[EuroDriver]

I have become a big fan of DSD128 and DSD256, listening to a handful of native DSD recordings and a lot of Redbook to DSD 128 and DSD256 conversions.

JRiver MC19 does the conversion competently, HQ Player and Foobar with the SACD plugin do it very well.  It's pretty close to Alchemy turning Redbook lead to DSD gold !

Last week I came across Bruno's blog about the Mola Mola DAC

http://www.mola-mola.nl/index.php/dak

I also had the opportunity to chat with Bruno at Munich High End last Saturday.  A really nice guy !

Just as the NC400 blazed the way for NCore, I am thinking can we DIY types do the same trail blazing for Hypex's discrete component DSD DAC ?

To keep the cost down and to keep the competition with the main stream to a minimum, I am thinking that the DIY unit should be DSD only and USB input only.  The unit does need a volume control so that we can hook it up to our NCores direct

Korg has pretty much this concept with their player software and their DSD only DAC http://www.musicradar.com/reviews/tech/korg-ds-dac-100-597226/

So what I am asking you to do is to register your interest.  As to the cost of the unit I would hope that Hypex would use a similar margin to what they do with the Nc400 and SMPS.

The more people we can get to support this, the better the chance we have of getting Hypex to do this

Let's wait until we have 5 registrations of interest before we post to other forums, then please post links to this thread in other forums you frequent.  I only lurk at Computer Audiophile and Pinkfish Audio, but am also registered at DIY Audio.

Suggested spec
- DSD 64, DSD 128, DSD 256, native and DOP format
- USB input with drivers for Windows, Mac, and Linux
- volume control with digital control interface
- display showing sample rate, volume level
- balanced output and single ended output simultaneously driven
- 12 volt DC input power so we can use linear power supplies or batteries

Welcome ideas

[cab]

Sounds great. I would ask for an 8 channel version for those with multichannel or active speakers.

[Nick77]

Sounds great, count me in with interest. 

[barrows]

I would love to see a DIY DAC from Hypex.  But DSD only would be a deal breaker for me.  Not everyone uses software players which have the option to do on the fly PCM-DSD conversions, and limiting a DAC like this to only those folks running certain SW is a losing proposition.
Additionally, there is no reason such a conversion scheme could not have exceptional PCM-DSD onboard in FPGA.  I would also trust Bruno Putzeys and his team to make this conversion algorthm much more than I would the JRiver developers.

USB only input is fine with me, but only if it is a USB Class 2 compliant USB interface for complete compaanility with OSX, Linux, and Windows (with a driver for Windows).

Certainly volume control should be available as well, and good output current drive for direct amplifier connection.

[santacore]

Very interested!

[EuroDriver]

Additionally, there is no reason such a conversion scheme could not have exceptional PCM-DSD onboard in FPGA.  I would also trust Bruno Putzeys and his team to make this conversion algorthm much more than I would the JRiver developers.

Hardware conversion of PCM to DSD is what the Mola-Mola DAC is going to be doing.  This is also what the EMM, Playback Design and PS Audio do at prices from 5k and up.  For SOTA performance Mola Mola might well be looking at a 6 k or higher price.  A DIY DAC with the same full format capabilities would certainly steal some sales from the Mola-Mola.  However a DSD only DAC that needs a software player to do the conversion, would limit the canibalization of Mola Mola DAC sales

[barrows]

Euro:  I respectfully disagree.  Whatever the Mola Mola DAC ends up costing, the vast majority of the costs associated with producing it are going to be the chassis cost and power supply cost.  The additional expense is that of the R & D.  Considering that the likely final price for a Mola Mola DAC will be in the 5 figure range, I very much doubt that anyone who has the spare cash to afford one, would rather spend time doing DIY (their time is too vaulable for that).
Bruno's converter architecture already requires lots of processing power just for converting DSD: his approach is not based on taking a 1 bit stream and directly converting it to analog, on the contrary, he is using 2 powerful SHARC DSP chips to oversample DSD to 32 bits and and then converting that 32 bit stream.  His approach to converting DSD already requires having the processing power, and the bulk of the development is done.  Are you suggesting he develop an entirely new topology for DIY only?  That would add expense, not reduce it.
Not sure what PS Audio, or EMM and Playback have to do with this, they all do things differently (and much differently from how Bruno is doing it, his approach is more akin to dCS and Chord).  Are you suggesting that all these approaches are similar?  They are not, PS Audio, EMM/Meitner, and Plaback are converting 1 bit data streams, Bruno is doing a multibit conversion.
Offering a DSD input only DIY DAC would severley limit its potential users and makes no sense.

[EuroDriver]

Offering a DSD input only DIY DAC would severley limit its potential users

Exactly !

