MSB Diamond DAC IV Reviewed By Stereophile

[asiufy]

There's no mystery. I've heard this directly from Larry @ MSB. He said there's no conversion. It just decodes DSD via software.
Since MSB doesn't use off the shelf parts, they're able to program the DAC to do whatever they want, and since the DAC runs at a fast enough speed, they were able to program it to decode DSD, via software, in 2 days.
Since dCS also has programmable DAC modules, they're also able to add DSD to their DACs via software. dCS' limitation is actually in hardware, in which they rely on specific ports for receiving the signal (USB, Firewire or dual AES), while MSB can handle up to 2xDSD on any port.
It's commendable what some people can do with off the shelf DAC chips. But, in my experience, rolling your own DAC modules give you a lot more flexibility, not to mention much better SQ overall. Heck, you can even custom order specific filters for your DAC IV, since they're also all software-based.


alexandre

 

[microstrip]

(...) Converting DSD to PCM brings a significant loss, no matter where the conversion is done (software in the server, DAC, etc.)

Even to DXD 32 bit, 352.8KHz? I think I remember Bruce referring that it was a negligible loss.

BTW, did any one listen to the Luxman's DSD-capable DA-06 DAC) (thanks DaveyF for pointing me the enjoyable article on the Philip O'Hanlon of 'On a Higher Note' listening session in Stereophile. A disturbing article ...

http://www.stereophile.com/content/luxman-debuts-its-dsd-dac-music-lovers

 

[asiufy]

Didn't try that, only 24/192. DSD2x beats that handily. Don't know if Korg's software is able to transcode DSD to DXD, will try, though, as my DAC can play both.



alexandre

 

[CKKeung]

Even to DXD 32 bit, 352.8KHz? I think I remember Bruce referring that it was a negligible loss.

BTW, did any one listen to the Luxman's DSD-capable DA-06 DAC) (thanks DaveyF for pointing me the enjoyable article on the Philip O'Hanlon of 'On a Higher Note' listening session in Stereophile. A disturbing article ...

http://www.stereophile.com/content/luxman-debuts-its-dsd-dac-music-lovers

There are so many new mid-price DSD-capable DAC coming out recently : Luxman DA-06, Auralic Vega, Invicta, exaSound e20, Meitner MA-1, Chord QuteHD, Matrix Audio X-Sabre, Lumin ...etc
Hope that the number of DSD music files available for sale can catch up.

 

[Orb]

Even to DXD 32 bit, 352.8KHz? I think I remember Bruce referring that it was a negligible loss.

BTW, did any one listen to the Luxman's DSD-capable DA-06 DAC) (thanks DaveyF for pointing me the enjoyable article on the Philip O'Hanlon of 'On a Higher Note' listening session in Stereophile. A disturbing article ...

http://www.stereophile.com/content/luxman-debuts-its-dsd-dac-music-lovers

Also worth noting 2L studio setup is DXD and some still prefer their DSD releases (which must be transcoded from DXD in the workstation), suggesting then that DXD is transparent and without any loss of information.

TBH though I am doubtful there is enough processing/resources power in a domestic DAC to transcode on the fly, even the latest 4-box dCS solution Paul Miller recommends not to transcode/upsample.
So realistically I would think it is only possible with a DAW/studio setup, and interestingly I do wonder with modern PCs with the right software and setup (alternative solution to doing it in the DAC).

Cheers
Orb

 

[microstrip]

(...) TBH though I am doubtful there is enough processing/resources power in a domestic DAC to transcode on the fly, even the latest 4-box dCS solution Paul Miller recommends not to transcode/upsample.
So realistically I would think it is only possible with a DAW/studio setup, and interestingly I do wonder with modern PCs with the right software and setup (alternative solution to doing it in the DAC).

Cheers
Orb

Orb,

After reading the documentation about the Vivaldi system I also could not stop thinking about it. Do we really need to own a very expensive on the fly separate upsampler? Why can't we have the upsampling done off-line in a PC to a DXD 352/32 file and then the file played directly through the Vivaldi DAC? It the DXD transfer over 2 AES cables proprietary of DCS?

