SMSL SU-1 Stereo DAC Review

[yovko82]

The ears are not the only one, for what this is done. Don`t forget it.

Please elaborate

Why, if you can`t remember and think only for ears.
Only a joker - MPX signal - for example... Is it enough or may be to list all the possible coding or decoding stuff, that can be made?

 

[staticV3]

Why, if you can`t remember and think only for ears.
Only a joker - MPX signal - for example... Is it enough or may be to list all the possible coding or decoding stuff, that can be made?

I'm sorry, I don't understand what you're saying

 

[yovko82]

I'm sorry, I don't understand what you're saying

That is why, you only think about ears. Sorry

 

[ahofer]

I'm sorry, I don't understand what you're saying

I don't either.

 

[rongon]

At the risk of entering yet another sample rate flame war....

The idea is that higher sample rates allow better filtering of ultrasonic noise by raising the Nyquist frequency (half the sampling frequency). If you want to pass a 20kHz sine wave cleanly and the recording's sampling frequency (rate) is 44.1kHz, you have to construct a filter that lets 20kHz through with no attenuation but that attenuates over -60dB at only 22.05kHz. That's a very steep filter, which is more likely to 'ring' (create short duration oscillations not present in the original signal).

Many DACs already upclock the sampling frequency internally to avoid this; they take a data set with 44.1kHz sample rate and upsample it to 352.8kHz before applying the filtering. I can see this at work by looking up the MPD audio tab in Moode Audio on my Raspberry Pi with TI PCM5122 DAC, for instance.

I think this is more of an issue in music production rather than reproduction, but it could be an issue in certain reproduction (playback) setups. For instance... Back around 1998 or so, I used to master demo CDs for musicians. They'd bring me their stereo mixdown DAT tapes (16bit/48kHz) or WAV or AIFF files and I'd master them to CD so they could take the master CD to a duplication service. I was using a whiz-bang PC with a fancy RME soundcard, running Steinberg WaveLab and Waves VST plugins for digital signal processing. I noticed that if I could work with 24-bit 48kHz (or 96kHz) files to begin with, I'd end up with a smoother and more easily listenable result than if I was given a 16-bit 44.1kHz file to begin with. I would do all my processing in 24 bits (or 32 bit float) and only downsample to 16 bit/44.1kHz when I was ready to print the final mastered file for CD. It sounded noticeably better that way. Why would that be?

Ever since then, my attitude has been that if you can transfer a precious analog master tape to 24-bit/192kHz PCM using great analog gear and a high quality ADC, then why the heck not do that? If you want to make that transfer to DSD128 or even DSD512, well why the heck not? More data is better, because you never know what that extra information might be used for in 10 or 20 years.

As for straight up playback, I noticed that the latest 24/192 transfer of John Coltrane "A Love Supreme" captured the bias frequencies of the tape deck used to play back the master tape. The bias frequencies (one was 38.4kHz) are captured in the digital file. It's useless information since it gets filtered out in the D-to-A process, but it was there in the analog domain when you played the tape, right? We're talking about one of the great recorded masterpieces of the 20th century, so it should be digitized with as much fidelity to the source as we can possibly attain. If you want the ability to play it back that way too, you'll need high sampling rates like 96kHz, 192kHz or whatever. It's technically possible, so why not do it? I would rather hear "A Love Supreme" just like it came from the 4-track tape deck than I would want to listen to a data-reduced copy, especially if it costs me nothing extra to attain that level of fidelity to the source.

Sorry, I got a bit carried away.

 

[Herbert]

At the risk of entering yet another sample rate flame war....

The idea is that higher sample rates allow better filtering of ultrasonic noise by raising the Nyquist frequency (half the sampling frequency). If you want to pass a 20kHz sine wave cleanly and the recording's sampling frequency (rate) is 44.1kHz, you have to construct a filter that lets 20kHz through with no attenuation but that attenuates over -60dB at only 22.05kHz. That's a very steep filter, which is more likely to 'ring' (create short duration oscillations not present in the original signal).

Many DACs already upclock the sampling frequency internally to avoid this; they take a data set with 44.1kHz sample rate and upsample it to 352.8kHz before applying the filtering. I can see this at work by looking up the MPD audio tab in Moode Audio on my Raspberry Pi with TI PCM5122 DAC, for instance.

I think this is more of an issue in music production rather than reproduction, but it could be an issue in certain reproduction (playback) setups. For instance... Back around 1998 or so, I used to master demo CDs for musicians. They'd bring me their stereo mixdown DAT tapes (16bit/48kHz) or WAV or AIFF files and I'd master them to CD so they could take the master CD to a duplication service. I was using a whiz-bang PC with a fancy RME soundcard, running Steinberg WaveLab and Waves VST plugins for digital signal processing. I noticed that if I could work with 24-bit 48kHz (or 96kHz) files to begin with, I'd end up with a smoother and more easily listenable result than if I was given a 16-bit 44.1kHz file to begin with. I would do all my processing in 24 bits (or 32 bit float) and only downsample to 16 bit/44.1kHz when I was ready to print the final mastered file for CD. It sounded noticeably better that way. Why would that be?

