[bhobba]

I have no doubts that these fancy outboard supplies are better than the supplied wall wart. However, that will not affect the jitter that is generated by the poor supply that feeds the clock and SPDIF output chip. And that is what will really make a difference.

Yep Pat.  It seems rather silly to spend more on a power supply when simply disconnecting the clock and clocking it externally, as the pacecar does, yields a much better result at a cheaper price.  Although I may eventually get a pacecar, because I don't know of other similar products, I think Steve's solution could be done cheaper.
http://home.socal.rr.com/audio_gestalt/default.htm

If you, or someone, could create such a device at a reasonable price (say below Aus $500) then you would have at least one customer (me).

Some people claim they can hear jitter of 2ps.  I would like that claim put to a blind listening test some day.  Burson audio has a clock with jitter about 5-10 ps - maybe a bit higher.  They claim never to have found anyone that can detect that level of jitter.

Of course that is using an external DAC.  The claim is with these power supplies, and mods, the internal DAC sounds 'fantastic', beating things like a Wadia (not that I am a fan of Wadia mind - but that is another story).  I have a lot of respect for some of the people making that claim, but believe a well designed external DAC being fed a low jitter signal will outperform it.

Thanks
Bill

[Occam]

Pat,

If you've a suggestion, please feel free to make them. Technical discussions of various vendor offered products (including those with Circles) is fine, especially relative merits, pro and cons of different approaches, but '$value' judgements of those specific products make my life in the Lab Circle as moderator heck,  and I make an effort to keep it out of this Circle. As a sophisticated member of the trade, you obviously know how to keep the discussion technical.

I'll try and dig up some links to schematics, and ps topology for the SB, unless someone else posts them.

If you wish to solve that LP program of optimizing whatever modifications/addons by whatever objective functions you choose for us, that would be just keen..... aa

TIA,
Paul

[art]

Paul:

I am not trying to knock the "competition" (they really aren't competition, since we are not in the business of doing those sorts of product mods. ) I am merely pointing out that it make little sense to stick on an expensive outboard supply, and ignore a key problem. Seems to me that at least one of those guys would have thought of that already.

Don't know much about SN's stuff. I get the idea he has an external clock that you pump into the unit. Yes, that is one way to solve the problem. Still, the chip that does the clock also does the SPDIF output. (I believe more than one of us has pointed out that the 48 kHz clock can be disabled.)

If you only have one internal supply, (and a SMPS at that) to power the big FPGA and all of that stuff, it is going to be hard to completely isolate the noise it places on the rail from the critical chips.


The problem arises since the PSRR of  CMOS gate is only 6 dB at the midpoint. Which is around the point when the logic levels are changing states. Not the right time to have things wandering around, due to noise on the rails.

As for how much jitter you can hear??????

Well, I dunno. I would not say that one can actually "hear" the jitter. More like one can better detect lack of jitter. Jitter has to be quantified as to both magnitude and frequency spectrum. To say "It is 20 pSec", without having the spectral content known, is not really helpful.

But, back to your question:

A lot of folks that I work with claim that when you get jitter measurements down into the single digits (pSec-wise), you can not detect much of a change.

It also depends if the jitter is Gaussian or data-dependent. The latter being more discernable, in a negative sense.

Jitter is much less of a problem in CD players. The main way data-dependent jitter is introduced is when they use the filter chip to generate the clock. SPDIF is a different animal. Tons of data-dependent jitter.

Pat

[bhobba]

Don't know much about SN's stuff. I get the idea he has an external clock that you pump into the unit. Yes, that is one way to solve the problem.

That is exactly what he does.

Still, the chip that does the clock also does the SPDIF output. (I believe more than one of us has pointed out that the 48 kHz clock can be disabled.)

Yes - sigh.  My only consolation is that it is probably is not audible.  But aesthetically I think it stinks.

If you only have one internal supply, (and a SMPS at that) to power the big FPGA and all of that stuff, it is going to be hard to completely isolate the noise it places on the rail from the critical chips.

