Review and Measurements of Pro-Ject Pre Box S2 Digital

[amirm]

Looking forward to your later thoughts on the distortion compensation feature. Jussi has mentioned (on that CA Forum thread) this is a feature of the ESS chip itself, not the DAC or XMOS interface of the DAC and that it only helps, it doesn't hurt in any of his measurements.

For sure. But there can be firmware bugs in how it is enabled. In the simplest case, it could be setting it to on when we ask to turn it off and vice versa. I pulled a muscle in my back and haven't been to my lab. Let me see how I feel a little later and I will try to get to the bottom of this.

 

[amirm]

But one potential significant advantage could be eliminating a path for ground loops or mains leakage. One would then hope the noise output is then regulated by whatever it's powering, by low noise linear regulators.

Your point is correct there. I have a couple of these banks and will do a test.

 

[Randem Tox]

I don't believe noise output is the advantage that's discussed with these powerbanks - at least it shouldn't be, for the reason you mention, i.e. the noise output can be no different to a wall wart.

Sure, that's one reason for going with the USB power-bank (even if that's not the impression I've gotten reading various posts on CA/HF about their use). However, the post I was responding to brought up the USB power-bank as an alternative to an external LPS - so my assumption there was that, given an LPS would still have the potential for a ground loop the issue of concern (by that poster) was noise on the USB power-line.

Either way, that's two possible benefits to the USB power-bank approach.

 

[Randem Tox]

But there can be firmware bugs in how it is enabled.

I believe this is documented as being the case for the Distortion Compensation ... in "Best" mode, at least, (from memory) the setting and display values are inverted. I'll have to find the post where this is discussed.

Given that Pro-Ject have just released a firmware updater for macOS, and have reposted the 2.12 firmware, I expect there will be more updates in the near future and hopefully this will get fixed, along with a proper fix for the MQA drop-out issue.

 

[Music1969]

Sure, that's one reason for going with the USB power-bank (even if that's not the impression I've gotten reading various posts on CA/HF about their use). However, the post I was responding to brought up the USB power-bank as an alternative to an external LPS - so my assumption there was that, given an LPS would still have the potential for a ground loop the issue of concern (by that poster) was noise on the USB power-line.

Either way, that's two possible benefits to the USB power-bank approach.

Sure. Some people probably assume USB mobile powerbanks have the noise/ripple and impedance of a car lead acid battery - of course they don't but more importantly they don't need to either since there is usually very good voltage/ripple regulation downstream, in good gear at least.

The S2 DAC marketing material says it uses low noise linear regulators. I'm sure Amir will verify once he performs surgery

 

[amirm]

OK, here is the test of distortion compensation enabled/disabled. Let me be exact on how I tested.

I power cycled the unit after being on for many hours.
I selected "Test" mode and pushed the knob in.
I checked Distortion Compensation and it is Enabled.

I measured the 1 Khz residual noise and distortion.

I then changed Distortion Compensation to Disabled which turned the mode to "user" (I made no other changes).

I measured again. Here are the results:

What we see here is that when we turn off Distortion compensation, we actually get a reduction in distortion in 2nd harmonic. There is a trade off elsewhere though with third harmonic actually increasing. We also see some small spikes with compensation enabled that are not there when we turn it off. So the overall picture is a mix.

Remeasuring THD+N at different frequencies we get:

So when we sum the total distortions+noise, distortion compensation is doing the opposite of what its name indicates.

Digging in further, we see this note from Benchmark Media on their DAC performance: https://benchmarkmedia.com/blogs/application_notes/dac3-introducing-the-new-es9028pro-converter

This is very much consistent with above measurement where the largest improvement does come in 2 and 6 Khz. The problem is, to get that improvement I had to turn Compensation to Disabled.

Anyway, I think the designer needs to examine this and make sure that the UI matches what it is telling the chip to do.

And that we need to be aware that this is no a panacea in that some distortions are traded for others.

Let me know if you all agree with this analysis.

 

[Music1969]

Let me know if you all agree with this analysis.

Hi Amir

As per Miska's posts on CA Forum, can you set the filter to 'linear phase, fast rolloff' (just "fast rolloff" on the GUI) with distortion compensation ON?

This should show as 'User' mode.

He mentioned on that CA Forum thread that the above filter setting in custom 'user' mode measures better than both 'test' and 'best' modes and in 'fast roll off' you see the differences of distortion compensation ON/OFF much clearer. He recommends staying away from 'test' or 'best' modes for measurements.

