SMSL SU-1 Stereo DAC Review

[radix]

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.

I think you are referring to this text?

11.1 Power Supply Decoupling Capacitors The LME49720 requires adequate power supply decoupling to ensure a low total harmonic distortion (THD). Place a low equivalent-series-resistance (ESR) ceramic capacitor, typically 0.1 µF, within 2 mm of the V+ and Vpins. This choice of capacitor and placement helps with higher frequency transients, spikes, or digital hash on the line. In addition to the 0.1 µF ceramic capacitor, it is recommended to place a 2.2 µF to 10 µF capacitor on the V+ and V- pins. This larger capacitor acts as a charge reservoir, providing energy faster than the board supply, thus helping to prevent any droop in the supply voltage.

I do not know what SMSL used, but maybe their engineers used what is needed for audio bandwidth signals based on the expected drive loads? Usually, it's OK to use larger PS caps (of the right voltage), but is it necessary? The board PS might be just fine for the use. Have you seen a problem with the output?

 

[levidos]

My son has an SMSL unit and likes it a lot. The CXN has very nice streaming software built in. I have a Cambridge streamer and I think it has the best streaming ergonomics I've seen.

Tbh I'm very happy with the WiiM's interface and most of my music is on Amazon Music so I'd keep using it anyway.

 

[ahofer]

Tbh I'm very happy with the WiiM's interface and most of my music is on Amazon Music so I'd keep using it anyway.

Well then certainly don't waste money on the CXN. Get any of the cheap recommended DACs here. Or even use the internal DAC in the Wiim.

WiiM Mini Review (Streamer)

This is a review and detailed measurements of the WiiM Mini Wifi/Bluetooth streamer. It is on kind loan from a member and is on sale on Amazon for US $89. As you can see the former factor is a "hockey puck." An included phone charger needs to be used to power the unit over USB-C. It did not...

www.audiosciencereview.com

 

[levidos]

Well then certainly don't waste money on the CXN. Get any of the cheap recommended DACs here. Or even use the internal DAC in the Wiim.

Isn't the DAC on the CXA81 better than the WiiM mini?

 

[ahofer]

Isn't the DAC on the CXA81 better than the WiiM mini?

Not much. Has some eccentricities. I doubt you could tell the difference.

Cambridge Audio CXA81 Review (Sample 2)

This is a follow up to my first test of the Cambridge Audio CXA81. Results there showed fair bit of noise which was uncharacteristic of other products I have tested from the company. There were a lot of speculation as to the source of the problem but no data to back any. Fortunately company...

www.audiosciencereview.com

 

[Herbert]

I think you are referring to this text?



I do not know what SMSL used, but maybe their engineers used what is needed for audio bandwidth signals based on the expected drive loads? Usually, it's OK to use larger PS caps (of the right voltage), but is it necessary? The board PS might be just fine for the use. Have you seen a problem with the output?

Correct, referring to this text. In Amir‘s multitone measurements we see a spike next to 6kHz. I also assume the PS could be anemic. The SU-1 seems to use only 0.1 uF. I would like to know whether such a large capacitance can be used after DC to DC upconversion (and switching of polarity). Unfortunately I cannot identify the parts that convert USB-power to +/-12V. Maybe the converter circuit is hidden in the “brick” close to the outputs?

 

[radix]

Correct, referring to this text. In Amir‘s multitone measurements we see a spike next to 6kHz. I also assume the PS could be anemic. The SU-1 seems to use only 0.1 uF. I would like to know whether such a large capacitance can be used after DC to DC upconversion (and switching of polarity). Unfortunately I cannot identify the parts that convert USB-power to +/-12V. Maybe the converter circuit is hidden in the “brick” close to the outputs?

It's likely that STC chip right by the power inlet. I cannot make it out from the photo. (I was looking at the photo upside down, I cannot tell which part is doing the DC-DC). Also, what is the value of those two giant caps? Those are likely the PS caps. The 0.1 uF caps are noise filters that should be positioned close to the opamp.

If you add the 10uF, the V+ is the obvious polarity (+ to V+ and - to ground). The second 10 uF is the opposite (- to V- and + to ground).