As for how much jitter you can hear?????? Well, I dunno. I would not say that one can actually "hear" the jitter. More like one can better detect lack of jitter. Jitter has to be quantified as to both magnitude and frequency spectrum. To say "It is 20 pSec", without having the spectral content known, is not really helpful. But, back to your question: A lot of folks that I work with claim that when you get jitter measurements down into the single digits (pSec-wise), you can not detect much of a change. It also depends if the jitter is Gaussian or data-dependent. The latter being more discernable, in a negative sense. Jitter is much less of a problem in CD players. The main way data-dependent jitter is introduced is when they use the filter chip to generate the clock. SPDIF is a different animal. Tons of data-dependent jitter.

Spot on Pat.  I am now starting to form the view a straight CDP may be the better choice right now.  Hopefully it will change as the potential of computer audio is obvious.

Thanks
Bill

[bhobba]

Whoops - sorry guys - the following was obviously written by Pat.  Didn't check my editing.

If you only have one internal supply, (and a SMPS at that) to power the big FPGA and all of that stuff, it is going to be hard to completely isolate the noise it places on the rail from the critical chips.

Sorry
Bill

[bpape]

If jitter is the only thing that "REALLY" make a difference... why is it so difficult to pick the best CD players.

I have no doubts that these fancy outboard supplies are better than the supplied wall wart. However, that will not affect the jitter that is generated by the poor supply that feeds the clock and SPDIF output chip. And that is what will really make a difference.

Exactly.  Nobody said the external PS was going to help jitter.

Bryan

[DSK]

... I am now starting to form the view a straight CDP may be the better choice right now.  Hopefully it will change as the potential of computer audio is obvious....

Bill, if you A/B a straight CDP and a modded SB/PS and prefer the CDP, then that is the correct choice for you. But don't make this decision based only upon a fear that there is a sub optimal part/circuit in the SB. The proof is in the pudding, not necessarily just the ingredients.

[NewBuyer]

The Transporter still uses a SMPS for the digital circuits...

The Transporter uses separate linear Super Regulators for DAC/clock/analog-out, and a low-noise SMPS for display/CPU.

Those who sell mods or use modded SB's (as seen in their sigs/links), should naturally defend their choice, and synergy is also critically important. I too am an "ears first" evaluator of this stuff!

I've recently borrowed a Transporter to give it an actual audition in my system. I wanted to do this for a long time, but tried other things first and didn't get to this until now. But now I don't want to give it back! 

While perhaps some would choose to buy/keep a modded-SB over a Transporter, I bet some others must go the opposite direction too. I've personally heard and greatly enjoyed different versions of modded SB3's with linear PS and some DAC combo's in my own system (never with Wayne's top $950 power supply though).

But even so, I've not personally heard any of these sound quite as satisfying, as this un-modded Transporter's analog-out in my setup. I expected it to sound great, but it has really surpassed my expectations - I didn't realize it sounded this good. I don't know what accounts for it, but to me, this is some seriously fine sound.

Well my ears have convinced me. I'm definitely buying a Transporter now, as a Christmas gift to myself!

[crooner]

Thinking about becoming the next Saul Marantz, of the wi-fi digital audio age. I already have the key basic design in my Super Squeezebox (better looking than the Transporter, in my opinion). I'll hire Pat as my Chief Engineer.
Can you say Sidney Smith??  aa

Watch the Super Squeezebox in action on YouTube! http://www.youtube.com/watch?v=sBDXxzkVY1U

[bhobba]

Bill, if you A/B a straight CDP and a modded SB/PS and prefer the CDP, then that is the correct choice for you. But don't make this decision based only upon a fear that there is a sub optimal part/circuit in the SB. The proof is in the pudding, not necessarily just the ingredients.

I have had a chance to think about it, and you are of course correct.  I can get a little carried away by techy stuff.  Great to see guys chiming in an keeping me on the straight and norrow.

Thanks
Bill

[brj]

I am not trying to knock the "competition" (they really aren't competition, since we are not in the business of doing those sorts of product mods. ) I am merely pointing out that it make little sense to stick on an expensive outboard supply, and ignore a key problem. Seems to me that at least one of those guys would have thought of that already.

Based on some conversations I've had recently, I know for a fact that it has occurred to at least one such party... and I suspect a few of the others as well.  The catch seems to be that what you are describing escalates the mods from those that can be done within the existing case to those that require a new case, which adds a whole new layer of costs and manufacturing complexity.  (This is especially true with a display involved.)

