Cheap SA9027 + ES9023 (or PCM5102) DAC, anyone try?

magman

Member
2011-11-01 7:55 pm
Kaunas
Hello,

I'm looking for inexpensive dac. Found this
SA9027 ES9023 24bit 96kHz Asynchronous USB DAC HiFi Sound Decoder with Case | eBay
And DA3 SA9227 PCM5102A 32bit 384kHz USB DAC HiFi Asynchronous Decoder Case | eBay

They look like same but only dac chip is diferent. which one is better?

Another option is to make AMB "gamma2" dac and feed i2s with something like whis: WL High End CM6631A 7 Pin USB Daughter Card Free Shipping 1504A | eBay instead of 60$ AMB "asynchronous USB-I²S module".

+use external power supply.

Any advice?
Thank you
 

sharpi31

Member
2005-10-20 12:57 pm
I bought the DA3 (SA9227 with PCM5102A) and this is working fine with Windows (with Savi driver) and RPi running Volumio. I haven't opened it up yet but believe this is the PCB:
 

Attachments

  • image.jpeg
    image.jpeg
    111.3 KB · Views: 2,268

sharpi31

Member
2005-10-20 12:57 pm
The 10uF electrolytics appear to be output coupling caps, and if so these can be removed/bypassed. After that I will check the capacitance before and after the AMS1117 regs, and add/replace if need be. I'd be keen to try a lower noise alternative to the AMS1117 but haven't yet identified the best option. I'm not going to spend any significant money, and am keen to keep everything within the existing case. Basic tweaks where these can make a difference.
 

sharpi31

Member
2005-10-20 12:57 pm
The 10uF caps were output coupling, so have been removed. I measured approx 12uF on the output of each AMS1117 reg, and also 12uF on the USB 5V connected to the 2x AMS1117 inputs. As a first pass I added 100uF from input to ground at each AMS1117. No change to reg output caps.

When I first listened to the DAC (before mods) it didn't have any obvious problems. It sounded nice - maybe a bit less HF and punchier in the bass than the ES9023 I2S RPi DAC I'd been using beforehand. There was a very slight texture to the sound (not roughness, but heading in that direction) but nothing overtly unpleasant.

I've listened for approx 10mins since doing the mods detailed above. The slight roughness has gone and the HF is more open. There is an increase in depth. Pretty much what I'd expect after removing un-biased electrolytic coupling caps (so quite possibly a psychological element going on!).
 

Attachments

  • IMG_0211.jpg
    IMG_0211.jpg
    560.7 KB · Views: 2,270

Alexium

Member
2010-05-25 12:44 pm
I'm going to buy the same DAC (but a couple bucks cheaper on Aliexpress). Two questions:
1. Don't you think those decoupling caps were necessary? Did you measure DC offset on the outputs?
2. Can you compare the sound quality to other, more thoroughly designed DACs with current output? Like PCM 1794?
 

sharpi31

Member
2005-10-20 12:57 pm
No need for output coupling caps with the PCM5102A - I've just measured +0.3mV DC on both L & R outputs.

The best DAC I have up and running right now is the SMSL M8 (ES9018K2M). I haven't done a proper comparison, but will do soon. I have a Buffalo IIISE almost completed too, so that will be a good standard for comparison. That will be used in voltage output mode into my TVC preamp.

All I can say for now is that the DA3 isn't doing anything obviously wrong (no audible noise, distortion, sibilance etc.) and has good dynamics and bass weight. Mids and top are cleaner and clearer since my mods.

I'm sure (hope?) I'll notice more when I do a proper side by side comparison with the SMSL and Buffalo. Having said that I'd be surprised if the differences were dramatic. Whether it's my hearing getting worse or budget DACs getting better (most likely both) I've not heard as much difference between DACs in recent years.
 

Jensen567

Member
2016-02-13 9:33 pm
Thought I might chime in as I recently purchased the bare PCB version with the PCM5102 (no real reason for that over the ESS other than I have a Ti bias).

SA9227 PCM5102A 32bit 384kHz USB DAC HiFi Asynchronous Decoder | eBay

I bought it mostly to play around with the SA9227 as I would like to use that chip in an upcoming project if possible. Still need to try to find some more info on how to program the chip, may be a proprietary tool to volume customers only. Anyway, that doesn't matter here, so back on topic.

I like the board, sound is very good, doesn't seem to add much if any color to the sound. No hiss or any other noise to speak of when it is powered but not playing anything which is better than I can say for some other cheap DACs I have tried.

I actually like it so much that I plan to embed this board into an APPJ PA1502 Tube headphone amplifier I have so I can play straight from USB when I am done probing at its chips :)

For comparison, my "daily driver" DAC which gets the most use is an Audiotrak Dr.DAC3

http://audiotrack.net/en/index.php?document_srl=2524&ckattempt=1&mid=dac

The Dr.DAC3 contains a VIA VT1731 USB controller and a PCM1794 DAC connected to a TPA6120 headphone amplifier.

