SA9227 PCM5102A Asynchronous 32bit 384Khz USB DAC

mroboto

Super Member
I already have of two of these DACs, just ordered the third.

I slightly modify them (swapping and adding few parts) for 2 reason,

1st. If you own this DAC and don't add filtering capacitors, you may have experienced some freezing problems, the USB SA9227 chip gets hot and stop transmitting signal, the issue is fixed for me by adding small and low cost capacitors.

2nd. it sounds great (better) once modified, must hear to believe, for a DAC in this price range (Including the price of the additional parts).

Original (Click on the link at the right of "Doccument joint" to see full size):
https://www.cjoint.com/c/HAfcRveUSlt

Modded:
https://www.cjoint.com/c/HEieOQLFqc6

With two Wima MKS2 10uf 50V as output coupling capacitors.

One 4.7uf 50V Multilayer ceramic capacitor on the regulator on the left in front of the SA9227 USB receiver chip.

One 10uf 50V Multilayer ceramic capacitor on the regulator on the right in front of the PCM5102A DAC chip.

Rear:
https://www.cjoint.com/c/HEieQ0Hvq76

And one 2.2uf 50V Multilayer ceramic capacitor on the USB 5V input.


The SA9227 Windows driver can be found here:
https://support.hifimediy.com/suppo...vitech-sa9227-uda38pro-uda18-uh1-uh1-digital-

The DAC support up to 24bit 96Khz without the driver, or without restarting Windows,

But to support higher resolutions, the driver is required, and you have to start a Windows session after the DAC is plugged in the PC (If you unplug and plug it back, it will be recognized as FS Audio USB instead of HD USB Audio).


If you look for a low cost but quite good sounding USB DAC, that will last forever without the need to restore anything (no electrolytig capacotors to replace, ever), this is a good recommendation, but not if you have no intentions to mod it, at least, add filtering capacitors to make it reliable.

If you still having issues after upgrades, (not me), but the SA9227 chip is still warm, you can also add a 10x10x10mm heatsink on the SA9227 chip to keep it cool.


Note: Some may prefer to remove the output coupling capacitors and have DC coupled output, it's possible with this DAC chip, but I prefer to be safe for my tube amps and speakers using the Wima capacitors for output coupling.
 
Last edited:
I replaced the 2.2uf that I previously added to the 5V DC input for a 4.7uf , and added a second 10uf for a total of 20uf for the PCM5102A DAC chip regulator (the regulator on the right), they work and sound very fine.
 
It does look like the voltage regulators don't have the usual, recommended decoupling caps on input and output. I'd use surface mount components, close to the pins. But I have access to a professional rework station, and access to such components.

Identify the regulator devices. Read the application note. Add the recommended decoupling components.
 
I don't understand why anyone would buy a product that needs to be modified to work properly.:dunno:

To buy a second and/or third makes no sense to me.

I have no problem with modifying something to make it sound/work better. However, modifying it to make it work properly out of the box is a gigantic red flag for me.
 
Last edited:
It does look like the voltage regulators don't have the usual, recommended decoupling caps on input and output. I'd use surface mount components, close to the pins. But I have access to a professional rework station, and access to such components.

Identify the regulator devices. Read the application note. Add the recommended decoupling components.

Do you own this DAC or tell this only by pictures I posted ?

It uses SMD capacitors on it, not sure if it use decoupling before the regulators, but sure there is some after.

The results of the simple mods I done are very good, I look no further.

I don't understand why anyone would buy a product that needs to be modified to work properly.:dunno:

To buy a second and/or third makes no sense to me.

I have no problem with modifying something to make it sound/work better. However, modifying it to make it work properly out of the box is a gigantic red flag for me.

They have designed this DAC starting from the SA9027 ES9023 design that I know about, this one is nearly identical except for the chips it uses, and the first DAC had no problem, probably they thought this one will be perfect too, but no, they have not tested this design extensively.

The SA9227 chip is sensitive to noise in the power supply, and since the chip is sensitive to noise, it overheats, some DACs that uses this chip uses heatsink on it, but they didn't added one, with proper supply, it heats less, and work as it should work.

I first bought one to try, modded it and was impressed by the results, bought another for a second system (New portable computer for my bedroom), I use the first one in my main system when I don't use my TubeDAC-11, I like to change sometimes, and since they work fine, sounds great, and are cheap, I bought a third one for the future, I have 4 computers in total.

The last mods I done taked it to an higher level.
 
Last edited:
I too own and use multiple computers (5 total, 3 laptops and two desktops). Unlike many I have one main stereo/AV system. Never in life would I buy or otherwise acquire any device that needs to be modified to work properly. I'm all for modifying something to make it sound better. Modifying it to make it work properly doesn't work for me.
 
Last edited:
I don't try to sell you this DAC, just inform about my discoveries.

This is a desing flaw that is easily corrected, I understanded this quickly with my first.

And since a Multilayer ceramic capacitor will last forever, this flaw will never have to be corrected again, and the sound improved a lot by the way.
 
Last edited:
One thing I don't like with this DAC on a laptop with Windows 7, is that the screen won't shut down at all when you set it to shut down after 15 minutes in power management.

But on Windows 10 , it shut down as it should.
 
Sometimes, it didn't freeze one day, then it froze twice the day after, it was random.

Had to unplug and replug it.

And If I wanted it to recognize frequencies over 96Khz again, I had to restart computer so it is recognized as HD Audio not FS Audio.
 
Last edited by a moderator:
Do you own this DAC or tell this only by pictures I posted ?

Just looking at the pictures.

U1 and U3 will be the linear voltage regulators. It is normal to place decoupling capacitors as close to the pins as possible. It's clear that there are no capacitors near the pins of those devices, even though there is space to do so.

They are SO-223 package devices. Probably something like an LM2937 LDO regulator*. Here's the TI datasheet, which includes recommended decoupling (100nF input, 10uF output), and a recommended layout (figure 27, p18), with decoupling right next to the pins. It's also worth noting the ESR spec on the output capacitor (10mR to 3R), required to keep the regulator stable; you can't just stick any old capacitor on the output.

http://www.ti.com/lit/ds/symlink/lm2937.pdf

* I can't see the markings, so I can't confirm exactly what part they are.
 
Last edited:
They both are AMS1117 regulators.

Adding a multilayer ceramic before, (4.7uf at the 5V input) , and on the output pins (4.7uf on the left SA9227 regulator) (and 20uf on the right PCM5102A regulator) is, I think, a good choice, and I understand why it sounds better.

USB standard specify 10uf max at a specified "?" impedance for current surge, I think I do not exaggerate too much with what I done.
 
Last edited:
I made a clean install of Windows 10 build 1809 on my laptop this morning, and now the screen won't turn off, exactly as Windows 7.

As soon as the DAC is connected to the USB port, the screen won't turn off by any mean when a Windows session is started.

I don't know how I did this spring on build 1803 to make this work properly, I don't know what is to do or not to do, what's the trick ?

I searched the internet in the past, and there was a command to show what devices wake up the system, and you can configure by another command but this command shows nothing, no device wake up the system, so I'm lost.
 
With Foo ASIO I don't have to restart Windows ever, and sound very improved over Direct Sound.

And he didn't freeze like when original, rarely I have a problem, only disconnect and connect again sometimes when I resume computer from sleep, but not always, maybe I touched the cable when Windows was sleeping ??

And I found a newer driver here:

https://www.driverscloud.com/fr/services/GetInformationDriver/72064-73543/speed-dragon-sa9227zip

PS: these DACs works great with 6ft USB cables, but won't or have problems with 15ft with some motherboards or portable computers.
 
Last edited:
I ended up adding a second cap to the SA9227 regulator, a 10uf ceramic, so now it's 14.7uf on this regulator.
 
My understanding from a similar thread on diyaudio is that you don't need the coupling caps because the pcm5102a is designed to be able to used directly. I removed the caps on mine, and shorted with a wire instead. I tried directly hooking up from cap pads, to the rca outs, skipping the resistors but for some reason, at least according to my cheap oscilloscope $30, 1khz sound wave isn't smooth. I took those removed coupling caps and put it on the regulator, waiting for ceramics to be shipped. Removing the caps seemed to improve the highs, while putting the caps on the regulator seemed to improve a little the bass. Hopefully putting caps both on the Vin and Vout will improve things further.

All things said, it probably already sounds around 85-90% to my hifime UH1 which uses the same chips, but costed 4x-5x more.

As to why the driver with windows is messed up. I think the reason why is that the sa9227 may be counterfeit. I use it on linux, but it's similar on windows. If you reboot with it plugged in, the usb device id reads as 262a:9227, which allows usb 2.0 speeds. If you plug it in while the computer is on already, the usb id instead reads as 262a:9217, and you lose usb 2.0 speed, and instead only can use usb 1.0. That means only 96khz files work. That's why if you unplug and replugin, you lose proper driver support.
 