 

[Nutul]

Urban Dictionary: /S

s/: Typically used in IRC or IM to indicate that the previous typo should be replaced by the correct spelling of the word.

www.urbandictionary.com

When I do, I rather use html5-like tags, as
<sarcasm>
</sarcasm>
or, more frequently
<rant>
</rant>

 

[Nutul]

Why is driver required for a USB DAC? And only in windows, but not in Linux? It's hard to think of another USB device like that.

Because in 2023 Windows does not yet implement the USB AUDIO PROTOCOL 2.0 natively in the kernel, so you need drivers to supply it. Linux, MacOS and Android (which is Linux) don't suffer of this "sickness" ;-)

 

[staticV3]

Because in 2023 Windows does not yet implement the USB AUDIO PROTOCOL 2.0 natively in the kernel, so you need drivers to supply it. Linux, MacOS and Android (which is Linux) don't suffer of this "sickness" ;-)

Windows supports UAC1 and UAC2 audio devices natively with no manual driver install required. It does this via a set of generic USB drivers that come preinstalled with the OS.

 

[Nutul]

Windows supports UAC1 and UAC2 audio devices natively with no manual driver install required. It does this via a set of generic USB drivers that come preinstalled with the OS.

I know, I should have said "natively and completely full-standard".

Why are these drivers required? For the A6, for the Z6 / Z8... everybody (ok, you get what I mean...) complains that without the drivers the units are limited in functionality / samplerate support / bitdepth and whatnot (somebody even argues about latency settings, clearly missing the point, but this is another story...).

Windows does support UAC2 to some extent, but not completely, e.g max 24bit and max 384kHz, and no DSD native transport, as I read about jumping in and out of forums.
Me, I have no complains, as I am on Linux, but reading here and there keeps me busy trying to understand.

 

[staticV3]

Windows does support UAC2 to some extent, but not completely, e.g max 24bit and max 384kHz

32bit 384kHz I believe is the highest that UAC2 can do, which the generic driver built into Windows fully supports:

 

[Nutul]

32bit 384kHz I believe is the highest that UAC2 can do, which the generic driver built into Windows fully supports:
View attachment 312981

Yes, you are right; although I remember having read of somebody complaining that he couldn't see any 32 bits option in the list (with the A6) until he installed the recently released driver.
Anyway, still no 768kHz (or higher), and still no native DSD more than 256 (4x). Not a great of a deal, but... other OSs can do it.

 

[chatnever]

I've just received mine.

First impression is that both musicality and transparency are excellent. Especially when comparing to my AK4493 based DAC from BRZ-Hifi, which used an unlicensed version of Amanero for DSD decoding.

The BRZ has a oled screen, volume controll, and bluetooth, but the sound is so-so, slightly dulled, not warm nor bright/detail enough, RCA output is pretty low, I've to crank up the volume up to the point it may distort the sound, before sending to the amp.

You might reconsider when purchase a DAC from BRZ.

Meanwhile, listening to the SU-1, the different is night and day. it is very musical, yet maintain the detail and airy, bass has more power, treble is delicate. Sometimes, I think the treble comes from a Sabre DAC, but the overall, SU-1 is much enjoyable (when comparing to my usb dac with es9018k2m, which sound a bit dry, analytical).

 

[Herbert]

It's likely that STC chip right by the power inlet. I cannot make it out from the photo. (I was looking at the photo upside down, I cannot tell which part is doing the DC-DC). Also, what is the value of those two giant caps? Those are likely the PS caps. The 0.1 uF caps are noise filters that should be positioned close to the opamp.

If you add the 10uF, the V+ is the obvious polarity (+ to V+ and - to ground). The second 10 uF is the opposite (- to V- and + to ground).

Many thanks, your help is very much appreciated but I think you poke around in the dark more than me. The AKM Dac can be clearly seen between the big caps, so they cannot be part of the supply as distances to capacitors always need to be short. But as I suspected, what I called a “brick” seems to do the conversion from 5V to +/-12 V. “S105O12x2-1W” indicates this. They call it “Power Purification” in their ads.

https://www.amazon.com/S-M-S-L-M200-AK4497EQ-High-Resolution-Bluetooth/dp/B0887VGTT3

Unfortunately googling the part number leads to nothing. I cannot imagine that they make the part themselves though SMSL is branded on it. Anyway, I would need to find out whether putting 10uF after the brick would be a benefit or a hazard. The parts labeled 101 could be ferrite beads
to suppress high frequency noise…?

