Moonshadow XMOS DAC - No I2S Output

[snoopy33]

Hi,

Does anyone have one of these boards working?

XMOS数字界面板，CPLD波形重整技术-月影科技

They are popular on aliexpress.

I'm using Linux, so no drivers required, and I'm feeding the I2S lines into a ES9028Q2M board which I've verified is working correctly.

And... no output.

Did a quick test of the output pins and have:

Data 3.28v
BCK 1.64v
LRCK 1.64v
MCK 1.64v

Which appears to be correct. A bit stumped.

Aliexpress seller was helpful until the protection ran out, which I didn't catch, then unsurprisingly silent.

Regards,

Andrew

 

[percival007]

Not familiar with Linux, so first things first, are you sure your USB (?) source is 'talking' to the PCB?

Are they talking the same Language? i2S can have a few 'dialects' ! Left Justified etc etc.

Have you a 'scope? Can you actually see some Data and Clocks on the i2S Lines? Those are just DC Voltages you quote which could be there when powered even with no Signal.

Cheers,

P.

 

[abraxalito]

If those voltages are what you got from a DVM then the 'Data' voltage looks suspicious (too high), the others are fine (as they're clocks at 50% duty cycle).

 

[snoopy33]

Hi, I used a PCM2704 based converter to ensure the DAC (a ES9028Q2M) is working, so the issue is with the XMOS USB-I2S.

For the XMOS, it's recognised fine. I've tested on Linux and Windows with same result so don't think it's a driver issue.

I note the 1.64V lines are the same as the output of the PCM2704 (1.64V is probably the digital on/off averaged by the multimeter). Whether they're talking the same I2S "variant" I'm unsure, but I'd like to hope so.

I don't have a scope but I do have a data recorder good for 20MHz. If I2S lines are less than this I can try capturing. Worth a shot!

 

[snoopy33]

If those voltages are what you got from a DVM then the 'Data' voltage looks suspicious (too high), the others are fine (as they're clocks at 50% duty cycle).

Thanks Abraxalito, I agree it initially raised an eye-brow. I've checked and it's not shorted with the 3.3V line. I'll get the data recorder onto it first.

 

[snoopy33]

Results:

Data is 0V when not playing (my error earlier). Once playing I can see signal. Looks clean
BCK is a square wave signal with some roll on/off. Not the cleanest.
LRCK is a similar square wave but much cleaner than BCK.

So indeed looks like a compatibility issue.

 

[snoopy33]

Pictures

Pictures attached

 

[abraxalito]

Those three signal plots look reasonable for I2S running at 2.8MHz BCK (so 32 bits per channel).

 

[snoopy33]

Thanks.

I've tried WIndows 7, 10 and Linux now, same on all.

It looks to me the I2S signal is malformed in some way.

 

[canvas]

It should work unless DOA. I've seen people using this USB bridge and claimed having better performance than regular XMOS.

 

[rfbrw]

Are they talking the same Language? i2S can have a few 'dialects' ! Left Justified etc etc.

Statements like this simply serve to confuse the issue. There are no dialects. There is I2S replete with its own terminology and there are other serial audio interface formats.

 

[rfbrw]

It helps to look at the data using LRCK as trigger.

 

[percival007]

Statements like this simply serve to confuse the issue. There are no dialects. There is I2S replete with its own terminology and there are other serial audio interface formats.

I apologise.

So, how about, instead of just telling me I am wrong, this being an amateur Forum, mostly, why not pass on some of your knowledge and educate us on the different serial data formats.
What, for instance is precisely meant by Left Justified and if a DAC or Receiver IC is set to receive i2S, why will it not work when fed with Left Justified Data.

That might help the OP in finding out why they have no sound.

Cheers,

P.

 

[snoopy33]

OK - I believe it's installed with bad firmware or similar. The I2S mustn't be standard.

I had an email going with Moonshadow the Chinese developer but they've not been helpful.

 

[abraxalito]

Does your data recorder have two channels? The most likely thing I reckon is the phase of the word select (WS or LR) is wrong with respect to the datastream. This could lead to always loading in zeroes.

 

[snoopy33]

It does but that would raise an issue with the board wouldn't it, not something I can resolve?

 

[abraxalito]

If you can show its the board's fault rather than your DAC's wouldn't you want to know?

 

[snoopy33]

My DAC is working (tested with another USB 2 I2S), and many people use the ESS9026Q2M. It's got to be this xmos board.

 

[weissi]

I had a bad USB connector on my Moonshadow interface that produced the weirdest things. Maybe worth checking that also.

 

[snoopy33]

Can't be that, it's enumerated fine. But thanks.