DAC Schematic Review and Layout Suggestions - WM8804 and PCM5102A

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
After I'm done my first board and confirmed that it works, I'll give away the remaining ones (you have to pay shipping though:spin:) and post the Gerbers online.

Hi, please let me know if you will have available an extra board, I would be happy to pay also for the board and not just for the shipping :) keep us updated how it sounds ... I remember an old thread where one had to power the 5102 with 3,6V instead of 3,3 otherwise there were some serious digital clipping issues at max. level
 
Disabled Account
Joined 2014
Well, you're welcome to try it out with a 3.6V reg (I believe the LP5907 regulator that I used in this board has a version with this voltage) and see it it sounds better. However, the datasheet states that the maximum voltage for the charge pump is 3.6V, and pushing it close to the limit might damage the chip over time.

Since those Chinese manufacturers are so cheap, I won't mind giving away my first run of boards for free (I ordered 5, but they have the habit of sending over 10+ boards for whatever reason). If there's interest, I'll make a second run with better boards (ENIG, etc.) and Jean-Paul's suggestions. Those you will have to pay for as they're more expensive;).

So far you and Jean-Paul are lined up for the first run of boards - I'll post more details once everything's arrived and ready.
 
Disabled Account
Joined 2014
Boards are here! I'm itching to start building soon...

F9RunA7.jpg


8FZbhBf.jpg
 
Disabled Account
Joined 2014
wow, really nice and clean looking boards and the manufacture process and shipping from china seems to be blazing fast; I'm curious about your first impressions regarding the sound quality :)

I'm currently building it and should be able to report on the quality soon.

Please consider also the PCM5142A converters. You know, they include DSP, and add just a few pins (I2C/SPI, GPIO outputs for additional I2S outputs). Board like this with the 5142 part would make a nice XOver for a 2-way (or even more with the GPIO I2S usage) speakers.

DSP and programming? Not for me (yet);)
 
Disabled Account
Joined 2014
I used a jumper cable over the bridge so that I could try it out. I just couldn't wait...

The DAC doesn't sound as neutral as I expected, but the sound is as a result more smooth and coloured. Distortion in my RMAA test is a bit high, but the details and instrument seperation in music come out fine. Then again, some people prefer their sound to be a bit different than the studio master (think tube amps).

RMAA test (my ALC892 doesn't have enough dynamic range and SNR for an accurate reading, but it does show that my distortion is a bit high), with my onboard output as a comparison

Lbg5kv5.png


I'll post more listening impressions later - right now I'm busy enjoying the music:D

pGWm6vm.jpg
 
Disabled Account
Joined 2002
Not bad at all. As it breaks in it will sound different in a few days. IMO a DAC should be neutral as coloration is not HiFi but that is debatable... I don't know this DAC chip that well but it can be that it also changes its character to neutral sounding somewhat after breaking in. As experienced by many this is the case with ES9023 for example.

Please remeasure after 3 days to check the differences (if measurable).
 
Last edited:
Disabled Account
Joined 2014
In general, I too prefer neutral rather than coloured equipment. I like my music the way the artist intended;), and I plan to use this DAC for some mixing and editing as well.

Are there specific tracks I should play for breaking it in? Some people recommend sine waves, others recommend music, etc.
 
Disabled Account
Joined 2014
After burning it in for many hours and installing the diode bridge (that finally arrived), performance has not changed. I decided to do some tests to try to isolate the problem.

The first thing (which I didn't bother to say earlier) was that RMAA reported that there was "high distortion in the recorded signal" during the calibration process every time I performed a test. A screenshot is below:

D2eGWak.png


Naturally, the DAC was on 100% volume during the test and outputting 1.4VAC while the 1kHZ/-1db test tone was being played. Later, I played several 0db test tones of varying frequencies and noticed that the DAC outputted around 1.5-1.7VAC on the RCA jacks, depending on the actual frequency.