Ever since then, my attitude has been that if you can transfer a precious analog master tape to 24-bit/192kHz PCM using great analog gear and a high quality ADC, then why the heck not do that? If you want to make that transfer to DSD128 or even DSD512, well why the heck not? More data is better, because you never know what that extra information might be used for in 10 or 20 years.

As for straight up playback, I noticed that the latest 24/192 transfer of John Coltrane "A Love Supreme" captured the bias frequencies of the tape deck used to play back the master tape. The bias frequencies (one was 38.4kHz) are captured in the digital file. It's useless information since it gets filtered out in the D-to-A process, but it was there in the analog domain when you played the tape, right? We're talking about one of the great recorded masterpieces of the 20th century, so it should be digitized with as much fidelity to the source as we can possibly attain. If you want the ability to play it back that way too, you'll need high sampling rates like 96kHz, 192kHz or whatever. It's technically possible, so why not do it? I would rather hear "A Love Supreme" just like it came from the 4-track tape deck than I would want to listen to a data-reduced copy, especially if it costs me nothing extra to attain that level of fidelity to the source.

Sorry, I got a bit carried away.

I have read somewhere that the algorithms of certain professional audio plugins work best at certain sample rates, like the 24/96kHz you mentioned. I do not have the time to dig out the sources. But this might explain your experiences. About „A Love Supreme“: The bias frequence was never meant to pleasure the HiFi - enthusiast.

 

[AnalogSteph]

I have read somewhere that the algorithms of certain professional audio plugins work best at certain sample rates, like the 24/96kHz you mentioned.

That specifically affects anything nonlinear. The result of any nonlinear processing is not inherently band-limited, which in the frequency-periodic world of digital audio can create some nasty aliasing artifacts. Case in point, the sound quality of mainstream releases became so much better after switching to oversampled brickwall limiters (pretty much the ultimate nonlinearity) around 2012.

Do note that some effects will oversample internally, those do not need special treatment. Others may also unintentionally scale with fs, that's no good either.

 

[BradC]

I'm new to posting but have lurked for a while.

Just wanted to say thanks for the detailed review!

I just added a SMSL SU1 (with a wiim pro) to let me finally stream to my vintage stereo. The combo sounds terrific and brings new utility to this old amp for a very reasonable price.

Wiims are starting to multiply in my house too

(And yes the speakers are way to close, the room is mid rearangment)

 

[AudioNewbie]

Is this a DAC that requires the power button to be turned on each time it's connected to a power supply?

 

[DSJR]

I'm new to posting but have lurked for a while.

Just wanted to say thanks for the detailed review!

I just added a SMSL SU1 (with a wiim pro) to let me finally stream to my vintage stereo. The combo sounds terrific and brings new utility to this old amp for a very reasonable price.

Wiims are starting to multiply in my house too

(And yes the speakers are way to close, the room is mid rearangment)

View attachment 312239

Oh hell - Linn/Naim dealer hat thought long buried has been located and inserted where the sun don't shine...

You don't want to read this probably so apologies but...

PLEASE - large speakers either side of your LP12 will still create some feedback if records are played at a spirited level (those 'Brik stands oscillate sideways too (been there and done that) and placing the SNAPS alongside the 32 preamp does introduce hum into the phono stage, if less than a HiCap does - I know it was forty years ago, but I did this with a friend's LP12/Naim 32-Snaps then Hicap-250/Isobarik system and imagined or not, losing the hum induction into the phono stage did 'seem' to make the sound better... A 32 preamp on phono (MC and MM) should only hiss when turned up full. Your situation as pictured can only put some hum into the vinyl input I feel...

The 72 filter boards will refine the sound from 'digital' sources well and a change to 72 improves things further (not sure why but many dems and samples indicated this)

Apologies again - I knew this gear well and owned/sold/installed plenty myself back then and honestly only trying to help you get a better sound from the rig you have.

I'll crawl back under my stone now...

 

[BradC]

Oh hell - Linn/Naim dealer hat thought long buried has been located and inserted where the sun don't shine...

You don't want to read this probably so apologies but...

PLEASE - large speakers either side of your LP12 will still create some feedback if records are played at a spirited level (those 'Brik stands oscillate sideways too (been there and done that) and placing the SNAPS alongside the 32 preamp does introduce hum into the phono stage, if less than a HiCap does - I know it was forty years ago, but I did this with a friend's LP12/Naim 32-Snaps then Hicap-250/Isobarik system and imagined or not, losing the hum induction into the phono stage did 'seem' to make the sound better... A 32 preamp on phono (MC and MM) should only hiss when turned up full. Your situation as pictured can only put some hum into the vinyl input I feel...

The 72 filter boards will refine the sound from 'digital' sources well and a change to 72 improves things further (not sure why but many dems and samples indicated this)

Apologies again - I knew this gear well and owned/sold/installed plenty myself back then and honestly only trying to help you get a better sound from the rig you have.

