Douk Audio U2Pro Digital Interface Upgrade is Being Planned and Waiting for Your Advice!

[Sharppain]

Did you try changing the UAC1.0/UAC2.0 switch on the bottom? I think you need UAC2.0 for 192kHz.

Absolutely!!! I did the UAC2.0 since I stream at 192khz minimum and wanted to use it also for the comp where I do HQPlayer but it failed in both sources.
Added several hours later: Already at home, I found my old U2 Pro, with the old XMOS chip (think 208) but with upgraded clocks (three TCXO) and it has power imput of 5V; I installed it using same I2S and Supra USB cable and (!) external iFi iPower of 5V. OooMG, all working perfectly and the I2S input is making difference.
So, it pitty for the new U2 Pro but it is bad power design IMHO.

 

[Alan O]

So, it pitty for the new U2 Pro but it is bad power design IMHO.

I did think it was a shame they dropped the option for external power in the new one, yes.

 

[audio_tony]

I've just bought a U2 PRO and on checking the clock it's spot on at 48kHz sample rate but off at 44.1kHz (I see 44128Hz)

Timestamp,Card0,Card1,Card2
19:53:57,44129,44100,44100
19:53:58,44128,44101,44100
19:53:59,44128,44101,44100
19:54:00,44128,44100,44100

Card0 is the U2 pro
Card1 is a Cambridge Audio 851D
Card2 is a Topping D10

I didn't expect to see the clock this far off but my understanding is that to get 44.1 kHz from a 24 MHz source, the XMOS chip has to use a Fractional-N Synthesizer or a Phase-Locked Loop (PLL). It basically has to perform complex "on-the-fly" math to approximate the 44.1k clock because 24 doesn't divide into 44.1 evenly.

Hence the error at 44.1k.

Am I right to assume that this is expected?

EDIT:

I tested this theory by recording one channel from the DAC connected to the U2 PRO and a second channel from the Cambridge.

There is an audible phase shift due to the mismatched clock timing at 44.1kHz.

I also recorded two sample files - one via the U2 PRO and another via a C Media DDC - I don't need to tell you which is which...

You can download the files here:

WARNING: these are LOUD 1kHz tones - ensure your volume is right down before listening.

https://dnsbits.net/downloads/tmpwav.wav

https://dnsbits.net/downloads/tmpwav2.wav

WARNING: these are LOUD 1kHz tones - ensure your volume is right down before listening.

TLDR; the U2 PRO is only suitable for playing at 48, 96, 192, 384kHz if you value pitch accuracy.

It's not ideal for 44.1kHz (which a lot of material is?) as it will play all 44.1kHz material 0.06% fast, which represents a pitch shift of about 1.1 cents.

 

[Roland68]

Absolutely!!! I did the UAC2.0 since I stream at 192khz minimum and wanted to use it also for the comp where I do HQPlayer but it failed in both sources.
Added several hours later: Already at home, I found my old U2 Pro, with the old XMOS chip (think 208) but with upgraded clocks (three TCXO) and it has power imput of 5V; I installed it using same I2S and Supra USB cable and (!) external iFi iPower of 5V. OooMG, all working perfectly and the I2S input is making difference.
So, it pitty for the new U2 Pro but it is bad power design IMHO.

I did think it was a shame they dropped the option for external power in the new one, yes.

You can use any external power supply by splitting the USB cable: Data (+/-) and the negative power line from the computer, and the positive and negative power lines from the external power supply. This applies to all USB DACs as well.

 

[Roland68]

I've just bought a U2 PRO and on checking the clock it's spot on at 48kHz sample rate but off at 44.1kHz (I see 44128Hz)

Timestamp,Card0,Card1,Card2
19:53:57,44129,44100,44100
19:53:58,44128,44101,44100
19:53:59,44128,44101,44100
19:54:00,44128,44100,44100

Card0 is the U2 pro
Card1 is a Cambridge Audio 851D
Card2 is a Topping D10

I didn't expect to see the clock this far off but my understanding is that to get 44.1 kHz from a 24 MHz source, the XMOS chip has to use a Fractional-N Synthesizer or a Phase-Locked Loop (PLL). It basically has to perform complex "on-the-fly" math to approximate the 44.1k clock because 24 doesn't divide into 44.1 evenly.

Hence the error at 44.1k.

Am I right to assume that this is expected?

EDIT:

I tested this theory by recording one channel from the DAC connected to the U2 PRO and a second channel from the Cambridge.

