I was going to use an exaU21 8 channel usb to i2s board, which by all accounts, was an excellent board. Not aware of any converters with more than 8 channels.
Not feasible for a studio setup. You need one master clock and a completely reliable setup of several sound cards using ASIO drivers.
The difference with clocking is really disconcerting. Ironically, it shouldn't matter for the RME card either, which also re-clocks and is supposed to have lower jitter with its internal clock. But the whole audio goodness coming from the DAM collapses once the external DCS converter is no longer the master...
Soeren, could you please chime in what must be going on here and if it can be fixed?
XMOS Multichannel high-quality USB to/from I2S/DSD SPDIF PCB - DIYINHK
And I just read it can even do 32 channels... and it would not surprise me if the master clock could be inserted... but indeed the drivers and overall reliability may be less good...
And I just read it can even do 32 channels... and it would not surprise me if the master clock could be inserted... but indeed the drivers and overall reliability may be less good...
Exa was excellent but sadly obsolete and unsupported today and diyinhk for some reason cannot even make a reasonably good sounding 2-channel board.
Master clock is really a concern for the ADC process, you can't directly mix tracks with different sample rate. But it doesn't matter much for the DAC process, as the output is analog with no clock embedded.... Of course, if you use DACs with different delay, you might get problems, so you should stick to the same DAC....
The soekris dam modules all sync their master clock to the incomming sample rate, with a resolution of 1 hz for the 44M/48M master clock. I am thinking about going to a higher resolution, maybe 1/64 hz (so it fit in 32 bit), starting with the dac2541 and then back ported to the dam modules.
Thanks Søren, I should have framed my question better:
I was wondering why the incoming S/PDIF signal's clocking (its amount and type of jitter) makes an audible difference here (since the DAM1021 does re-clocking) and why doubling the S/PDIF output voltage (to professional AES/EBU level) also makes a very audible difference.
And if something can be done to make the DAM1021 less or not susceptible to the incoming clocks idiosyncracies.
The resolution of 1hz is news to me, I read here somewhere that the clock was syncing at a rate of 400 hz...
I was wondering why the incoming S/PDIF signal's clocking (its amount and type of jitter) makes an audible difference here (since the DAM1021 does re-clocking) and why doubling the S/PDIF output voltage (to professional AES/EBU level) also makes a very audible difference.
And if something can be done to make the DAM1021 less or not susceptible to the incoming clocks idiosyncracies.
The resolution of 1hz is news to me, I read here somewhere that the clock was syncing at a rate of 400 hz...
Could be your source or cabling, test at various places showed no difference between the sources, in fact, many had SPDIF as the best.... Also you can connect the output of a RPi I2S output directly to the I2S input on a dam1021, and the RPi is really jittery, per design. The reclocking fifo works.
Check your receiver circuit and cabling.... I am planning to double the fifo size, but that should only make a difference with high jitter sources.
There are different theories floating around by people who don't know much.... The clock get updated at most once per second.
Early on I did describe the digital pll in details, but that was many postings ago....
Last edited:
Fedde reported the same thing (post #8977), also with a RME card.
I use the receiver circuit from the manual, transformer coupled.
How much jitter do you consider to be "high" wrt to the DAM1021?
Also, shouldn't it be either or? Either the FIFO reclocks perfectly no matter what you throw at it or it shouldn't? The RME card is specced to output jitter below the 1ns mark.
Thanks Søren.
Yes, but he also said "But going to a good USB to I2C/spdif board is... " Implying that maybe the RME card isn't that good....
One thing is the schematic, another thing is the actual implementation....
When the 1mS FIFO underflow/overflow, which will cause the dam1021 to loose sync.... Until then the FIFO will eat whatever you feed it.
Sorry, but I can't really address those subjective "different" sound statements, especially as plenty of other people get good results....
The RME card is, of course, a lot better than 1ms. RME gear has long been the de facto standard in professional studios for digital I/O. The transformer balanced digital output of the RME card and the transformer balanced DAM1021 input are connected by a 3m cable. There's not much room for things to go wrong here...
As for the implimentation of the input receiver, it is reallly simple and I wired it all point to point. And it is all transformer balanced symmetrical lines, again, not much room for things to go wrong there.
The LED of my DAM1021 syncs and then just stays on.
But since the FIFO is supposed to reclock everything perfectly, it all shouldn't matter as long as the signal arrives intact, which all my test recordings show it does indeed...
But I will try and see if I can move the pulse transformer even closer to the DAM1021 board.
What's the correct way to connect ground and chassis (enclosure) with the DAM1021, especially regarding balanced connections?
In my setup the XLR outputs at the enclosure have pin 2 and 3 wired, but pin 1 (shield) of the outputs is connected to chassis (but not to the DAM audio ground). Same for my own balanced analog output board. All boards are isolated from the chassis ground (literally) they stand on. The AES/EBU input is floating.
So the only chassis connection of the DAM1021 board is through the PSU ground.
Is this the recommended way to do it, or should there be a connection of audio ground directly to the chassis?
I don't have any hum or RF problems.
In my setup the XLR outputs at the enclosure have pin 2 and 3 wired, but pin 1 (shield) of the outputs is connected to chassis (but not to the DAM audio ground). Same for my own balanced analog output board. All boards are isolated from the chassis ground (literally) they stand on. The AES/EBU input is floating.
So the only chassis connection of the DAM1021 board is through the PSU ground.
Is this the recommended way to do it, or should there be a connection of audio ground directly to the chassis?
I don't have any hum or RF problems.
pin1/shield connected only to the chassis, grounding of the dam1021 through star ground that combines chassis/pin1/signal reference as explained in the article from David Davenport.
Audio Component Grounding and Interconnection
I made some experiments regarding grounding to chassis, even using fat speaker cables to have ultra low impedance.
The differences in detail/"3D-ness" of the signal are not subtle. Audio ground to pin1 directly to chassis worked better than everything else.
So this another very important area that DIYers may easily overlook.
Bottom line for me is, that the stock board sounds awesome if implimented correctly.
The differences in detail/"3D-ness" of the signal are not subtle. Audio ground to pin1 directly to chassis worked better than everything else.
So this another very important area that DIYers may easily overlook.
Bottom line for me is, that the stock board sounds awesome if implimented correctly.
Wouldn't the soundstage sound "nervous", if the idea of out-of-sync clocks makes you "nervous"? 😉
Last edited: