SMSL M500 DAC and HP Amp Review

[RickSanchez]

Would be handy to put these measurements side by side with DX7 Pro like someone did with the Sabaj unit.

That was me. I’m traveling right now but happy to put up a SMSL M500 vs. DX7 Pro comparison later in the week (and before the 11/11 sale.)

 

[ryohnosuke]

@amirm
In a group they have done many tests with the M500. The point is that the SMSL activates the audio filters when it is in MQA and this should not happen. MQA Limited spoke about it and said the M500 is not doing it correctly. When it is MQA it must block all types of filters because what it reproduces is supposedly the original master. Please tell us if you notice audible differences in MQA, regards.

What group?

 

[Jaysz]

Liking this dac with my hs7's and krk sub

Going to use xlr splitter's rather than sending signal through sub filters.
Still not sure what filters are doing or what led colours mean or why have to scroll through menu to switch screen on if of as cant even see the menu
or why firmware on screen shows different to computer.

 

[gpauk]

Anyone have any idea how much MQA adds to the price of the product?

 

[Patrick1958]

Anyone have any idea how much MQA adds to the price of the product?

The Matrix X Sabre pro MQA is listed 1.945 €
The Matrix X Sabre pro (non MQA, no longer listed in seller shop) was 1.595 €
350 € difference.
But then that's Matrix Audio high end gear.

 

[MediumRare]

For those of you considering whether to buy the M500 or the DX7 Pro, the DX7 Pro has full control over which outputs (XLR/RCA/Headphone) are active: All, one, or any combination are possible. Plus the display is very clear about what's on. And it's very easy to switch using the remote. Just sayin'.

 

[gpauk]

The Matrix X Sabre pro MQA is listed 1.945 €
The Matrix X Sabre pro (non MQA, no longer listed in seller shop) was 1.595 €
350 € difference.
But then that's Matrix Audio high end gear.

Hmm... So rather more than a license fee... Shame they took away the user choice - but then that's a big uptick in unit profit so no surprise I guess!

 

[amirm]

Anyone have any idea how much MQA adds to the price of the product?

I have not read any info on that. I can say from experience (we developed our own format which we licensed and help set prices on open standards in patent pools) that on the low side it is a few cents per device to high side of $2.50 which was the MPEG-2 video license. MPEG got away with that high fee given its dominant position in the world.

 

[VintageFlanker]

For those of you considering whether to buy the M500 or the DX7 Pro, the DX7 Pro has full control over which outputs (XLR/RCA/Headphone) are active: All, one, or any combination are possible. Plus the display is very clear about what's on. And it's very easy to switch using the remote. Just sayin'.

The M500 has for itself an unusually high max output (6.90V, same as ADI-2 DAC). If used as a preamp, it may deal with any power amp sensitivity.

 

[NielsMayer]

Glad to see it do so well especially the amplifier, about as good as DX3 low gain? Nice!

It's really weird though, I saw reports of people complaining about anemic XLR output? When it goes to max 6.8V?! Odd.

I suppose the output impedance is 2ohm SE and thus ~4ohm balanced?

The comments i made about anemic output were for USB mode only. It was comparatively and jarringly "hot" on other outputs, which means you had to be careful to turn down volume to 10 or less when switching inputs, or you'd be blasted by sound.

The update to the 1.04 -> 1.07 indicates, per @SMSL-Mandy "The new firmware. Mainly solve the problem that some users have proposed before, the Android phone sound is small when connecting with M500" ... I'd say the reason for the delay/wait for Amir's review was that SMSL wanted to ship the upgraded firmware. Which is good because that is the unit people will be purchasing, after all the complaints about issues with 1.04 firmware.

Since android uses Linux underneath, the "android phone" fix should also apply to my use-cases of Linux and Android-TV (NVIDIA Shield Pro 2015). Alas, since the firmware upgrade requires windows or mac,I cannot verify and am sending my unit back via amazon prime free returns (which was why i was willing to gamble purchasing an unreviewed unit in the first place).

But like I said, on Coaxial input fed via USB from a Topping D10, or by a Raspberry Pi with DigiPi HAT, it sounds really good, better than the Topping DX7S it was to retire.

One test I'd like to see @amirm add to all his DAC tests is a 50Hz square wave signal fed into the DAC and measured at the XLR outputs under specified XLR balanced cabling impedance (600 ohms i believe) load. So as to see what the output op amps slew rate and output capacitors do to the bass response of the unit, especially in driving a "real world" length of XLR cable say 6-10 meters (and maybe a second measurement with 10-20cm of cable, aka little inductive and capacitive load, as "control"). Because it's easy to measure a sine wave down to 20Hz, but most of the "bass" we hear is a fundamental and overtones, so if analog outs can't pass a nice 50Hz square-wave with relatively sharp edges, and without phase delay between fundamental and overtones, the bass will sound all wrong (aka lacking in "thumpiness" and "attack" characteristics).

