- and 24-bit data representing a 1kHz tone at –90dBFS, the increase in bit depth dropped the noise floor by more than 30dB (fig.5), indicating that the Benchmark's resolution is at least 21 bits. This is as good as a DAC can currently get!"

I see also that JA tested the 'HGC' version which is slightly different in that it parallels 4 channels of the 9028 dac vs 3 channels for the 'DX' version that Amir tested. Though this is unlikely to influence the unbalanced linearity in isolation, maybe the other differences between the models has compromised the unbalanced outs in favor of the balanced outputs?

 

[derp1n]

When you turn the volume down, you lose dynamic range. It’s what you would expect.

What? Your post makes little sense because the volume/dynamic range thing doesn't apply here the way you're trying to apply it.

Amir's measurements of the Topping D10 and Behringer UMC204HD give 18.3 and 17.6 bits, respectively. It boggles the mind the DAC3 would perform so much worse.

 

[dallasjustice]

What? Your post makes little sense because the volume/dynamic range thing doesn't apply here the way you're trying to apply it.

Amir's measurements of the Topping D10 and Behringer UMC204HD give 18.3 and 17.6 bits, respectively. It boggles the mind the DAC3 would perform so much worse.

Amir’s measurements show that the DAC3’s DR pushes the limits of his AP. There’s only about -120db of real DR on the AP’s ADC. If you carefully review the OP’s first post, you’ll see the DAC3 and the RME are basically equivalent in terms of bit linearity. They are both basically 20 bit converters. Amir calls it at 19 bits, but that’s a close call, IMO. JA’s stereophile method of determining bit linearity is a different but equally legit method of evaluating bit linearity. He called it at 21 bits.
https://www.stereophile.com/content...preamplifier-headphone-amplifier-measurements


Both the RME and the DAC3 are scraping the limits of the measurement gear.

Big picture take away:
1. I’d bet good money that no human being will reliably be able to discern the difference between the DAC3 and any other converter in its league.
2. All audio gear is simply a tool used to obtain the desired fidelity. The DAC3 puts out a lot of juice. This is critically important for my setup. I knew that before I bought it. It was important to me because the JBL M2 horn tweeter needs a DAC with high output, assuming normal amp gain. This may seem counterintuitive since the horn is very sensitive. However, the D2 compression driver loses output precipitously in the upper end of its passband. The only way to get smooth output to 20khz with the M2 is to give it a lot of power and then DSP the target curve from there. The target curve will take out some DR but there will be plenty of range leftover for extremely dynamic playback when the user starts off with lots of voltage. Here is more info about my setup:
https://audiosciencereview.com/foru...m2-audiolense-digital-crossovers-w-subs.2369/

 

[derp1n]

Amir’s measurements show that the DAC3’s DR pushes the limits of his AP. There’s only about -120db of real DR on the AP’s ADC. If you carefully review the OP’s first post, you’ll see the DAC3 and the RME are basically equivalent in terms of bit linearity. They are both basically 20 bit converters. Amir calls it at 19 bits, but that’s a close call, IMO. JA’s stereophile method of determining bit linearity is a different but equally legit method of evaluating bit linearity. He called it at 21 bits.
https://www.stereophile.com/content...preamplifier-headphone-amplifier-measurements


Both the RME and the DAC3 are scraping the limits of the measurement gear.

Big picture take away:
1. I’d bet good money that no human being will reliably be able to discern the difference between the DAC3 and any other converter in its league.
2. All audio gear is simply a tool used to obtain the desired fidelity. The DAC3 puts out a lot of juice. This is critically important for my setup. I knew that before I bought it. It was important to me because the JBL M2 horn tweeter needs a DAC with high output, assuming normal amp gain. This may seem counterintuitive since the horn is very sensitive. However, the D2 compression driver loses output precipitously in the upper end of its passband. The only way to get smooth output to 20khz with the M2 is to give it a lot of power and then DSP the target curve from there. The target curve will take out some DR but there will be plenty of range leftover for extremely dynamic playback when the user starts off with lots of voltage. Here is more info about my setup:
https://audiosciencereview.com/foru...m2-audiolense-digital-crossovers-w-subs.2369/

What does any of this have to do with my original post?