 

[Orb]

Yep totally agree Micro,
comes down to how good the solutions/musicplayer are for consumers and their "lite" version desktop/laptop they are using with the DAC - messy environment to upsample/transcode in so they will not all be equal even ignoring their different approaches to the algorithms/filtering-noise-shaping/etc IMO.

But I do wonder if CUDA orientated high performance GPUs can or will eventually be involved in such solutions, or even needed.
Cheers
Orb

 

[opus111]

There's no mystery. I've heard this directly from Larry @ MSB. He said there's no conversion. It just decodes DSD via software.

Software sounds to me like conversion. If there was dedicated hardware then the software would be redundant no?

Since MSB doesn't use off the shelf parts, they're able to program the DAC to do whatever they want, and since the DAC runs at a fast enough speed, they were able to program it to decode DSD, via software, in 2 days.

If their DAC can run at a 2.8MHz symbol rate (that's 64Xfs for RBCD) that's some mean multibit converter. Why wouldn't they offer 64X upsampling for RBCD if this was the case? I'm myself dubious it would offer any benefits but I'm pretty certain there's a market for it. Then having their DAC running at such a lick, they'd also need some kind of hardware filtering to remove the OOB hash which wouldn't be needed with PCM.

The story does not stack up from a technical perspective.

Since dCS also has programmable DAC modules, they're also able to add DSD to their DACs via software.

And again, software would be used to convert from DSD to some kind of low-bit format suited to their 5bit wide DAC. Just as DSD is converted to 'DSD-wide' in some workstations - though 'wide' is actually 8bits.

It's commendable what some people can do with off the shelf DAC chips. But, in my experience, rolling your own DAC modules give you a lot more flexibility, not to mention much better SQ overall.

I've not encountered evidence myself that rolling your own DAC chips beats the best integrated ones, but then I've yet to hear MSB's solution. I also don't come across compelling technical stories for why doing it yourself is better but more than happy to be pointed to some.

 

[Bruce B]

Correct me if I'm wrong, but the MSB transport will only output DSD using the proprietary DSD link. Sending it out any other way (short of HDMI) would be illlegal.....

There are several units that output SACD's DSD natively... Playback Designs, EMM Labs and dCS

 

[edorr]

There are several units that output SACD's DSD natively... Playback Designs, EMM Labs and dCS

All encrypted though....

 

[opus111]

I'd guess dCS can do DSD natively because their 5bit DAC already is running at the right clock speed (2.8MHz for 44k1). So they just need to downgrade to using just two out of their 32 available levels.

 

[Bruce B]

All encrypted though....

Nope.... just connect it to a Sonoma and hit record or go from a format converter to SDIF-3 and record with a Tascam DV-RA1000HD

 

[asiufy]

Software sounds to me like conversion. If there was dedicated hardware then the software would be redundant no?

Of course it's hardware+software, opus.
What I meant to say is that they only needed a software upgrade in order to be able to "decode" DSD.

If their DAC can run at a 2.8MHz symbol rate (that's 64Xfs for RBCD) that's some mean multibit converter. Why wouldn't they offer 64X upsampling for RBCD if this was the case? I'm myself dubious it would offer any benefits but I'm pretty certain there's a market for it. Then having their DAC running at such a lick, they'd also need some kind of hardware filtering to remove the OOB hash which wouldn't be needed with PCM.

The story does not stack up from a technical perspective.

They don't do 64x, but 32x. Read up on their programmable software filters, please.
If it's a benefit it's up for discussion, but there's a considerable sonic difference between the 16x filter and the newer 32x one.

I've not encountered evidence myself that rolling your own DAC chips beats the best integrated ones, but then I've yet to hear MSB's solution. I also don't come across compelling technical stories for why doing it yourself is better but more than happy to be pointed to some.

Well, my evidence is just the contrary. dCS and MSB (and Playback) are so far ahead of the other "off the shelf" guys ... There are only 2 DACs that I thought were "almost there", that, AFAIK, use off the shelf chips: Light Harmonic and Bricasti.

alexandre

 

[microstrip]

Nope.... just connect it to a Sonoma and hit record or go from a format converter to SDIF-3 and record with a Tascam DV-RA1000HD

Bruce,

Just an hypothetical question - can some one having a DCS Vivaldi driving directly a power amplifier, that does not have analogue inputs, use a DSD5.6 ADC connected to the RIAA preamplifier and listen in real time to his LPs?

