New M-dac power supply

[realysm42]

Got something a little special ordered for the M-dac :eyebrows:

ETA about 2 weeks; will keep you posted.

[technobear]

Intriguing http://www.technobear.btinternet.co.uk/emoticons/hmmm.gif

You know of course that JW is developing his own, the MPAX which adds a few more bells and whistles as well as being a better PSU.

Not aware that he has a prototype yet - or does he? :eyebrows:

[sq225917]

You do know that the PSRR on the Mdac is already better than -120db for most sections and around -150db for the digital line to the dac chip itself? Those numbers don't really leave much scope for improvement without major internal surgery. Certainly any external supply is unlikely to offer any benefit- maybe you could scrape a tiny bit by running a transformer for every separate voltage rail, but gains would be minuscule.

[MCRU]

Intriguing http://www.technobear.btinternet.co.uk/emoticons/hmmm.gif

You know of course that JW is developing his own, the MPAX which adds a few more bells and whistles as well as being a better PSU.

Not aware that he has a prototype yet - or does he? :eyebrows:

yes he does, it was on show at munich and as to simon's post why would john make an external psu himself if nothing can be done to improve the mdac (or slightly as you say)?

worth a punt at least simon surely?

[Marco]

You do know that the PSRR on the Mdac is already better than -120db for most sections and around -150db for the digital line to the dac chip itself? Those numbers don't really leave much scope for improvement without major internal surgery. Certainly any external supply is unlikely to offer any benefit- maybe you could scrape a tiny bit by running a transformer for every separate voltage rail, but gains would be minuscule.

You're not suggesting that this (undoubtedly 'adequate'), but puny little plastic thing with its cheapo 'kettle lead', is the best that can be done, are you?


http://img171.imageshack.us/img171/6771/mdacpsu.jpg (http://imageshack.us/photo/my-images/171/mdacpsu.jpg/)


http://img266.imageshack.us/img266/8109/mdacpsu2.jpg (http://imageshack.us/photo/my-images/266/mdacpsu2.jpg/)

Numbers? Shmumbers... Let's see what some big 'fuck off' toroidal transformers can do instead, in a more elaborate power supply design, from someone like lurcher, Paul Hynes or Anthony TD, and then have a listen, and bugger the numbers! ;)

Marco.

[realysm42]

You've really taken issue with that thing haven't you Marco :lol:

I'm aware of JW's stuff yeah; it's not where mine's coming from, that's all Ill sat for now.

And Simon, I have no idea what you're saying even means; to a simpleton to me (and from what I've read time and again on here) is that vanilla psu's tend to be a major limiting factor in whatever equation they're placed into, so I'm happy to have a try with something else :)

[Marco]

You've really taken issue with that thing haven't you Marco :lol:


Lol - it's fine, but merely 'adequate for purpose'. Looking at what is inside it, it really can't be anything else.

I'm afraid that in audio I'm from the school of 'over-specced to fuck is best', so I expect the (DC) PSUs I use on my components to be as linear as a very linear thing, weigh around 20kg, and be packed full of 'bad boy' toroids ;)

Quite simply, there's NOTHING which has more impact on the performance of audio equiment than the power supply: it is the 'engine' which drives everything else, so in that respect, I'd rather have a Rolls Royce one at the helm, than a 'toytown' one from a Mini Metro!

Marco.

[sq225917]

Marco, you miss the point entirely. You are assuming the size of the fuel filler cap on the car has anything to do with the way it drives- in this case it doesn't.

The Mdac has multiple stages of internal regulation with banks of low ESR organic electrolytic caps specially made for the purpose. It literally does not matter what you do on the outside, if the internal stages have noise already down between -120db and -150db, fannying about with an external transformer will make exactly the square root of SFA difference. Ask Paul and I'm sure he'll explain why.

The transformer is more than 'adequate' for the task at hand, in fact its current delivery capability is several times greater than the current draw. it effectively provides nothing to the regulation of the power supply it just delivers a few rails at the correct voltage. So adding a bigger one would do nothing, adding one for every voltage rail 'might' reduce modulation and noise by a tiny amount. Adding big ass wires to it will make no difference, they are already 10x bigger than they need to be to deliver the paltry half amp it uses split across 6 rails.

The PSU upgrade that john in making adds a whole load of signal bypass and control circuit functionality as well as upgrading the PSU, the actual PSu upgrade part of the package comes from replacing the separate internal PSU board that proceeds all the rest of the dac circuity and reworking the main pcb itself to add further different stages of regulation right on top of the active circuit devices themselves.

