[AKSA]

Folks,

I have just posted an updated NEWS section to the website, here:
http://www.aksaonline.com/news.html    

I discuss progress on the retail GK-1, upgrades to the Nirvana (and a small price rise for the Nirvana, now posted), progress on the DAKSA, the new AKSonic Loudspeaker, advertising strategy, and more.

It is my hope that this keeps people up to date with happenings at Aspen - we are certainly not standing still!

Sincere thanks for all your continuing support,  

Cheers,

Hugh

[tg3]

Great news, Hugh.

I am especially interested in the DAKSA. However...

The DAKSA will be the only *complete* DIY DAC kit available with absolutely all components and with comprehensive, easy to follow instructions.

Do the Scott Nixon DacKit and TubeDAC kits not qualify?

Your offering certainly sounds more interesting!

I think you will have real winner with the DAKSA. Might I suggest that the unit be 'embeddable' into CD transports for the DIYer who does not want yet another box connected with even more cables!

[Jens]

Excellent news!

Keep up the good work, Hugh!

[AKSA]

Hi Anthony,

You are right, dammit, should've done more homework.  My apologies to Scott Nixon and TubeDAC;  their kits are indeed complete!

We too considered a tube output stage but dismissed it since we envisage the DAKSA will be used with the GK-1, for which it is designed, in fact....

We thought about using the tube as an I/V converter, but then shelved this idea too because we felt the Didden conversion regime was far more sophisticated and would give better results.

Thanks for the encouragement, Jens, will email you privately a little later today.

Cheers,

Hugh

[AKSA]

Ben and I have been burning the midnight oil.  

Two weeks ago Ben had completed the initial chipset and topology.  He has continued with refinement however and only three days ago he explained the finer points of the DAKSA's digital processing in a very pleasant cafe in the eastern suburbs of Melbourne.  Things are now set in concrete, chips are arriving, both purchased and samples, and Ben is progressing the pcb design swiftly.  Suffice to say his ideas are unique, and there will be no other DAC like this in the world.  Its principle of operation is revolutionary in the Red Book world, nough said....  

The principal outcome will be this:  This DAKSA will be able to lock down jitter to considerably less than 10 picoseconds REGARDLESS OF THE TRANSPORT USED!!  The DAC chips will be controlled by an ATMEL microprocessor, which will do all the signal routing and control the clocking using a high precision crystal oscillator.  The interconnection will be standard SPDIF;  no additional interlinking of clocks will be necessary.  A secondary outcome will be that the output impedance will be less than 0.5 ohm, making it almost immune to interconnecting cable variables.

Just this evening I finished the current to voltage converter pcb, having already completed the schematic and testing of the prototype last week.  This pcb was complex and consumed about forty hours.  There will be more refinements to follow.  The entire analog output stage uses ten transistors, features dazzling linearity, has built in discrete regulation to +/-24V (giving enormous headroom and low distortion) and fits on a double sided pcb approximately three inches square.  This will be incorporated onto a single pcb for the entire DAC, and DIYers should have no difficulty assembling the DAKSA.

I should caution anyone expecting this to be ready by Xmas.  This is extremely unlikely;  for starters several thousand dollars of sourcing is yet to come, not to mention thorough documentation, which can take up to three months.  Needless to say, no deadlines will be given.  

The GK-160/150 is also coming along well.  Laser cutting of bases and top covers is now partially complete, and all is moving nicely.  I should have some available for sale around Christmas.  This GK-1 will certainly be very special, and will look a million dollars.  The chromed transformer covers are just gorgeous!

Thanks go out to all AKSA supporters, past, present and future, who have spurred us on to greater heights.  

Cheers,

Hugh

[Agisthos]

The principal outcome will be this: This DAKSA will be able to lock down jitter to considerably less than 10 picoseconds REGARDLESS OF THE TRANSPORT USED!!

It's Interesting that you mention this. From what I gather the problem with dac/transport combos was that the transport quality itself had an impact on the sound.

but some of the latest dacs that have been coming out have been using special methods to reduce jitter and reviewers claim that the Transport is no longer an issue, you can use any and the sound is not lessened.

How it should be imo.

[Tinker]

It's Interesting that you mention this. From what I gather the problem with dac/transport combos was that the transport quality itself had an impact on the sound.

but some of the latest dacs that have been coming out have been using special methods to reduce jitter and reviewers claim that the Transport is no longer an issue, you can use any and the sound is not lessened.

How it should be imo.

< Long post ahead>

I heartily agree.

Transports are a problem indeed, and the interlink between transport and DAC is a major source of jitter. Jitter within a CD player is bad enough, owing to complex interactions between components. It has become popular to retrofit high stability clocks in players to try and reduce jitter. Reclocking a whole transport works up to a point, but a clean clock only goes so far in many players owing to power supply modulation, signal dependent switching artifacts and the multitude of logic gates interposed between clock injection at the transport and the recovered data stream at the DAC chip.