I'll crawl back under my stone now...

The speakers are just that close in the picture as I am rearranging the furniture in the room. Usually they will live a lot further apart, because they sound crap at bookshelf speaker spacing

I haven't noticed any phono hum since I had the SNAPs rebuilt by Naim recently.

I'll do a little research into the filter board differences on the 72, and have been researching some 3rd party board upgrades for the tape input. There is some very very low level hiss (barely audible when music isn’t playing, I cant hear it over even low volumes), but for higher fidelity I have more modern stereos in other rooms. This system is my nostalgia machine that now let's me add streaming to my office.

Cheers,
Brad

 

[Herbert]

Is this a DAC that requires the power button to be turned on each time it's connected to a power supply?

No. I shut my complete stereo setup off with a mains switch. After providing power again, the SU-1
remembers / stays on the last chosen input, at least when running it from an USB-power supply…

 

[radix]

I've started using the SU-1 between my RPi 4 (volumio) and Anthem STR preamp. When I used USB, Volumio would go away when the Str was off and then take a while to come back up on the network when the STR was turn back on. I am also using speaker groups in Volumio. The on/off of the USB was just causing a lot of headaches. With the STR being always on with the RPi4, Volumio seems much happier.

 

[AudioNewbie]

No. I shut my complete stereo setup off with a mains switch. After providing power again, the SU-1
remembers / stays on the last chosen input, at least when running it from an USB-power supply…

That's nice. Thank you!

 

[Herbert]

That's nice. Thank you!

De rien! But I have to mention that on coax (I did not try Toslink yet) i have 3 seconds dropouts i.e. when turning a light switch.
But I live in an old house without proper mains grounding. I will put in a dedicated 1:1 SPDIF transformer to check whether this changes anything. But evidently, the SU-1 has no decoupling from coax SPDIF ground which as far as I know is recommended - but would of course raise costs.
A proper transformer (i.e. from the company "Pulse Electronics" PE 65612) is about $6.

 

[radix]

De rien! But I have to mention that on coax (I did not try Toslink yet) i have 3 seconds dropouts i.e. when turning a light switch.
But I live in an old house without proper mains grounding. I will put in a dedicated 1:1 SPDIF transformer to check whether this changes anything. But evidently, the SU-1 has no decoupling from coax SPDIF ground which as far as I know is recommended.

toslink is likely the bandage you're looking for if you have ground issues.

 

[Herbert]

toslink is likely the bandage you're looking for if you have ground issues.

But limited to 24bit/96kHz. Toslink is also not my main source for listening.

 

[Herbert]

BTW while doing this mod I also want to put in additional 10uF electrolytic capacitors at the voltage rails of the lme49720 opamp the SU-1 uses.
This is recommended in the datasheet of Texas Instruments.
But I have no clue what DC-DC converter is being used by SMSL to provide the +/-12V the opamp needs from the 5V input.
Any ideas what IC is being used? I am asking this question because maybe putting in a higher capacitance at the opamp's supply
is ok with a dedicated dual supply but contraproductive with a DC-DC upconverter.

 

[DSJR]

The speakers are just that close in the picture as I am rearranging the furniture in the room. Usually they will live a lot further apart, because they sound crap at bookshelf speaker spacing

I haven't noticed any phono hum since I had the SNAPs rebuilt by Naim recently.

I'll do a little research into the filter board differences on the 72, and have been researching some 3rd party board upgrades for the tape input. There is some very very low level hiss (barely audible when music isn’t playing, I cant hear it over even low volumes), but for higher fidelity I have more modern stereos in other rooms. This system is my nostalgia machine that now let's me add streaming to my office.

Cheers,
Brad

I do apologise for slipping back to 1983, I just cannot help it

My current hearing issues and this overdamped room here have made me look at Isobariks again, but with thoughts of a more modern less hard toned amplifier able to drive the three ohm or so midrange load they present. There was a really good era for passive 'Briks starting in March 1983 with the MDF carcasses and lasting until KEF changed the B110 mid driver in 1985 or thereabouts, but I can't remember the serial numbers now sadly. My second walnut pair were dated August 1984 and they were pretty good.

72 filter boards in a 32.5 went part way, but the 72 as an entity 'better still' we thought, the biggest differences heard in the then early domestic digital sources. There was a variable gain input board made, but this had extra filtering over 15kHz I remember and back then with more youthful hearing, none of us liked it I remember! I only bang on about it because the up-front and rather brightly lit 'dead' sound got rather better as the years went on (the 52 preamp was a revelation but at a high price) and the biggest sonic improvements were to that new fangled 'digital' source that people like 'us' were so suspicious of

 

[levidos]

I have the following setup: Wiim Mini, CXA81, KEF LS50 Meta, and thinking of upgrading my DAC. I was thinking to get the CXN V2, but would I get the same improvement by using this SMSL SU-1?