Observations between the Matrix SPDIF 2 and Douk (Nobsound) U2 Pro

I have this curiosity about i2s which I’ve played around with for a little while, having built an i2s streamer mentioned elsewhere using a Pi4 and FifoPi and HDMI Pi from Iancanada. Along with that, I also played with a Breakout box that takes HDMI and outputs various standards including PS compatible i2s.

So within the past couple weeks, I also got my hands on the Douk, then got a deal on a used Matrix. So here’s my observations so far.

Both have no problems outputting PCM properly.

The Douk has a constant signal on pin 15 that sets the pre-emphasis/de-emphasis flag, which only affects 16/44.1. This can be eliminated by pulling or covering pin 15 from the HDMI cable, or cutting the trace between pin 45 of the XMOS chip and pin 15 of the connector - It’s a direct connection between the pin with no resistor in the path, and the trace is easily accessible.

The Douk also reverses L+R channels with DSD. This is verified with a bunch of free test files I have that comes in various bit/sampling rates in both PCM and DSD, so you can easily here the channels reversing on DSD (it opens with a piano spanning both channels). It does fine with PCM/Flac.

The Matrix has no swapping problem on DSD, and does not set the pre-emphasis flag,

The Douk has no problem running off USB power (it has the ability to be externally powered).

The Matrix DOES draw significantly more current than the Douk. The first clue I noticed was that the USB B connector was 3.0, which has a 900ma standard compared to 2.0 which has a 500ma standard. Along with that, they spec that if using an external supply, it needs to be at least 6 volts and 800ma. So when it downconverts internally to 5v, it will likely be near the 900ma spec.

When I tried plugging it in to my Pi streamer, I got a low voltage warning from the OS. I have been running the Pi (with hats) through the Iancanada FifoPi rather than the USB C supply input because that’s been sufficient, and the FifoPi cleans the power somewhat. But the powering the 5v through the FifoPi limits the total 5v power to 2A. That may have been fine with older Pis that only had USB 2.0 ports, but the Pi4 has 2 USB 3.0 ports. I upped the output of my external 5v to 5.23 volts to account for the sag, and no more low voltage warning. I’m going to power the 5v directly through the GPIO and bypass the FifoPi, so that should take care of it - or maybe I’ll use a spare 12v external I have along with a buck converter to bring it down to 6 - maybe.

There’s no such problem with my music only Windows PC with “normal” USB 3.0 ports. Depending on your computer/streamer/whatever, you may want to be cognizant of the higher current needs (which exceed the normal 2.0 spec), or as with many users, just power it externally.

Edit: I haven’t gotten around to measuring the actual current draw of the Matrix yet. If anyone has tested the actual draw, please post the number (so I don’t have to dig my stuff out).

I played music via my laptop through the matrix and it was fine. I powered the matrix via the usb plug coming from my Sonore as well. No issues. I did end up getting a Small green computer LPS for it, was not that expensive so I added just in case and to avoid any issues. I have only had issues once where roon behaved weird and I had to reboot the Sonore and Matrix in order to clear it up. Software, and firmware can suck. But other than that one time its been performing great.

I could be wrong, but I believe the Matrix has linear power regulators and uses one regulator for USB, but two linear in series for an external supply. Anyway, the Matrix is speced for USB 2.0 so it can’t really get more than 500mA at about 5V over USB and, because it’s a linear supply, wouldn’t draw less current from a higher voltage source, just more power.

I ran my Matrix over USB 2.0 for quite and while, now it’s connected via a USB 3.0 hub.

1 Like

It should be reasonable on a full desktop or laptop in general. Just putting it out there that if plugged into a USB 2.0 port (spec’d at 500ma max), it may exceed the capability of the port (depending on the computer and how the ports are set up). It can potentially cause performance issues because of too much voltage drop (even if it seems like it’s working properly), or possible damage the USB controller of the computer because of excessive amperage draw. Again, depending on the computer.

For something as small as the Pi, you definitely need to be cognizant, since a Pi is limited to 1200ma through all 4 ports COMBINED - and that’s with adequate supply amperage. So if the Matrix draws the full 900ma standard for 3.0, it would only leave 300ma for the remaining 3 combined. I haven’t had the time to measure the actual current draw yet. So if using a Pi, one might definitely want to run an external supply.

Edited for bad math

It LOOKS like USB does some filtration but no regulation, and the power port has one regulator.

Could it be their ‘spec’ of 2.0 may just be referring to data? On the Pi, I have run things that draw the full 500ma off the 2.0 port with no voltage sag.

The Matrix explicitly states 500mA @ 5V for USB. The Matrix on USB 2 is not a problem here, but I know that many USB 2.0 hubs didn’t really provide 500mA for all outputs concurrently, so before I got a USB 2.0 hub that supplied the speced 500mA on all outputs in parallel I used to have problems with some external USB hard drives. With a good hub I never had those problems.


