Review and Measurements of SMSL VMV D1 DAC

[derp1n]

This is the Feature of the ESS.

Why doesn't it show up with all DACs using the chipset, then? John Siau from Benchmark's explanation is (more or less) that many manufacturers are leaving out parts in the name of cost savings. Which could make sense for a $200 device, but would be surprising on a $1300 flagship DAC like SMSL's D1.

Measured under different settings.
View attachment 15369

If this was in response to my question, Amir's graph is IMD vs level, not THD.

 

[helloworld]

Why doesn't it show up with all DACs using the chipset, then? John Siau from Benchmark's explanation is (more or less) that many manufacturers are leaving out parts in the name of cost savings. Which could make sense for a $200 device, but would be surprising on a $1300 flagship DAC like SMSL's D1.


If this was in response to my question, Amir's graph is IMD vs level, not THD.

If you look at the imd test, it looks like it’s better than oppo 205 besides the hump part. I am just guessing, maybe oppo did something to get ride of the hump but sacrificed part of imd performance? Benchmark is very good but they use 9028pro chip, maybe it’s the 9038 problem?

 

[derp1n]

The hump is definitely specific to the 9038. IIRC correctly it's to do with a particular setting that can be changed on the DAC chip, the trade-off being higher 2nd harmonic distortion. I'm a bit fuzzy on the details now though, and most manufacturers do not change this setting. I think Benchmark did something a bit more sophisticated than just changing the chip setting, but I don't recall the details...

If you're referring to the THD compensation feature, can you point to some proof that it's implicated in this?

Benchmark's solution is linked from my post #101.

Really, adding more and more speculation and badly remembered information isn't adding anything useful to this conversation.

 

[andreasmaaan]

If you're referring to the THD compensation feature, can you point to some proof that it's implicated in this?

Benchmark's solution is linked from my post #101.

Really, adding more and more speculation and badly remembered information isn't adding anything useful to this conversation.

Comment deleted.

 

[mindbomb]

The Allo Katana also avoided the ess imd issue.
https://audiosciencereview.com/foru...and-applepi-raspberry-pi-dacs.4164/post-97196

although now that I look at it, it looks like the hump is present in the katana, just at -10db rather than the usual value.

 

[helloworld]

The Allo Katana also avoided the ess imd issue.
https://audiosciencereview.com/foru...and-applepi-raspberry-pi-dacs.4164/post-97196

although now that I look at it, it looks like the hump is present in the katana, just at -10db rather than the usual value.

No enough data for katana’s IMD test

 

[amirm]

Btw, there is a reason why AES17 is recommended when measuring DAC.

There is no such recommendation. AES has specified a filter to be used when needed such as a DAC that has heavy noise shaping. Or a class D amplifier. In both cases the ultrasonic content may upset the measurement gear (e.g. slew rate limiting of the front-end). So AES has specified that should you need to filter ultrasonics, what accuracy filter should be used.

With respect to the test here, we are using sample rate of 192 kHz and that absolutely should be tested to Nyquist. Here is AES-17 spec itself:

And Audio Precision's note on the same:

DACs that support 192 kHz sampling need to show clean response to 96 kHz. There would be no use for them if the bandwidth is limited to 20 kHz.

You can see an example of ultrasonics causing problems (in this case the auto-scalar): https://audiosciencereview.com/foru...ex-xb120bt-and-dual-am505bt-car-stereos.3695/

We are not seeing issues like this in DACs.

 

[amirm]

Thanks for the review at first.
And can you change the Brand name VMV D1 for us.
This is the connection picture when we measured.Connect the earthing port to ensure the correct data.
View attachment 15344
View attachment 15345View attachment 15346 Here.
Looking forward to hear from you.Thank you very much.

I had tried this before I posted my review but it does not fix the problem. Here are the FFT results with (black) and without (red) grounding jumper:

As we see, there is a slight reduction in high frequency spectrum but there is also a rise in mains leakage (60 and 210 Hz).

This is what it looks like for Balanced again:

Since we are dealing with switching/high frequency noise, how the device is grounded naturally makes some difference but the root cause is internal.

 

[gvl]

Balanced looks very clean. However it was mentioned the AP's XLR inputs are transformer coupled with very good common mode rejection, and real world XLR inputs aren't necessarily as good at that.

https://www.audiosciencereview.com/...easurements-of-benchmark-dac3.3545/post-90519

It would be interesting to see if the HF junk is present on a single pin of the balanced output or it's cancelled internally.

 

[amirm]

As a way of comparison, here is Topping DX7s RCA outputs vs SMSL VMV D1 with grounding jumper (-70 dBFS):

The switching noise in SMSL D1 (red) is quite obvious.

 

[amirm]

And here are the results from 48 to 192 kHz Sampling on SMSL D1 (grounded):

These are all with 90 kHz bandwidth which allows us to see the switching noise. Amplitude changes but the effect is there at all sample rates.

 

[amirm]

A-weighting (blue) definitely cleans up all sins:

So hopefully SMSL is seeing the problem now after turning that off.

 

[bennetng]

How about disconnect the USB cable and use toslink? Apart from the switching noise I also like to see how the D1 deals with toslink jitter.

 

[gvl]

A-weighting (blue) definitely cleans up all sins:

View attachment 15399

So hopefully SMSL is seeing the problem now after turning that off.

One must wonder why was it on in the first place.

 

[Krunok]

With respect to the test here, we are using sample rate of 192 kHz and that absolutely should be tested to Nyquist.

While I have absolutely no intention to defend SMSL DAC in any way I must say this measurement seems completely detached from the purpose of any audio device, unless they are designed for bats and not for humans.

DACs that support 192 kHz sampling need to show clean response to 96 kHz. There would be no use for them if the bandwidth is limited to 20 kHz.

I failed to understand the use of 192 kHz anyway as I cannot possibly imagine that perceived SQ can benefit from it, hence I see no point in measuring it.

I mean, why stop at 192, why not measure 384 or higher?

 

[amirm]

How about disconnect the USB cable and use toslink? Apart from the switching noise I also like to see how the D1 deals with toslink jitter.

Toslink only goes up to 96 kHz. So here is that compared to 192 for USB:

Something is wrong here. We have our large switching noise but now a continuous amount of oscillation with Toslink (red).

 

[bennetng]

Toslink only goes up to 96 kHz. So here is that compared to 192 for USB:

View attachment 15400

Something is wrong here. We have our large switching noise but now a continuous amount of oscillation with Toslink (red).

Interesting. Could you show the J-Test at 48k and 96k? Something horrible may show up I suppose?

 

[derp1n]

Did Michael J Fox draw the red sinusoid?

 

[amirm]

And here is balanced which again, shows excellent response:

 

[Krunok]

Did Michael J Fox draw the red sinusoid?

Was this supposed to be funny?

 

[Krunok]

Toslink only goes up to 96 kHz. So here is that compared to 192 for USB:

View attachment 15400

Something is wrong here. We have our large switching noise but now a continuous amount of oscillation with Toslink (red).

It definitely is, but do you care to blind test this DAC against DX7s as you did with the Schiit?