Hi all! I found several shortcomings of this DAC. Maybe it will be useful for someone.
All modifications you make at your own peril and risk! The device is sensitive to electrostatic discharge! You need PRO skills and special equipment!
I decided to measure my DAC with my old Asus Xonar u7 and found that the DAC gives pretty bad results. There was also a terrible ripple at the type-c socket.
So, I dug deeper and found a more frightening ripple at the negative output of the DC-DC converter. It was 1Vpp and 71Hz.
Further, I found out that the
TPS65131 based DC-DC converter
does not work normally in power saving mode and generates a lot of low frequency noise. This is clearly a flaw in the chip. This can be corrected by adding an RC filter to the converter's negative feedback and turning off the power save mode.
[You need PRO soldering skills and a microscope to make this!] These mods fixed the ripple on the negative side of the DC converter, but fixing the positive side was not so easy.
[Careful! I burned the chip before I could.]
When I turned off the power saving mode on the positive side of the converter, it stopped oscillating and I had +5V at the output. When the positive output of the converter is loaded with beafy capacitors and the power saving mode is disabled, a voltage surge may occur and the OVP protection will turn off the positive side. What the hell?
So, I left the positive side in power saving mode. It still has a slight ripple, but is stable. The the power saving mode of negative side was turned off. Ceramic caps can also be added to the electrolytic to filter out HF noise.
How to disable power save mode for negative side of DC-DC converter (if it has high ripple): cut the trace near the via from 11th pin and connect 11th pin to ground. Add a 6.8pF cap and a 51K resistor to negative feedback.
Another thing that can be fixed is the USB connection. XMOS is controlled by an
ADM1085 sequencer chip. This chip keeps the XMOS off for a while until the supply voltage becomes stable. Replacing the timing capacitor with 100nF increases the XMOS startup delay time to 0.48s. A DAC with this fix starts normally with almost any power supply. It draws about 600mA, and any 1A charger or 1A power supply will suffice.
Earlier, reboots were necessary to successfully start the device.
My tests have shown that this DAC is sensitive to power and USB noise. I was unable to kill the USB noise. As mentioned earlier in this thread, insulating the
Ground and
+5V contacts of the USB-B cable with kapton tape helped reduce the noise. Only the USB data lines remain. (Perhaps it was a ground loop.)
Last way you can improove DACs performance is low noise linear power supply.
Interesting, that I haven't done a listening test yet ))). I'll add an update later. Also,
my measurements may not be correct or there may be some errors. This is my experience.
I also can’t understand where these spikes at 8KHz and 16KHz come from. They decrease when the music is paused and increase when XMOS wakes up and the music starts playing. They are a little visible in the tests of other devices. But that's another story.
P.S.: answer is
HERE,
HERE and
THERE.
Good luck!