 

[radix]

Many thanks, your help is very much appreciated but I think you poke around in the dark more than me. The AKM Dac can be clearly seen between the big caps, so they cannot be part of the supply as distances to capacitors always need to be short. But as I suspected, what I called a “brick” seems to do the conversion from 5V to +/-12 V. “S105O12x2-1W” indicates this. They call it “Power Purification” in their ads.

https://www.amazon.com/S-M-S-L-M200-AK4497EQ-High-Resolution-Bluetooth/dp/B0887VGTT3

Unfortunately googling the part number leads to nothing. I cannot imagine that they make the part themselves though SMSL is branded on it. Anyway, I would need to find out whether putting 10uF after the brick would be a benefit or a hazard. The parts labeled 101 could be ferrite beads
to suppress high frequency noise…?

I do really think those 2200 uF caps are the power supply caps. At that size, they only filter low frequencies and do not need to be particularly close to anything so long as the traces are large enough. high frequency filter caps need short traces. I don't know what else one would use 2200 uF caps for in a DAC. As they are 10V, they're likely on the 5V supply to the AKM, or maybe on the 5V reference voltage output for the AKM's analog stage. It's all guessing.

Anyway, my point was that the SMSL engineers would not leave out a 10uF capacitor if it was really needed. Have you or anyone measured voltage sag to the opamp during high output?

 

[Herbert]

I do really think those 2200 uF caps are the power supply caps. At that size, they only filter low frequencies and do not need to be particularly close to anything so long as the traces are large enough. high frequency filter caps need short traces. I don't know what else one would use 2200 uF caps for in a DAC. As they are 10V, they're likely on the 5V supply to the AKM, or maybe on the 5V reference voltage output for the AKM's analog stage. It's all guessing.

Anyway, my point was that the SMSL engineers would not leave out a 10uF capacitor if it was really needed. Have you or anyone measured voltage sag to the opamp during high output?

Also large caps need short traces in digital circuits. SMSL could also simply have left out the two 10uF caps recommended by TI to cut costs - while the caps at the DAC are a must. They did cut the SPDIF transformer at the Coax input, also not a must. But I just put one in, the dropouts are completely gone now. The circuitry is standardized, the transformer just needs to be inserted between the Cinch-jack and pcb. A very delicate job as the 3-cinch socket has to be completely soldered out and as the pins are hole-through, it is soldered from both sides of the pcb. Still heating from the bottom is sufficient but while gently bending the pcb and prying out the socket in very little steps one has to take care not to rip the delicate traces from the pcb.

 

[morillon]

Also large caps need short traces in digital circuits. SMSL could also simply have left out the two 10uF caps recommended by TI to cut costs - while the caps at the DAC are a must. They did cut the SPDIF transformer at the Coax input, also not a must. But I just put one in, the dropouts are completely gone now. The circuitry is standardized, the transformer just needs to be inserted between the Cinch-jack and pcb. A very delicate job as the 3-cinch socket has to be completely soldered out and as the pins are hole-through, it is soldered from both sides of the pcb. Still heating from the bottom is sufficient but while gently bending the pcb and prying out the socket in very little steps one has to take care not to rip the delicate traces from the pcb.

to check if there is not a transformer already on the source side before this somewhat delicate manipulation...
;-)

 

[Herbert]

to check if there is not a transformer already on the source side before this somewhat delicate manipulation...
;-)

Still more convenient than opening the lid of any SPDIF Coax- source one encounters. If there was one on the source side, there would be no drop outs. In studio setups, transformers on both sides are mandatory. This is why Toslink came up -much cheaper than a dedicated transformer to break ground loops.

 

[morillon]

Still more convenient than opening the lid of any SPDIF Coax- source one encounters. If there was one on the source side, there would be no drop outs. In studio setups, transformers on both sides are mandatory. This is why Toslink came up -much cheaper than a dedicated transformer to break ground loops.

there is no need to have them on each side e etc.. just a precaution... and the transition to aes...

in absolute terms even if their performance is very good it is still preferable to have only one absolute performance side, and even more if we go up in resolution
((you can use optocoupler too)