Anyway it's for Hypex to decide.  Hypex really went out of their way with the NC400 to give the DIY community something very special at a price that common folk could afford.  I am just hoping to show that there a lot of DIY Hypex fans who would like to get their soldering irons busy on a Hypex 1 bit 100 MHz PWM DAC

[bruno]

The Mola-Mola DAC asynchronously converts everything to 3.125 MHz (32-bit) which is then turned into PWM with 5-bit resolution using 7th order noise shaping. This has about the same useful, that is noise-free, bandwidth as 256x DSD and way better SNR. Crudely speaking it's got about the same information carrying capacity as 6 to 8 times DSD - the latter having less capacity than its bit rate suggests on account of the low entropy of 1-bit sigma delta modulated data. The difference with other low bit schemes is that by using PWM every switch in the DAC carries the full signal which means using 32 of them only lowers thermal noise and jitter sensitivity.

Some of the more business minded folks at Hypex have already noted that a DIY DAC would make quite a splash. The idea keeps popping up every now and then. All you need to do is wait for it to resurface during a moment of relative calm.

Note that, should that happen, I can't make any guarantees about the conversion method used. I'm an engineer. I like to look at the complete problem -which, significantly, includes cost- and then work out a solution that best fits that problem. One-bit sigma delta is a useful format for building converters. Not the best, but properly implemented the results are quite fine. That's clearly not a clinching argument for or against choosing 1-bit no matter what. Neither is DSD: DSD means using 1-bit sigma delta data as a storage, production and release medium. It's a monumentally stupid idea because it implies that any processing you may want to do requires you to go to a multibit representation and then convert to one bit again. You can't edit, change gain or anything. As a result, DSD recordings that haven't gone through either multiple AD/DA conversions or repeated remodulation are incredibly thin on the ground (less than 100 in the tens of thousands of DSD titles available).

The whole DSD ethos is predicated on the idea that decimation and upsampling filters can't be made to work well. Sound familiar? Let me give you a hint. Even today the place is rife with folks who do not believe class D can ever sound good and who think negative feedback never solved any problems. I wouldn't have gone into class D, much less something like Ncore which relies on frightening amounts of feedback, if I were in the habit of following prevalent beliefs. This has always stood me in good stead and I don't see why I should suddenly feel compelled to do a DSD DAC simply because the concept has such great powerpoint appeal. I went into class D, not because I thought it would sound great but because of the practical advantages (efficiency) and it didn't seem to me to be fundamentally broken in terms of potential sound quality. I spent the next 15 years working to solve all the problems that stood between class D and great sound.

The fact that there are still class A/B amps around is obviously because the methods and solutions aren't universally understood. If there weren't any bad sounding class D amps, why should anyone want to consider a linear amp? And if there weren't people who didn't understand class D but still wanted to build amps, who would there be left to build linear amps? Same for solid state vs valve and feedback vs non-feedback.

I think PCM is a practical and efficient way of storing, processing and releasing audio data. I don't think it's got any problems that can't be fixed. The solutions and underlying principles are much more generally known than the ones that were needed to get class D on the rails. The fact that people "do DSD" as a data format only shows that those principles aren't universally understood.

_______
edit & PS: I wish to apologize that I won't be able to read or respond to this thread further. With exciting stuff to do there's exceedingly little time left to write about it.
PS2: I should add that EuroDriver's idea of using the computer as the front end and running the DAC as a fixed 1-bit back end struck me as delightfully quirky. It wouldn't fly as a product idea on practical grounds (requires a computer) but you can get above average results from it. I can imagine if someone built it and put all the design files online it could easily spin out into a kind of gainclone like popularity. An XMOS chip, a clock, a pair of flip-flops clocking out return-to-zero coded data and a decent output filter should deliver quite satisfying results (see the old converter paper on the Grimm site for the RTZ arrangement). It's insanely sensitive to wideband jitter so expect a lot of frustration doing the clock and the board layout. I'm not volunteering, mind you.

[EuroDriver]

Some of the more business minded folks at Hypex have already noted that a DIY DAC would make quite a splash. The idea keeps popping up every now and then. All you need to do is wait for it to resurface during a moment of relative calm.

Many thanks for the dissertation above !

Praying for a moment of relative calm 

[barrows]

Mr Putzeys: thanks for taking the time to share your experience here, it is well apprecaited.

EuroDriver, I suspect we will see a Hypex DIY DAC product at some time, and I suspect that if it comes, it will offer superlative performance for both DSD and PCM decoding.
In the mean time, let's just hope that Mola Mola finally gets their supply issues sorted, and their product to market, so that their team can catch its breath, it has been an extended road for them learning the ins and outs of being a high end audio manufacturer!