No DAC can every be completely free of transport "sound", but it is entirely possible to reduce the contribution to something negligable.

A technique gaining popularity is asynchronous resampling. In this scheme a rock solid clock is used to resample the incoming data using a high order polyphase oversampling filter, which removes short term jitter (down to the purity of the new clock) but cannot correct for frequency drifting and some other types of modulation in the source (nothing can!). This technology has matured in recent years with the need to glue all kinds of incompatible digital systems together. The complete severing of transport and DAC clock simplifies system integration hugley and gets rid of problems with clock coupling. The resampling process, however, does introduce a little bit of distortion and needs decent bit depth to maintain signal integrity.

Another common technique is to use a PLL (Phase-Locked Loop) which chases after the input clock (from the transport) steering a high-quality clock toward the frequency of the incoming transport clock, but interposes a low-pass filter to remove short term cycle to cycle clock phase noise. These are great, but have a few limitations, the main one being that they merely attenuate jitter (admirably) but can't remove all types and frequencies and certain designs can "motorboat" limiting the achieved jitter reduction. People seem more comfortable with the PLL idea because the data from the CD is not altered by resampling, but simply passed on bit-for-bit through the reclocking process.

A very expensive approach is to use a RAM player. Jitter is possible here  too. A workable RAM player is expensive, but becoming a fiscal possibilty very quickly. But more on that some other time...

In short, there are several techniques which can attenuate jitter markedly. The typical jitter on the recovered clock for separate DACs is about 200pS RMS, even if the rtasnprot is better than this! There is empirical literature which shows that 100pS jitter is easily audible, and some quick mathematical approximations will show that less than 50pS is needed to realise the noise floor that CD is capable of. This would be fine were jitter just "noise", but jitter creates distortion. Since jitter is the source of some of the worst kinds of distortion audiophiles deserve to profit from the advances in jitter-reduction techniques. The AKSADAC is going to be a balls-and-all approach stopping short of cesium clocks and small computers normally used for measuring phase differences in relativity experiments, made from readily obtainable and time-tested components. How this is going to be done is being kept under wraps for now, but I can say it is related to a digital implmentation of a PLL.

I hope this long post opens up a can of worms, of sorts. I am convinced that CD player technology is the weak link in many systems that are otherwise extremely hi-end (e.g. AKSA). This is all the more irritating since a DAC/player is hard to build. Hence the AKSADAC, a kit to combat this shortcoming.

Adios for now.
T.

[blizzard]

Hey Tinker,
  Thanks for the info.  That was great.
  You wrote:

A very expensive approach is to use a RAM player. Jitter is possible here too. A workable RAM player is expensive, but becoming a fiscal possibilty very quickly. But more on that some other time...

If you could give more detailed information on the RAM type players, I would appreciate it.  This is a very interesting topic.  I have also always felt that the CD technology is most often the weakest link.

             Thanks for sharing,
                    Steve

[Agisthos]

Yeah I second that, thanks Tinker very informative

What do you think about the upsampling and oversampling? I have seen some net reviews of some new dacs that only use 16 bit 44k but the sound is as good as anything else so prehaps all that sampling is not needed?

this caught my eye in regards to ram type players

a recent Hifi Choice review (march)  of the Teac P-70 Transport and D-70 Dac. Each piece cost 6000 pounds.
It has both types - Ram and PLL and you can select between them. Both were as good as each other, but when using the Ram buffer they claim that there was no difference between a budget cd player and the 6k transport!!

[Tinker]

Yeah I second that, thanks Tinker very informative

What do you think about the upsampling and oversampling? I have seen some net reviews of some new dacs that only use 16 bit 44k but the sound is as good as anything else so prehaps all that sampling is not needed?

this caught my eye in regards to ram type players

a recent Hifi Choice review (march)  of the Teac P-70 Transport and D-70 Dac. Each piece cost 6000 pounds.
It has both types - Ram and PLL and you can select between them. Both were as go ...

Hi Agisthos and Blizzard. I began writing a long and considered response to this, which I will develop a little and place on the Aspen website with Hugh's permission, rather than post it in full length here.

For now let me potentially put the cat among the pigeons by saying that IMHO oversampling is a mixed bag but probably a necessary evil for most systems (although it has some up sides too!) and the generally positive reviews of RAM players further illustrates how crucial solving the problem of jitter is.

T.

[Agisthos]

Im guessing you must be Ben right? who is developing the Daksa. Will it use optical or coaxial connection?

[AKSA]

Agisthos,

Yes, Ben is Tinker......

It will have coaxial connection.  No optical.

