I thought that OSX simply transcoded to linear PCM for the device.. thus whatever you play the OS simply transcodes it to LPCM. The only difference is the sample rate and sample bit depth sent. The filters are then executed on the LPCM. This 'dumb device' only taking LPCM is because there's no codec within the xmos except they provide the Cirrus support for the evaluation board. My understanding is that the Windows drivers are because windows itself doesn't support audio class 2.0 and the transcoding for the higher sample rates/depths.
It sounds like the DAC is simply applying FIR1, IIR, FIR2 to the LPCM stream, using a zero insertion upsampling for any lower speeds - first FIR1 upsamples to the 352/384MHz, second FIR2 upsamples to the final 3MHz for switching.
I suspect that the 32bit stream is simply accepted and rendered down (as the system is 27bit anyway).
So when people speak of NOS as a buzzword.. in this case given the system is always upsampling, the meaning here is really switch off any sound effecting filter parameters and IIR CD filter.. as the system still needs to upsample to it's switching speed which technically is oversampling.
It sounds like the DAC is simply applying FIR1, IIR, FIR2 to the LPCM stream, using a zero insertion upsampling for any lower speeds - first FIR1 upsamples to the 352/384MHz, second FIR2 upsamples to the final 3MHz for switching.
I suspect that the 32bit stream is simply accepted and rendered down (as the system is 27bit anyway).
So when people speak of NOS as a buzzword.. in this case given the system is always upsampling, the meaning here is really switch off any sound effecting filter parameters and IIR CD filter.. as the system still needs to upsample to it's switching speed which technically is oversampling.
Last edited:
Applying just the "filters" is not everything that happens.
Usually you also see an attenuation of the signal prior to the filters to avoid clipping.
Since these onDAC filters are realtime filters, the filter designer needs to add
quite a (fixed) safety margin (?dB :: -1/-2/-3dB) for each attenuation/filter stage.
That means the signal (without confirmation) sees potentially 3 times digital attenuation (2x filter, 1x volume control)
and two times filtering and that on the DAC only. I assume the volume control comes as final DSP stage.
And all that is done @32bit. From my experience doing DSP based convolution/crossovers/resampling/...
I always experienced advantages of using 64bit DSP in ""multi-stage"" DSP chains btw.
The 32bits used on the DAC "could" IMO therefore also be a slight limitation.
Talking about NOS in the context of this DAC would be of course more then misleading.
Yep. NOS matches R2R.
Both expressions suggest an old-school, analoglike still digital experience.
But from what I understand it's just the first FIR configured with 1.0 coefficients what the custom-filter
folks call NOS. Which of course suggests ...
As a matter of fact. The Soekris DAC is a hightech digital device. As with every device,
what matters is the final result. And that seems to be very competitive.
And not to forget, there is no solution out there that comes without compromises.
To lower/limit these compromises DIY-Audio has actually been invented. (allegedly)
Enjoy.
Usually you also see an attenuation of the signal prior to the filters to avoid clipping.
Since these onDAC filters are realtime filters, the filter designer needs to add
quite a (fixed) safety margin (?dB :: -1/-2/-3dB) for each attenuation/filter stage.
That means the signal (without confirmation) sees potentially 3 times digital attenuation (2x filter, 1x volume control)
and two times filtering and that on the DAC only. I assume the volume control comes as final DSP stage.
And all that is done @32bit. From my experience doing DSP based convolution/crossovers/resampling/...
I always experienced advantages of using 64bit DSP in ""multi-stage"" DSP chains btw.
The 32bits used on the DAC "could" IMO therefore also be a slight limitation.
Talking about NOS in the context of this DAC would be of course more then misleading.
Yep. NOS matches R2R.
Both expressions suggest an old-school, analoglike still digital experience. But from what I understand it's just the first FIR configured with 1.0 coefficients what the custom-filter
folks call NOS. Which of course suggests ...
As a matter of fact. The Soekris DAC is a hightech digital device. As with every device,
what matters is the final result. And that seems to be very competitive.
And not to forget, there is no solution out there that comes without compromises.
To lower/limit these compromises DIY-Audio has actually been invented. (allegedly)
Enjoy.
Last edited:
Quick question - the board has a second i2s connection (essential for future proofing imoho) but I have a couple of questions:
* can it be selected through the source select in uManager?
* I assume it’s not isolated?
* does the board start up muting the inputs (the output buffer would mute output)
* is it possible to shutdown the xmox when using the internal i2s?
* can it be selected through the source select in uManager?
* I assume it’s not isolated?
* does the board start up muting the inputs (the output buffer would mute output)
* is it possible to shutdown the xmox when using the internal i2s?
Yes. See page 8 in manual.
Correct.
Output would be muted during startup. The EN_PHO and EN_LIN should be used to enable the outputs (non mute).
No. Just don't connect USB Power if you want it shutdown.
Hmm, that's right, there should be no capacitor at C260. but it should be at C263.... The stock I have here is correct, so probably just a single mistake that got though.... I assume you can move it yourself ?
You face somewhat of a problem mixing 2 different DACs to play in one and the same system when one is USB (pleisyncron) and s/pdif which is syncron. It could men that the timing between the two DACs may change between songs and/or albums. How much... Dont know - there is no fix delay promised over USB.
I run a similar system now using 2pc of 1121 and a Hypex DCLP - all on opto using a self built true parallel opto multiplier. While this is not the same as hawing one-chip 4 channel DAC, I'm not to worried. Mixing USB and s/spdif... I wouldn't.
//
I run a similar system now using 2pc of 1121 and a Hypex DCLP - all on opto using a self built true parallel opto multiplier. While this is not the same as hawing one-chip 4 channel DAC, I'm not to worried. Mixing USB and s/spdif... I wouldn't.
//
I think this is just a small part of the problem. Even if Blitz uses an external USB/i2s/spdif board (not a great idea), the fifo will further delay the output of the DAM.
Perhaps it just cannot be done without a serious compromise.
Unless the subwoofer gets a little upgrade with another DAM
I try to setup the dam1941 with a streaming device over the 2nd I2S port directly which can give me high quality playback of my NaS files from 44k/16bit to 192k/24 bit AND spotify/tindal support.
My choice has been a RP4, Ian's fifopi ultimate and volumio.
Well, i learned, that the DAM1941 was able to play the 24bit recording, but no 16bit input is accepted / understood. Is this always the case ? No 16bit on I2S ?
Unfortunately Volumio cannot resample spotify, so i did not got what i hooed to realized so far.
Has anyone came to same issue and how did you solve it ? Is there a potential for a firmware update to enable as well 16bit ?
My choice has been a RP4, Ian's fifopi ultimate and volumio.
Well, i learned, that the DAM1941 was able to play the 24bit recording, but no 16bit input is accepted / understood. Is this always the case ? No 16bit on I2S ?
Unfortunately Volumio cannot resample spotify, so i did not got what i hooed to realized so far.
Has anyone came to same issue and how did you solve it ? Is there a potential for a firmware update to enable as well 16bit ?
All dams only take 32 bit I2S signals, you should be able to set your player software to always output 32 bit.