I mostly use tube amps, so I feel safer with an output coupling cap, even if i am tempted to try without.

When you use ASIO (Driver 64 bit + FooASIO) the sound is exactly the same when you reboot or not after plugging it in, I do not worry about counterfeit parts at all
I use them for about 3 years now, more than my TubeDAC 11, sometimes I roll them, but always come back to these little USB DACs

The ceramic on regulators are between Vout and Ground, and I think it's the better way to suppress noise to ground, not adding Vin noise to Vout

Have 14.7uf (10+4.7uf 50V ceramic caps) on the regulator in front of the SA9227 chip
And 20uf (10+10uf 50V ceramic caps) in front of the PCM5102A chip.

4 models of tube amplifiers to listen with (SE 300B / PP EL84 / SE KT88 / SE 6L6)

Il my main system I roll
Modded Paradigm 7SE speakers (Excellent with all the amps)
Dynaudio BM15 (better with the SE KT88 or PP EL84)
 
Last edited:
I'm only using solid state stuff, and planning to use it with my headphone amp because it seems to lack kernel support volume control (unlike my other sa9227 devices) meaning that if you adjust the volume on the computer, you have to use software mode, which I believe isn't bit perfect vs hardware kernel mode.

Hmm. I don't think that you should hear a sound difference unless perhaps you're output 192khz files, even then I'm not sure most can hear a difference. I also think that it's a real pcm5102a dac chip because it sounds too similar to my uh1 in sound signature, just that currently it's a little muddier. I only have an old slow windows 7 machine to play around with. And I don't like using it unless needed for legacy purposes. Can you check using https://www.uwe-sieber.de/usbtreeview_e.html, whether when you unplug your device and replug it in, that the device id is 262a:9227 (device id of a real chip) vs 262a:9217, where you lose usb 2.0 speeds thus only able to play 96khz files? When you reboot though with it plugged in, 262a:9227 loads up instead, and you get to use usb 2.0 speeds again. But if you unplug and replug afterwards, the device id reads as 262a:9217. This happens on linux and it looks like the same was happening on my old windows 7 machine.

As to the caps. Did you configure it like this?


In the video he puts an electrolytic cap on the vout and ground (although does it matter if you use a ceramic cap instead?), and a ceramic cap on the vin and ground.

Thanks.
 
The caps are as see in pictures in first post

https://www.cjoint.com/doc/18_05/HEieOQLFqc6_IMG-2112.JPG

https://www.cjoint.com/doc/18_05/HEieQ0Hvq76_IMG-2113.JPG

And yes, I remember something as such as 262a:9227 when plugged in before Windows startup and 262a:9217 when plugged on live Windows, and 24bit 96Khz Max. when plugged this way.

But Foo ASIO seems to fix the problem for me, the sound is the same in ASIO with or without reboot, anyway, the DAC is always plugged to my PC, and doesn't freeze anymore since the addition of the caps, so no need to unplug ever, except in case of bad contact with cable, that doesn't happens too often, I have a good 6ft silver plated and shielded cable with gold plated contacts.
 
I received some ceramic caps today. I left to electrolytics that had placed on the vin and ground on both regulators. And then placed a 4.7uf on the vout on the one for the sa9227 and 10uf on the one for the dac. It sounds pretty close to my hifime uh1 but that one's bass is still a tad stronger. I would say that the sound is 95-98% the same. Perhaps the main difference is that the hifime has way more caps, and clocks for the dac while this one just has a clock I think for the sa9227?

Anyways since I don't have any music above 96khz. It's pretty decent considering the cost. I actually just bought the dac to experiment with in order to understand what people are talking about when they talk about removing coupling caps, or trying to clean up the power supply signal. Kind of amazing how much difference making such small changes makes to the sound. When I first got it I thought it was kind of unlistenable.

My other sa9227 dac with a cs4398 chip was also missing caps on its regulator, but after I put some on its regulator the sound doesn't sound as "mushy" and there's more definition. Although I guess because of the dac chip, the sound is more laid back.
 
Maybe there is more bass with your other DAC because it have some kind of OP AMP for increased gain stage after the DAC chip ?

Adding stages change the sound, you like it, or not.

I purchased this DAC in first place because the DAC chip doesn't require any OP AMP to get line stage level.
 
Back
Top Bottom