I think part of the concern is also that we're also talking about mods to a consumer unit, and while one Logitech employee has indicated that the SB3 will be around for a while, consumer units tend to rev faster than higher end audio products.  Developing extensive mods for something that may be replaced in short order, and one where you may need to worry about case manufacturing, entails a greater level of risk than creating external boxes and mods that fit within the existing case.  I hope it will happen eventually as either a DIY effort or a more ambitions professional mod, however, and would love the chance to audition such a unit.  (Mgalusha's DIY effort sems closest to this very desirable end-state.)

And I hope you will continue your technical discussions, Pat... I'm not DIY inclined just yet, but I find your threads very informative.

[art]

I'm glad some of you guys find them helpful.

Yes, I see your point. It is not uncommon for manufacturers to buy a stock "bog standard" CD (or DVD) player, and use it for the guts of a mega-$ unit. In a lot of those cases, there is a substantial redesign of the guts. At least when it comes to the part that affects the sound quality. (C'mon........you can't expect all of us to know how to make a micro that tells the radial and focus assembly to find the right part of the disc....at the right time, etc., etc.) In those cases, it only makes sense to buy a complete unit, and rework the daylights out of it.

And it shows in the price tag.

However.......

When you take a $300 unit (and one that packs a lot of performance into it) and try to do the same thing by throwing money at it indiscriminately......well............. ..then I start to question the rationale of the approach.

To do what I suggest, you need a new case. And a larger one. So, that costs money. Especially since it will have to look good for the price it will undoubtedly command. And then there is the price of audiophile-grade parts, etc., blah, blah.

Ok, yeah it can be done, but.......

When you add it all up, you just spent more than you probably would for The Transporter. (I think that is what they call the big mojo unit.)

So why not just buy one of them and be done with it??????? If you still fell the need to solve the problem of that money that must be burning a hole in your pocket, well I am sure that there will be some modder waiting to get first crack at it.

"Problem solved"!

Oh...btw.....when I had Dan's unit here, I tried sticking a large (1000 uF or so) cap on the rail, right at the clock/output chip. No help. At least on the 'scope.

Pat

[DSK]

...
When you add it all up, you just spent more than you probably would for The Transporter. (I think that is what they call the big mojo unit.)

So why not just buy one of them and be done with it??????? ...

Simple reason is that (from all the reports I've read) the tweaked out SB/PS outperforms the Transporter.... and this was before the new Bolder/Aspen Nirvana PS which though very expensive takes the SB up another significant notch. OK, so the SB is $300 + ~$500 for mods + ~$1500 for Nirvana PS .. total of ~$2300 versus ~$2000 for the Transporter. I would far rather pay the extra ~$300 for the increased performance and ~$2300 is still cheap for a top class digital front end IMHO.

Yes, I would prefer an internal PS (for aesthetic reasons) and the ability to play 96/24 discs (for the handful that I have) but I don't need all the other Transporter features and don't want to accept lower performance just for these attributes.

I'm sure others will see things differently and weight things differently but this is (my) answer to your question.

[JoshK]

Since we are discussing the technical merits of mods/techonology without listening to it... with regards to USB

Those looking at upgrading the power supply hope to get better performance without spending the money on something like a pacecar.  Wayne's power supply plus mods costs more than a pacecar - hence my reservations about it for use as a digital source.  Hugh's is a bit cheaper - but at 2/3 the cost a pacecar it begins to look not that cheap.

The pacecar, well that is a USB dac or USB interface right? I don't know how the pacecar implements its USB, but regards to USB in general, there are some key points about the USB interface which keep getting brushed under the proverbial rug...

1) USB is inherently not galvanically isolated and we all know that computers are a haven of a noise and RFI/EMI.  I don't want my audio ground being tied to this, do you?  You may be able to use a transformer, I don't know.
Ethernet is galvanically isolated.
2) Don't most USB dac chips generate their clock from the computer?  No thank you. 
Ethernet is a FIFO stack type scheme that doesn't get its clock from the computer as far as I understand it.
3) Reports I've read implied that the USB to SPDIF chips were worse than the really bad SPDIF chips in terms of jitter producers.  You know what they say about polishing a turd? 

So I see this as many less than ideal solutions.  I am not sure why the transporter doesn't fair better, it seems that they atleast a better technical solution.  Could it be that we like jitter?  It masks sins and adds "soundstaging"?  Don't know.