Likewise love to see a "teardown" or circuit diagram of the XLR output circuitry of the M500 vs the Topping DX7Pro -- what kind of op-amps, coupling capacitors?, DC coupled (if only...), real balanced (using a pair of amplifiers) or "fake" balanced output (using a single amplifier 1/2 noise and distortion, but can't drive a long XLR line), etc ????

 

[dkinric]

The M500 has for itself an unusually high max output (6.90V, same as ADI-2 DAC). If used as a preamp, it may deal with any power amp sensitivity.

Good point.
@March Audio , do you see any issue with one of your new Purifi based amps handling this type of high output signal?

 

[NielsMayer]

It's built around an ES9038PRO, so no. ESS chips use a redundancy in bit depth to avoid that problem.

See http://www.esstech.com/files/3014/4095/4308/digital-vs-analog-volume-control.pdf for details.

 

[Soniclife]

So I tried this. Quite fascinating. When I turn on DSP in Roon, the "LED" indicator on M500 changes color from green to magenta. I can't make heads and tails out of the colors from either M500 manual or what they have posted from MQA:

However, I popped up the Roon pipeline with DSP on and off and got this useful information:

View attachment 37226

So what is happening is that when you turn on DSP, roon internally decodes the MQA stream. It then ups the bit depth to 64 bit, applies the DSP and then embeds the MQA flag that says this is MQA but is not the original bitstream ("ORFS"). The signal at this point is decoded PCM stream with just the MQA flag. I suspect the DAC is still allowed to upsample to 192 kHz (this content is 96 kHz so I can't test it).

Some info from roon on what they do, it's all sensible and matches what you found.
https://kb.roonlabs.com/Roon_x_MQA

 

[armani006]

Is it bad idea to attach M500 preamp outputs with two ARCHEL2 PRO AMP as balanced amplifier, one for one channel? Archels are very very transparent for signal, M500 has ideal THD and Dynamic range. Seems like it can be good system, it is also good that M500 can control volume of pre-amp outputs..

 

[Soniclife]

By chance, I found some test tones in Tidal that are MQA encoded! They use a frequency of 375 Hz for some reason. I performed a quick comparison, streaming the Tidal MQA versus straight PCM from my analyzer to M500. Here is the comparison (at 44.1 kHz):
View attachment 37229

Everything is identical until we get near end of this spectrum (90 kHz) with a rather strong spike to -60 dB. My guess is that the gentle MQA filter is allowing an aliasing component to show up there. Not an audible thing but information nevertheless.

I don't understand how an aliasing artefact can be larger than the real signal?
Can you repeat this test with a none MQA DAC, with roon doing the software decode. And then maybe get roon to upsample that with their slow rolloff filter.

 

[filo97s]

The figure 8 power cable is not as optimal as an IEC connector but does make for a lighter weight cord.

*average audiophile triggered*

 

[amirm]

I don't understand how an aliasing artefact can be larger than the real signal?

Larger? The real signal is at 0 dB (the tall peak way to the left). The aliasing component is 60 dB lower at -60 dB.

 

[Soniclife]

Larger? The real signal is at 0 dB (the tall peak way to the left). The aliasing component is 60 dB lower at -60 dB.

But the real signal is at 300hz or something, not high frequency, how can that leak.

If you test with this track with a none MQA DAC does the noise floor change if you disable software decode?

 

[Rusty Shackleford]

Indeed, you are quite right .

I appreciate very much that Amir made this measurement and found out about that huge spike artifact caused by MQA!

If people could enlighten me, there one 2 things I don't quite understand as they don't seem to be really consistent to me .

First, re. such huge ultrasonic artifacts:
- On one hand, during the fantastic TotalDAC discussion on ASR this summer ( the whole thing was so incredible that I read the 100 pages twice!, and that really motivated me to join ASR ), such ultrasonic components were found out, a few subjectivists said that it didn't matter as they cannot be heard, yet what I perceived as the "dominant", "enlightened" view an ASR was that it was really bad because it could quite negatively affect the electronics of further devices (e.g. an amp for speakers) downstream or the behavior of transducers ( tweeter, headphones).
- On the other hand, when it is created by MQA it is OK because it cannot be heard .

Second, re. apparently fraudulent marketing claims:
- When this is the case with a device such as Schiit, TotalDAC, PS Audio, it is really bad (and OK to speak of " scam ").
- With MQA it would be OK because part of the necessary marketing discourse for establishing a new standard .

Nailed it!

 

[VintageFlanker]

Would love to see this tested by Amir.

Spoiler: it won't perform well.