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.
Disabled Account
Joined 2014
Ok, just did a test with a headphone volume control attached inline and turned it down until RMAA reported that there was no clipping. The voltage after the pot was 1.4VAC. Distortion has been reduced, but take the results below with a grain of salt - the volume control is after all a cheap and crappy piece of equipment.

U8iAgMW.png


Honestly, (while I'm preparing version 2 of the board) I'm also wondering if there are any layout/schematic issues in my design that are hindering performance.

And BTW I do have boards avaliable, but you may want to wait until Version 2 first.
 
Using the volume control I think you showed that full scale output is not clipping. Checking the audio spectrum might provide some more clues.

The multimeter you have looks to be a solid meter for home and auto measurements, but audio signal measurements might be a little off. The most sensitive scale for AC is 250Vac. I doubt that last decimal point is guaranteed to be that accurate. Especially for non 60Hz measurements.

Some other tests to check the accuracy of the Realtek measurements would be to measure the noise with the line inputs shorted together. Then measure noise from the DAC playing silence (the DAC is supposed to have auto-mute). Then measure noise playing the smallest signal possible.

Trying to improve the circuit without being sure of the measurements would be difficult.

If you really want a version 2, maybe tackle the problem you already encountered: not all devices are aligned to a 2.1vrms = 0dB line level.

If you can provide a scalable output, that could be a useful feature. If you add some sort of peak detector or rms to DC converter (LTC1968 for example), then you could use the more sensitive DC ranges of your meter to match the output to whatever input you connect to.
 
Disabled Account
Joined 2014
To be honest, my second revision will probably include a 3.6V regulator for the charge pump, better part placement, and a better 3.5mm jack (the current one requires me to route the signals in loops, which may be affecting my sound). I also want to remove the trace running under the PCM5102A and create a ground plane on the top layer underneath both chips.

I'm looking at recording interfaces right now, which have much higher input ranges as well as lower THD and noise measurement ranges than my Realtek. When they go on sale, I'll pick one up and use it to do some proper measurements.

To be honest, I don't think the measured noise will be much lower with the DAC on mute. I'm already pushing the Realtek ADC (which by the way is only 16-bit) to its limits already.
 
Might I make a suggestion? I believe the clipping at full scale is caused by gain error in the line driver. (shown in the datasheet, it's one of the joys of silicon resistors...).

If you are relaying out a board, the PCM5122, even though it's mainly an I2C part can also do HW control with more GPIO pins. One of the pins can do an analog gain switch from 2.1VRMS to ~1VRMS. This *should* maintain the SNR, whilst reducing the output from getting near the power supply rails.

Good luck,

Rochey
 
Disabled Account
Joined 2014
Might I make a suggestion? I believe the clipping at full scale is caused by gain error in the line driver. (shown in the datasheet, it's one of the joys of silicon resistors...).

If you are relaying out a board, the PCM5122, even though it's mainly an I2C part can also do HW control with more GPIO pins. One of the pins can do an analog gain switch from 2.1VRMS to ~1VRMS. This *should* maintain the SNR, whilst reducing the output from getting near the power supply rails.

Good luck,

Rochey

Would using a 3.6V regulator on the charge pump provide the same effect? I feel that I'm kind of wasting the PCM5122 by using it like a hardware-controlled device...
 
Disabled Account
Joined 2014
BTW, this project is not dead (I've gotten messages & PMs asking me about this). While I've been busy with schoolwork, studying for exams, and such, I'm still thinking of making a V2 board and am laying out the PCB in my spare time.

Basically, the board will include more effective routing and part placement (I've been looking at many DAC designs as well as the EVM boards), ground planes under the chips, support for film caps on the output, a 3.6V regulator for the charge pump, and possible headers for people who don't plan to use the board-mounted 3.5mm and RCA jacks.

Possible ideas on the horizon (which I might implement) include the use of the better performing CS4351 DAC chip instead of the PCM5102A as I now have a linear power supply and can supply the 9-12V required for the chip. I'm also thinking of using a remote (relay) switch using a pushbutton in the front rather than having the switch on the side as it currently is.

I might as well say this now, but I've been listening to my DAC for quite a while and probably has been fully "broken in", if that makes a difference or not. When compared to my other equipment, SQ is rather nice but not completely neutral, with a slightly warmer sound. Clipping is not evident during normal listening, and is barely noticeable at 0db points. I have some more listening impressions here.

Stay tuned.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.