The datasheet for my soundcard (Realtek ALC892) states that the input voltage for the ADC is 1.5V. During 0db peaks during the test, it is entirely possible for the ADC to clip since the output voltage of my DAC can go up to 1.7V.

Maybe someone can elaborate on this? The problem here is that I don't have a more tolerant soundcard to measure this with...
 
Hi, maybe that it's because the soundcard input you mentioned, but as I said earlier, several memebers reported in the past nasty clipping on the pcm5102 when outputing full scale digital signal; this could be overcome only by feeding it with 3,6v instead of 3,3 for some extra headroom (the same was done to ES9023 as an upgrade for ES9022 ); i don't know if that's the problem here but just for the fun of the experiment you could also try to feed the dac using a 3,6v ldo (micrel ones are pretty cheap)
 
From the DAC spec sheet, full scale output is 2.1Vrms. At -1dB, the reading should be 1.67Vac which is what you measure.

The realtek input impedance is spec'd at 64kohms, while it's input voltage is spec'd at 1.5Vrms. So it does look like there is a chance of clipping unless there is something else in the connection (some other external load?).

Also, check the supply voltages with the device running. The DAC supply should read 3.3Vdc. Also, put the meter into AC and measure the supply voltage. Ideally, this reading should be close to the lower limit of the meter (unless you have a really good meter). If you read more than say 0.1Vac, then maybe the supply is not adequate.

Try lower amplitude test waveforms. -0.18dB waveforms should fit the Realtek input.

Also, can you check the Realtek card with another know DAC board? While you say the distortion numbers are high, it could be the Realtek.

Are you running the USB from a laptop on battery power just to check for ground loops?

I have not read the whole thread, so if this has already been mentioned, then sorry.
 
Disabled Account
Joined 2014
Just saying, but I never get full scale output of 2.1Vrms. Playing a 1kHz test tone at 0db, max volume only gets me around 1.7VAC on the RCA output.

I have no external load, I measure off the RCA jack directly.

Both the AVDD and DVDD supplies of the DAC report 3.3VDC, and 6.5VAC. If what you said is true, then the supply looks very inadequate. However, I'm running my device off a 700mA, 10VDC adapter, a 1.5A 7805, and four LP5907 regulators that supply 250mA at 3.3V. According to the chips' datasheets and their power consumption readings, my supply is in spec.

Lower amplitude test waveforms (I brought it down step by step until RMAA stopped saying it was clipping), give me around 0.003% distortion, but at the cost of dynamic range and noise floor measurements.

Yes, I've tested the Realtek with it's own DAC and the results are fine (see previous posts). However, I measured the Realtek's DAC outputs with my mutimeter and only found 0.9VAC on the outputs while playing the same 0db signal at max volume. Strange...

My DAC doesn't run off USB anymore; I've updated the design. The new schematic is here (note that it's an old revision and the RCA jacks were left out, but otherwise everything else should be the same), and everything is transformer-isolated to prevent ground loops.

What I really need is a scope, but I unfortunately don't have one. I do plan to implement a 3.6V regulator for the charge pump in my second revision, but for now I want to iron out my bugs first.
 
Last edited:
Well To me it seems like you are getting good results and the lack of dynamic range is due to the Realtek input limitation.

I would check the meter you are using. Is it true RMS or average reading? What is the make and model of the meter? What sort of test probes are you using?

You could use a resistor voltage divider to fit the DAC output range into the Realtek. The DAC cannot drive less than 1k. 4k and 10k resistor should scale 2.1vrms to a little less than 1.5vrms. (The Realtek input is supposed to be 65k)
 
Disabled Account
Joined 2014
This is my meter, and I'm pretty sure it's just average reading. However, I'm playing straight-up sine waves into the meter, so there shouldn't be a problem. I use the included probes and directly touch them to the board to measure the voltage, or in the case of the Realtek I touch it to the connected 3.5mm cable. However, I've measured the cable's resistance and it's too small to affect the output voltage.

I'll need to get parts for the resistor divider, and possibly a breadboard too.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.