 

[amirm]

Hmmm. I don't know why changing filters would change distortion characteristics here. Here are the last test with the filter changed from Brickwall to Fast Rolloff with both Compensation enabled and disabled:

Green and Cyan are the new measurements with the different filter. As you see, in both cases there is no change in distortion. Green lands on yellow and Cyan on top of red.

 

[Music1969]

Green and Cyan are the new measurements with the different filter. As you see, in both cases there is no change in distortion.

Yes this is consistent with what Miska noted, which I mentioned earlier, that there's no harm done with distortion compensation enabled. Your measurements show no advantage of using it but also no harm is done also.

He seems to prefer fast roll off. It seems certain filters are causing increased distortion then?

Your findings look consistent with Miska's - 'test' and 'best' modes are best avoided and overall performance of the DAC is good.

 

[Music1969]

Hmmm. I don't know why changing filters would change distortion characteristics here.

Maybe as a follow-up, for completeness, could you compare distortion measurements for each of the 7 digital filters?

And just verify if distortion compensation makes any difference (or harm) for each of the 7 filters?

 

[amirm]

Yes this is consistent with what Miska noted, which I mentioned earlier, that there's no harm done with distortion compensation enabled. Your measurements show no advantage of using it but also no harm is done also.

No, no. The measurements show that filter changes don't make a difference. Enabling and disabling Distortion compensation makes a significant difference.

 

[amirm]

Maybe as a follow-up, for completeness, could you compare distortion measurements for each of the 7 digital filters?

And just verify if distortion compensation makes any difference (or harm) for each of the 7 filters?

That's is a lot of measurements. As I noted, changing filters make no difference in distortion/noise measurements in the two cases I tried.

 

[Music1969]

No, no. The measurements show that filter changes don't make a difference. Enabling and disabling Distortion compensation makes a significant difference.

'Test' mode is using a different filter, so it's not an apples to apples comparison though, right?

'Test' mode distortion is higher than 'user' mode in fast roll off yes? It's using a different filter...

I'm definitely not saying the filters result in different distortion measurements, but I'm asking the question. Why is test mode measuring worse than user mode.

 

[amirm]

To be clear there is no "user mode" per se. If you make any changes, it will change "Test" (or "Best") into "User." In my first set of measurements I provided today, all I did was enable/disable distortion compensation and it was that which forced it into User mode. It didn't change any filters.

 

[Blumlein 88]

Sorry, I don't remember where, but I read this distortion compensation reduces the 3rd harmonic distortion. Your test results bear that out. No one mentioned it made other things worse in the process. Probably because all subjectivists know even order distortions are good, and 3rd order is bad juju. How weird design by marketing can become.

I still await you testing for Velvet sound in AKM DAC chips.

 

[amirm]

OK, here are all the filters with and without distortion compensation enabled:

So as you see the filter choice is invariant in this equation.

But this does show some extreme issues with Optimal transient and to much smaller extent the min phase modes.

 

[amirm]

Sorry, I don't remember where, but I read this distortion compensation reduces the 3rd harmonic distortion. Your test results bear that out. No one mentioned it made other things worse in the process. Probably because all subjectivists know even order distortions are good, and 3rd order is bad juju. How weird design by marketing can become.

I still await you testing for Velvet sound in AKM DAC chips.

All of this would be more clear if some chip companies had not started to keep their specs behind non-disclosure agreements. I like to Read what ESS is saying here.

 

[ManglerFi]

Thanks amir for the extra testing here. These results are quite interesting. So with distortion compensation on we get less 3rd harmonics but a higher THD+N overall but with the setting turned off we get a lower 2nd harmonic and lower THD+N? Certinialy would have thought that the THD+N would have been lower with the setting on.

Not knowing much of anything about harmonics, what kind of effect do they have on the audio and do different order harmonics affect the audio differently?

 

[DonH56]

Odd-order harmonics tend to sound more like a raspy buzz where even harmonics are not as obnoxious. Odd harmonics also cause intermodulation distortion that lies near the main signals so again tends to sound worse than even order IMD products that lie near DC and at ~2x the frequency and up.

Amir, you need to plot performance vs. phase of the moon, just in case certain phases cause it to work better. There is precedence as this old data sheet shows:

 

[amirm]

That's a riot Don!