Cheers,

Hugh

[Tinker]

Im guessing you must be Ben right? who is developing the Daksa. Will it use optical or coaxial connection?

Ahhhh! My cover's blown.  

Yes, following on Hugh's remark: most hi-end systems have both optical and coaxial output, and nearly every DVD player I have seen, including the $85 ones has, a coaxial ouptut. Probably because the two-box (ie separate receiver) DTS systems use digital coax for 5.1 transmission.

A while ago there was some debate on transmission and optical looked like it ought to be better than coaxial as the light pulses wouldn't suffer the edge degradation of an electrical pulse in 75ohm RCA coax. However, numerous folk in the pro-audio community made measurements and found that other problems with optical technology made it a much more jittery interlink system. Opto couplers are widely used to electrially separate systems. However, in the hi-speed digital world other forms of isolation are often preferred (not that CD is high-speed by digital standards).

Of course you could retrofit an optical input if you wanted...

T.

[Agisthos]

coaxial seems to have the same connector as normal line connectors. but you cannot use a normal interconnect cable right?

[EchiDna]

coaxial seems to have the same connector as normal line connectors. but you cannot use a normal interconnect cable right?

you could use a normal interconnect, it will work... but you ideally should use a 75 ohm coax interconnect. Personally I'd like to see the option of BNC over RCA...

[blizzard]

Hey Tinker,
  Now that your cover is blown:
  Just wanted to say that I saw the talk you gave at the AKSAfest.  It was outstanding.  You did a great job -- very practical and understandable.  I had a great time watching it -- wish it was longer.

  I'm really looking foward to your RAM DAC explanation.

            Thanks,
                 Steve

[Tinker]

coaxial seems to have the same connector as normal line connectors. but you cannot use a normal interconnect cable right?

you could use a normal interconnect, it will work... but you ideally should use a 75 ohm coax interconnect. Personally I'd like to see the option of BNC over RCA...

Both Agisthos and EchiDna are right. You can use an ordinary RCA, and I have never seen this fail, but the cable plug and socket should have a characteristic impedence of 75Ohms to ensure the higest integrity and hopefully lowest jitter. This would matter most on long lines.

I have both BNC and 75ohm RCA (they do make 'em!). You could retrofit a BNC socket to any design (nice gas-tight socket!), although I believe many of these are 50Ohm for networks. Now we're really splitting hairs.

Iremember not being game to mod my first decent Rotel so I made a whacky cable out of monster video link which had an RCA on one end and a BNC on the other.

T.

[Tinker]

Hey Tinker,
  Now that your cover is blown:
  Just wanted to say that I saw the talk you gave at the AKSAfest.  It was outstanding.  You did a great job -- very practical and understandable.  I had a great time watching it -- wish it was longer.

  I'm really looking foward to your RAM DAC explanation.

Thanks for the kind words Steve/blizzard. I really enjoyed the trip to AKSAfest. I still can't believe how switched on all the AKSAPhiles are. And I really enjoy talking and listening to folks who are as mad about this stuff as I.

As mentioned, I am going to put up some stuff on the Aspen site about this stuff. I don't want to keep folks waiting too long, so the first installment on filtering will arrive soon. Part II will be jitter, PLLs and RAMDACs.

Consider the logical extreme of the RAM idea: to acutally "burn" the the entire CD to memory chip at high speed and then play it from there. No moving parts. I have talked (in concept, no software yet) with a friend about this and we are toying with a fanless micro server to do this, although the best idea would be no moving parts at all, use a flash card. However,  1GB flash is over $1200 so I put that one on hold...

What a tangled web... I still agree that transports are problematic, but I believe that at their best with the right DAC they can be more than adequate for very high fidelity playback.

Cheers,
           T.

[blizzard]

Hi Tinker,
  I too have felt that eventually we would get away from moving parts as we have in CD/DVD players and also hard drives in computers.  I don't know how much longer it will take.  But the way science/physics (especially nano-technology) is progressing, I feel that it is just around the corner.  In keeping with that, I'm sure the $1200 price tag of that 1GB flash will be dropping soon too.  Specifically, the question would be how soon is soon?
  You  have some great cutting-edge ideas.  Technologically speaking, your future concepts should prove very interesting.  

                Cheers,
                     Steve

[JohnR]

Consider the logical extreme of the RAM idea: to acutally "burn" the the entire CD to memory chip at high speed and then play it from there. No moving parts. I have talked (in concept, no software yet) with a friend about this and we are toying with a fanless micro server to do this, although the best idea would be no moving parts at all, use a flash card. However, 1GB flash is over $1200 so I put that one on hold...

I still don't understand why it can't just be read from disk and buffered....

Anyway, is there a target price on the DAKSA kit? (Maybe one was mentioned at AKSAfest but I've forgotten if so )