The Dr.DAC is definitely a bit clearer and cleaner, but it also costs 10x as much, and I have not modded my SA9227 board at all, just installed the driver in Win 10 and have listened for probably 12 hours over the past few days since I got it. Would recommend if you don't want to spend a bunch of money but still want a quality DAC.

Most of my listening is done with headphones, AKG K7XX, through the TPA6120 in the Dr.DAC (I am using the RCA input directly to the amplifier to test the SA9227 board). Some of the listening was done with my tube amps (same headphones), again the SA9227 board was mostly transparent here, comparable to using them with the RCA outputs on the Dr.DAC which bypass the amplifier.
 
The 10uF caps were output coupling, so have been removed. I measured approx 12uF on the output of each AMS1117 reg, and also 12uF on the USB 5V connected to the 2x AMS1117 inputs. As a first pass I added 100uF from input to ground at each AMS1117. No change to reg output caps.

When I first listened to the DAC (before mods) it didn't have any obvious problems. It sounded nice - maybe a bit less HF and punchier in the bass than the ES9023 I2S RPi DAC I'd been using beforehand. There was a very slight texture to the sound (not roughness, but heading in that direction) but nothing overtly unpleasant.

I've listened for approx 10mins since doing the mods detailed above. The slight roughness has gone and the HF is more open. There is an increase in depth. Pretty much what I'd expect after removing un-biased electrolytic coupling caps (so quite possibly a psychological element going on!).

Are you sure that 10uF are in series with output? I would have thought they are between Analog VDD and analog GND?
 
The 10uF caps were output coupling, so have been removed. I measured approx 12uF on the output of each AMS1117 reg, and also 12uF on the USB 5V connected to the 2x AMS1117 inputs. As a first pass I added 100uF from input to ground at each AMS1117. No change to reg output caps.

If this device is host-powered only, you should take into account that everything > 1uF is not USB-conform when not having inrush-current-limit or softstart.

The slight roughness has gone and the HF is more open. There is an increase in depth. Pretty much what I'd expect after removing un-biased electrolytic coupling caps (so quite possibly a psychological element going on!).

What is to expect from un-biased vs. biased electrolytic caps?

Regards.
 

sharpi31

Member
2005-10-20 12:57 pm
My DAC is connected to a Raspberry Pi so is essentially connected to the 5V PSU for the RPi (not a typical USB power implementation). The additional capacitance doesn't cause issues here, nor on the Windows laptops I've also tried. Good to know this may be problematic in some setups - I wasn't aware of this limitation.

The PCM5102A output is 2.1Vrms (2.97Vpp) centred on 0V, so the output signal will swing to almost -1.5V. Polarised electrolytics don't tend to like reverse bias of this magnitude, and subjective experience (own & others) is that coupling electrolytics tend to sound worse when not biased properly. The PCM5102A is designed to not require output coupling caps, so no reason to keep them (unless the minuscule DC offset is a problem).
 
Thanks for the reply.

It might be problematic if some kind of potentiometer is following the outputs before connecting the the next (biased coupled) amplifier stage. -1.5V isn't something I would care/mind about but this depends on the caps used. Some "like" it, others might not but they will some sort of "reformat" to this over time.

(Beside this, I never use elythics in direct signal paths)

For the cap limitation on USB, well most host controllers can deal with it but it isn't compliant to the specs. Most possible trouble s when they do some kind of enumeration or fancy data-line toggling, like charge controllers or current-controllers. Depending on the current demands of the DAC a simple current limiting series resistor helps and gives additional filtering (RC) as a benefit.
 
On a related note, I just discovered on Ebay and then ordered a DAC that combined the SA9023 and ES9023 (see pic below). Still waiting for it to arrive. I'm hoping that it runs in ADAPTIVE USB MODE, unlike the board that is pictured in post #2, which I think runs in asynchronous mode.

I need adaptive mode because I sometimes used multiple USB stereo DACs connected to an R-Pi, which I am using an a multichannel DSP crossover unit by running IIR filters in software under Linux (Raspbian).

What I would like to know is how I can verify that the SA9023 is indeed implemented using adaptive mode. Does that show up if I plug in the DAC to some computer (Windows, Linux) and run some query (e.g. aplay, use Windows device manager, etc.)?


s-l1600.jpg
 
Last edited:
If this device is host-powered only, you should take into account that everything > 1uF is not USB-conform when not having inrush-current-limit or softstart.
Regards.

One simple workaround and improvement might be to get a USB power splitting cable, like this none:
USB 2.0 A Female to 2 Dual USB Male Hub Power Adapter Y Splitter Cable Cord | eBay

And then use an external USB power adapter (or a USB battery), no limit on the caps and most likely a lot less noise from the psu.
 
The 10uF electrolytics appear to be output coupling caps, and if so these can be removed/bypassed. After that I will check the capacitance before and after the AMS1117 regs, and add/replace if need be. I'd be keen to try a lower noise alternative to the AMS1117 but haven't yet identified the best option. I'm not going to spend any significant money, and am keen to keep everything within the existing case. Basic tweaks where these can make a difference.

Did you try shorting the output caps before doing the mod, or is that not a good idea?