I think you ultimately pick your poison and deal with its flaws.  Myself, I like the SB2/3 solution and plan to replace the internal switcher with individual linear supplies.  How hard is it to make a but of linear supplies?  Not hard if you just use the blog standard LM317T used in most all consumer CD players.  Better regulators take quite a bit better understanding of things.

[jhm731]

To do what I suggest, you need a new case. And a larger one. So, that costs money. Especially since it will have to look good for the price it will undoubtedly command. And then there is the price of audiophile-grade parts, etc., blah, blah.Ok, yeah it can be done, but....... 

Pat -

It has been done. DB System (http://www.db-system.ms/) built SBs with multiple linear voltage rails and clock
inputs for their customers when the SB2 first came out.

[Nuuk]

and don't forget the at-tunes SB+! 

But that rather confirms what Pat is saying about the cost of making something to that spec with a new case!

[dwk]

This is a bit old and may be straying a bit OT, but some things jumped out at me

Since we are discussing the technical merits of mods/techonology without listening to it... with regards to USB


The pacecar, well that is a USB dac or USB interface right?

No, that's not correct. There are several versions of the Pace Car, and the one for the SB doesn't use any USB. The Pace Car is all about exporting a good clock from the DAC and slaving the source to that clock. In the case of the SB this is just the 11.28 or whatever crystal clock. Steve replaces the crystals on the SB with a clock feed from the Pace Car, and then uses a synchronous reclock circuit at the dac to clean up any remaining problems.

I don't know how the pacecar implements its USB, but regards to USB in general, there are some key points about the USB interface which keep getting brushed under the proverbial rug...

There may be some folks that aren't aware of the problems of USB, but I don't think there are many 'in the know' that are sweeping them under the rug; there aren't that many truly working to solve the problems though.

1) USB is inherently not galvanically isolated and we all know that computers are a haven of a noise and RFI/EMI.  I don't want my audio ground being tied to this, do you?  You may be able to use a transformer, I don't know.
Ethernet is galvanically isolated.

Since I currently use analog out from my soundcard and am reasonably happy, I think the 'problems' with PC power tend to be over stated (although I'm running fully balanced which tends to help). However, when talking about true 'state of the art' you probably do want to isolate. You can do this over USB with an optical USB cable/extender. Not entirely cheap at ~$200, but that's mere pennies in the context of this thread.

2) Don't most USB dac chips generate their clock from the computer?  No thank you. 
Ethernet is a FIFO stack type scheme that doesn't get its clock from the computer as far as I understand it.

There are a couple ways to do USB. The cheap/easy/generic way is to use isochronous mode, in which case the PC USB clock is the master clock. This is indeed a rather poor way to do things, and you need to clean up the clock to have anything useful.
However, what the good USB solutions do is use async mode, where the procotol between the PC and the DAC is more complex. Basically, flow control is implemented so that the DAC can be the master clock. The USB device will send commands telling the PC to speed up/slow down as needed, and then buffer and clock out according to a good local clock.  This is what Gordon Rankin at Wavelength does, I believe.  Steve's Pace Car works on a similar principle, but I'm not sure whether it uses async or adaptive mode; the net result is the same though - the master clock is not derived from the PC or USB clocks, and so the performance is dictated entirely by the local high-precision clock.

3) Reports I've read implied that the USB to SPDIF chips were worse than the really bad SPDIF chips in terms of jitter producers.  You know what they say about polishing a turd? 

They can be, but I suspect that anybody doing this is using an ASRC or some other clock-cleansiing approach downstream. You can also do USB->I2S directly, and with a decent ASRC afterwards it's reportedly quite good. (the Twisted Pear Opus stuff is an easy way to do this)

So I see this as many less than ideal solutions.  I am not sure why the transporter doesn't fair better, it seems that they atleast a better technical solution.  Could it be that we like jitter?  It masks sins and adds "soundstaging"?  Don't know.

I'm not sure I'd call the Transporter/SB approach 'better', but it's certainly a viable approach.  The key with any digital setup is to ensure that a good clock local to the DAC is the master, and everything else is slaved/sync'd to that to avoid drift. Everything else is implementation details.

My disclaimer is that I don't actually use any of these approaches, since at the moment none are viable for multi-channel output. I do think that Steves Pace Car ideas are the most generally applicable to what I'm trying to do which is why I've paid attention, but I don't own any of his stuff.