I have quite a different question: Is there a Matrix or similar product that can change the I2S signal from HDMI of the SACD transport into USB or Ethernet ?
Thanks in advance !

Ted, What USB 3.0 hub are you using?

I have not seen one. But there are HDMI Audio extractors that remove the SACD and put in spdif (DOP - DSD over PCM) that can then be coax into a DAC. But not HDMI to I2S. They are less than 100 on ebay. If you want a high end one, DBob or something like that its called.

I got this type of hdmi-i2s extractor (it does various outputs).

It’s sold under various brands, and you can even by the bare board (no enclosure) for less as well. Like the Douk, though, it does set the pre-emphasis flag, so you’ll have to disable pin 15 on the cable, or unlike the Douk, you can desolder one resistor in the path (the Douk would require cutting the trace, as it has a direct trace between pin 15 of the output connector and pin 45 of the XMOS chip).

Edit: Like the Douk, it also reverses L+R with DSD. Unlike the Douk, it has one measly 24Mhz clock (which may not make much a difference because of the DS’ conversion/upsampling/reclocking.

I forgot to mention, I got the Douk with the “stock” clocks. I didn’t want to pay the upcharge for the upgraded clocks because I had spare REAL CCHD957 oscillators that I never used from Digikey. Plus, with so many counterfeit oscillators running around in China, I couldn’t be sure what would end up in the circuit.

The stock clocks (3) were mixed brands. The main 22 and 24 clocks were SCTF and HHKJ, which are OK. When I replaced them with the Crysteks, I could not really discern a difference, likely because of two factors:

  1. The DS upconverts/converts/reclocks everything anyway.
  2. My ears are too old (which likely saves me a lot of money these days).

I’ve had good luck with Plugable products. At the moment I’m using their 10 port USB 3.0 hub and their 10 meter USB 3.0 active extension cable.
Plugable USB 3.0 10-Port Hub with 48W Power Adapter – Plugable Technologies
Plugable USB 3.0 10M (32ft) Extension Cable with Power Adapter and Bac – Plugable Technologies

They work well with five portable USB 3.0 disk drives, my tools and my DS… There are a lot of hubs that didn’t.


Rock on! Thanks!

1 Like

I stand corrected. when I played I was only interested in the Coax, the one I have indeed does have I2S. I am going to have to remember that one. Thanks.

Update for now. I haven’t had the time to dig too far yet, but here is the result of modest tests, monitoring the 5v pins on the Pi. The Pis is an easy test board because it’s tiny, has tiny supply needs, so is sensitive to how it’s loaded. Plus you can monitor the voltage of the board itself via its GPIO pins (2 or 4 and ground).

I used a certified 2.0 cable in the 2.0 ports for both to keep things as even as possible.

The following observation are only rough averages as even when a computer is static, there’s stuff that goes on that draws power.

Static voltage drop with the Douk was about 20-30mv
Static voltage drop with the Matrix was about 50-60 mv

For both, it SEEMS like processing DSD taxed the circuit the most:
Douk playing 256 DSD caused a voltage drop of an additional 50mv (over static)
Matrix playing the same file caused a voltage drop of an additional 110/115mv (over static)

So total voltage drop caused by Douk = about 70ish mv
Total voltage drop caused by Matrix = bout 150ish mv

I was going to try my cheap usb power meter (that has 3.0) ports. It can actually show the voltage AND amps passing through to the device, but for some reason I can’t get it to work with the Matrix, so will have to play with it, or possibly get a better one. I’ve only used it in the past to check if a port was working properly - not measure current. It’s not that it didn’t display current, it does, I just never paid attention to it because that was never something I was interested in for my past testing. I can’t get it to pass the DSD file to the Matrix (but it does to the Douk).

At least it give me an excuse (to my wife) for getting a better tester! More toys are always a good thing!

Don’t forget to disable pin 15, or it will set the pre-emphasis flag. It only affects 16/44.1 files though (according to Master Ted). Either cut the trace to the port, desolder the inline resistor (I THINK it was R47), or simply use a piece of tape and cover pin 15 of one side of your cable. I find it pretty difficult to reliably pull the entire pin from a cable, and I think it’s harder to cover the contact of the female port.

For WIW USB 2.0 allows voltage to fall to 4.75V at the device and be in spec. I need to get a good inline USB voltage/current meter doohickey. Thanks for the info.

1 Like

For which solution above? The matrix? I have not heard of this before.

If you are using either the Chinese HDMI breakout box for i2s, or the Douk. The Matrix does not set the pre-emphasis flag. Though to be honest, I would have to dig through my ripped collection to see if it is NOT setting the flag when it should. I don’t know off hand which, if any, of my ripped files are SUPPOSED to have the pre-emphasis flag since many CDs were incorrectly coded (both with and without).

No way I am going to be checking each disc for that. Glad I have the matrix and did not use the cheapo extractor.