 

[edorr]

Nope.... just connect it to a Sonoma and hit record or go from a format converter to SDIF-3 and record with a Tascam DV-RA1000HD

Interesting. Does the Sonoma have vendor specific format converters? Not sure how the manufacturer specific links between transport + DAC work (I thought it was either encrypted or some propietary extension of the transmission protocol), but I'm pretty sure you cannot feed say the dCS DAC with the EMM transport, or the MSB transport into the EMM DAC.

 

[opus111]

Of course it's hardware+software, opus.
What I meant to say is that they only needed a software upgrade in order to be able to "decode" DSD.

Then you were using a different meaning of 'conversion' (to mean modify/upgrade) to the originator of the topic. Lasercd was using it to mean processing/rendering the 1bit signal into PCM form and that's the meaning I was using.

Well, my evidence is just the contrary. dCS and MSB (and Playback) are so far ahead of the other "off the shelf" guys ... There are only 2 DACs that I thought were "almost there", that, AFAIK, use off the shelf chips: Light Harmonic and Bricasti.

Which off the shelf DAC is Light Harmonic using? Have you heard an implementation of a segmented current source off-the-shelf DAC chip?

 

[asiufy]

Then you were using a different meaning of 'conversion' (to mean modify/upgrade) to the originator of the topic. Lasercd was using it to mean processing/rendering the 1bit signal into PCM form and that's the meaning I was using.

Don't know if that's the case. All I'm saying is that, through software, the DAC was configured to accept 1bit signals at 2.8/5.6Mhz. The hardware to do this was already there.

Which off the shelf DAC is Light Harmonic using? Have you heard an implementation of a segmented current source off-the-shelf DAC chip?

You'd have to ask them, or perhaps JackD knows. I don't think they mention this in their literature, but I don't remember them mentioning using discrete components either. Actually, for their new model, they mention using separate DACs, one for multibit, and a separate S-D one for DSD, a bit like AMR did at some point.
As for your question, you'd have to give me examples of actual products (not DAC chip model #s), since I don't know what's a "segmented current source DAC chip"
As I said before, even though I have preferences like everybody else, I don't have any prejudice against this or that. If something sounds better to me, I'll move on to that. This is what made me go from dCS to MSB, for instance.

alexandre

 

[opus111]

All I'm saying is that, through software, the DAC was configured to accept 1bit signals at 2.8/5.6Mhz. The hardware to do this was already there.

Yes sure, but that's somewhat tangential to the original question which was whether the MSB played back DSD natively or via some kind of processing. It seemed at first you were saying 'natively' but now its clear they're using software (or reconfigurable hardware) to do the job but still outputting via a multibit (not single bit) DAC. So I'm wondering what prompted the original question - do MSB claim in their marketing that they're doing the job natively?

As for your question, you'd have to give me examples of actual products (not DAC chip model #s), since I don't know what's a "segmented current source DAC chip"

AMR's RBCD DAC section is based on this form of DAC to give but one example. Zanden is another, Audial yet another.

 

[lasercd]

Yes sure, but that's somewhat tangential to the original question which was whether the MSB played back DSD natively or via some kind of processing. It seemed at first you were saying 'natively' but now its clear they're using software (or reconfigurable hardware) to do the job but still outputting via a multibit (not single bit) DAC. So I'm wondering what prompted the original question - do MSB claim in their marketing that they're doing the job natively?

I don't think they explicitly claim that it is able to see native DSD but I think its somewhat implied. BTW - although I use the Bricasti M1 I'm a big MSB fan. I use the UMT and love it. I raised the point simply because I'm trying to understand how a ladder DAC can accept native DSD when I've read on this forum and been told by manufacturers that it isn't possible. Just hoping to gain some knowledge.

 

[opus111]

Ah well you've come to the right place to gain some knowledge. A ladder DAC like MSB's can accept DSD if there's a digital filter in between to do the conversion. The filter needs to be a low-pass to get rid of all the ultrasonic noise, and it will need to have very strong rejection of all that hash or it will get aliased back into the audio band. An FIR (finite impulse response, no feedback) structure will do the trick - but as the input is only single bit this can be a multiplier-less filter, hence relatively simple in hardware terms.