[JoshK]

Doug, very good points, I thank you.  I know just enough to be dangerous.

FWIW, I respect Steve's technical knowledge and wasn't questioning his approach, since I knew he would have done it well.  My comments on USB were meant more generally to mass gear out there that has USB inputs, as if that was the next holy grail.  I should have stated that.

Since I currently use analog out from my soundcard and am reasonably happy, I think the 'problems' with PC power tend to be over stated (although I'm running fully balanced which tends to help). However, when talking about true 'state of the art' you probably do want to isolate. You can do this over USB with an optical USB cable/extender. Not entirely cheap at ~$200, but that's mere pennies in the context of this thread.

Fair enough.  Good to know about the optical isolator/extender, but I'm married to RJ45 at this point myself.   Do you know if there are tranformers appropriate to USB?

There are a couple ways to do USB. The cheap/easy/generic way is to use isochronous mode, in which case the PC USB clock is the master clock. This is indeed a rather poor way to do things, and you need to clean up the clock to have anything useful.
However, what the good USB solutions do is use async mode, where the procotol between the PC and the DAC is more complex. Basically, flow control is implemented so that the DAC can be the master clock. The USB device will send commands telling the PC to speed up/slow down as needed, and then buffer and clock out according to a good local clock.  This is what Gordon Rankin at Wavelength does, I believe.  Steve's Pace Car works on a similar principle, but I'm not sure whether it uses async or adaptive mode; the net result is the same though - the master clock is not derived from the PC or USB clocks, and so the performance is dictated entirely by the local high-precision clock.

I am understand, generally (big picture), what async is about and agree that this provides a clean solution.  I also figured Steve would do something like this.  Again, I was only throwing out that USB creates problems, that need such as solution, just as SPDIF does.  My general sense was that many (at least the technically unaware) think that computer audio is an improvement over a transport, thinking it takes the jitter away. 

You can also do USB->I2S directly, and with a decent ASRC afterwards it's reportedly quite good. (the Twisted Pear Opus stuff is an easy way to do this)

I was aware of the USB > I2S, but understood it was fairly new.  I believe Steve uses this approach to USB.

I'm not sure I'd call the Transporter/SB approach 'better', but it's certainly a viable approach.  The key with any digital setup is to ensure that a good clock local to the DAC is the master, and everything else is slaved/sync'd to that to avoid drift. Everything else is implementation details.

That is a good point.

P.S. Steve if you are reading this, I am not meaning to drag you into this by bringing your name up.  I was trying to discuss the problem in general and trust you have already created a valient solution. 

[Nuuk]

Very nice work of you. However, since most of us especially in my case that I am a DIYer (considering as an amateur) and do not clearly understand what components (resistors and capacitors) are used to modify with Newava S22083 which I have in hand in the SB3 according to your picture. I have not only a little understanding to your explanation but also hardly recognize the values of resistors you are using and what positions did you connect to. Therefore, would you please provide us some of the lists with exact values of the resistors and to which positions should these resistors be soldered. Thanks for your effort and time.

I would also like to thank Pat for the information in this thread but add my request to have the actual mods 'spelled out' for those of us not able to work it out for ourselves.

Will do some additional listening/break-in and try and determine which output sounds better.

Are you able to tell us which one you prefer yet? 

[jhm731]

Very nice work of you. However, since most of us especially in my case that I am a DIYer (considering as an amateur) and do not clearly understand what components (resistors and capacitors) are used to modify with Newava S22083 which I have in hand in the SB3 according to your picture. I have not only a little understanding to your explanation but also hardly recognize the values of resistors you are using and what positions did you connect to. Therefore, would you please provide us some of the lists with exact values of the resistors and to which positions should these resistors be soldered. Thanks for your effort and time.

I would also like to thank Pat for the information in this thread but add my request to have the actual mods 'spelled out' for those of us not able to work it out for ourselves.

Will do some additional listening/break-in and try and determine which output sounds better.

Are you able to tell us which one you prefer yet? 

I can't do a quick A/B, since I only have one BNC input.

Switching Pat's new digital cable from one output to the other, both outputs sound almost the same. At times,
I think the Newava S22083 output sounds a bit faster, but I wouldn't bet the ranch on trying pick them out in a DBT. 

The key is both outputs sound much better than the stock SB3 outputs.