There is an audible phase shift due to the mismatched clock timing at 44.1kHz.

I also recorded two sample files - one via the U2 PRO and another via a C Media DDC - I don't need to tell you which is which...

You can download the files here:

WARNING: these are LOUD 1kHz tones - ensure your volume is right down before listening.

https://dnsbits.net/downloads/tmpwav.wav

https://dnsbits.net/downloads/tmpwav2.wav

WARNING: these are LOUD 1kHz tones - ensure your volume is right down before listening.

TLDR; the U2 PRO is only suitable for playing at 48, 96, 192, 384kHz if you value pitch accuracy.

It's not ideal for 44.1kHz (which a lot of material is?) as it will play all 44.1kHz material 0.06% fast, which represents a pitch shift of about 1.1 cents.

Thanks for the info; unfortunately, that makes the device somewhat superfluous.
I had actually wondered right from the start how they managed to pull that off with just a single 24 MHz crystal.
Thanks to you, we now know the answer: they didn't manage it.

 

[L-Train]

I've just bought a U2 PRO and on checking the clock it's spot on at 48kHz sample rate but off at 44.1kHz (I see 44128Hz)

Timestamp,Card0,Card1,Card2
19:53:57,44129,44100,44100
19:53:58,44128,44101,44100
19:53:59,44128,44101,44100
19:54:00,44128,44100,44100

Card0 is the U2 pro
Card1 is a Cambridge Audio 851D
Card2 is a Topping D10

I didn't expect to see the clock this far off but my understanding is that to get 44.1 kHz from a 24 MHz source, the XMOS chip has to use a Fractional-N Synthesizer or a Phase-Locked Loop (PLL). It basically has to perform complex "on-the-fly" math to approximate the 44.1k clock because 24 doesn't divide into 44.1 evenly.

Hence the error at 44.1k.

Am I right to assume that this is expected?

EDIT:

I tested this theory by recording one channel from the DAC connected to the U2 PRO and a second channel from the Cambridge.

There is an audible phase shift due to the mismatched clock timing at 44.1kHz.

I also recorded two sample files - one via the U2 PRO and another via a C Media DDC - I don't need to tell you which is which...

You can download the files here:

WARNING: these are LOUD 1kHz tones - ensure your volume is right down before listening.

https://dnsbits.net/downloads/tmpwav.wav

https://dnsbits.net/downloads/tmpwav2.wav

WARNING: these are LOUD 1kHz tones - ensure your volume is right down before listening.

TLDR; the U2 PRO is only suitable for playing at 48, 96, 192, 384kHz if you value pitch accuracy.

It's not ideal for 44.1kHz (which a lot of material is?) as it will play all 44.1kHz material 0.06% fast, which represents a pitch shift of about 1.1 cents.

Wow, thanks for pointing this out. I had some concerns earlier in the thread about jitter from the PLL but didn't think that the 44.1 kHz frequency accuracy would be out by that much. I probed the I2S output's LRCLK using my Rigol DS1054Z oscilloscope using its hardware frequency counter and saw exactly what you observed as well:

44.1 kHz:

48 kHz:

The point I probed was after resistor R15:

Spoiler

I hope Douk Audio can look into this and improve the 44.1 kHz LRCLK with a firmware update.

 

[audio_tony]

I hope Douk Audio can look into this and improve the 44.1 kHz LRCLK with a firmware update.

I'm not familiar with what the firmware is capable of in terms of clock division etc. (if that's even possible) however the limiting factor is the use of a single 24MHz crystal.

This is why other DDCs have three separate crystals.

A 12 or 24 MHz crystal just for the USB data handling.
A 24.576 MHz crystal for 48k audio.
A 22.5792 MHz crystal for 44.1k audio.

44,100Hz * 512 = 22,579,200Hz (22.5792 MHz)
48,000Hz * 512 = 24,576,000Hz (24.5760 MHz)

The U2 PRO has the single 24MHz crystal as my understanding is that the focus is on USB performance - they are using a multiplier (or divisor in this case I think) of 500 vs 128, 256 or 512 which is the industry standard.

The clock for 44.1kHz is likely calculated internally by some maths routine and it is this that is responsible for the error at 44.1kHz.

Therefore with a 24MHz clock;

24,000,000 / 44,100 = 544.217687074 ...... (not the 500 they are using hence the timing error).

24,000,000 / 48,000 = 500