So if Realy is having someone remove the internal PSU board and replace it with something much better, then yeh he's likely to hear an improvement if it's done well and the replacement parts do better than -120db (which is pretty damn good). But buggering about outside the box isn't going to do anything.

Of course we are assuming he means a PSU upgrade rather than a lovely new mini server or similar.

[Marco]

Hi Simon,


The Mdac has multiple stages of internal regulation with banks of low ESR organic electrolytic caps specially made for the purpose. It literally does not matter what you do on the outside, if the internal stages have noise already down between -120db and -150db, fannying about with an external transformer will make exactly the square root of SFA difference. Ask Paul and I'm sure he'll explain why.


Ok, I will, and I'll ask Paul to contribute his thoughts here, which should be interesting. I'll also ask Nick (lurcher) and Anthony TD to contribute, both of whom are also experts in PSU design, for high-end audio purposes.

I should add that, you know me, and if what you claim is true (and it may indeed be), but the ONLY way I'd believe it is by listening, and using my ears as the judge, not simply by going by what the 'numbers' say.

Therefore, if Nick (lurcher) or Paul Hynes designs an off-board PSU for the M-DAC, then I'll ensure that I listen to them and make the necessary comparison, with the unit used as 'stock', and then with the various off-board PSUs, in the equation.

I've not heard an M-DAC, so it'll be interesting exercise anyway comparing it against my 'old school' modded Sony, and see how the latter shapes up against it :)

Marco.

[realysm42]

How comes you've stuck with that for so long Marco, nothing you've heard yet bettered it?

Have you looked for anything else?

[Marco]

Yes to the last one, and no to the first one. Everytime I compare it with something modern, and arguably technically 'better', the old timer has the audacity to win!!

So it gets bored trying... ;) Both Nick and Anthony have heard my DAC and know of its sonic capabilities.

Perhaps this could be a new challenge, though?

Marco.

[Marco]

The other thing too, and this is aimed at Simon...

If there wasn't any mileage in making a more highly specified off-board PSU for the M-DAC, then why is John Westlake giving advice on this thread on pfm, as to how best to do precisely that:

http://www.pinkfishmedia.net/forum/showthread.php?t=120306


...with people building stuff like this, in order to gain a sonic improvement:


http://img21.imageshack.us/img21/9392/pfmpsu.jpg (http://imageshack.us/photo/my-images/21/pfmpsu.jpg/)


Surely if there was no sonic benefit to be had, like Simon suggests, the manufacturer (who, after all, knows best) would say so, and wouldn't be wasting people's time guiding them through building a DIY PSU? :scratch:

Marco.

[sq225917]

Like I said they are all bypassing the internal AC/DC psu and providing regulated DC directly to the internal regs, not just replacing the external AC supply with a 'bigger transformer'.

[Marco]

Sure, I don't really care what the method is - it's the end result that matters!

Clearly, there are potential sonic benefits to be had by using a good DC (as opoosed to AC) power supply, to power the M-DAC, which means that the supplied AC one is a compromise, just as I originally suspected :)

Marco.

[anthonyTD]

Looking at the picture of the outboard power supply, it would seem that this is only the bare AC part, ie; the DC rectification and regulators are actualy inside the unit its powering, if this is correct then as long as the transformer is rated well for the load its intended to supply,[good basic regulation] then the rest is mostly up to how good the rectification, and regulator circuits perform inside the unit.
On another note, some people think that having banks of huge value capacitors is all that is needed for a good stiff power supply, unfortunately capacitors do not have a linear impedance with frequency, hence the need for well designed active regulators that have ultra low impedance way beyond the audio frequency bandwidth.
There are many people here who are well aware of what is needed to make such power supplies, so there is no need to teach people how to suck eggs.:)
A...

[Paul Hynes]

Hi Marco,

In principal any power supply can be improved with a little thought and attention to detail.

Firstly, quoting noise levels of –120 dB to –150 dB means little without qualifying the measurement spectrum and operating conditions for both the regulation and the load. All regulators have a limited bandwidth and it’s just a question of where their regulation ability starts reducing due to bandwidth limitations. Once the gain of the regulator error amplifier starts to roll off with increased frequency, the regulation capability reduces, as does the supply line rejection. Whilst the cascaded regulation used in the Mdac may well measure –120 dB to –150 dB at audio frequencies it can be an entirely different matter at 1 MHz, 10 MHz and 100 MHz. The cascaded low noise regulation combined with the SR7 used on your SL1210 easily achieves supply line rejection exceeding this well past 100 KHz, and does not run out of regulation capability until it gets well past 100 MHz. Upgrading the mains transformer is clearly audible even with this power supply as you are well aware.

Most regulators are virtually ineffective above 10 MHz. I should point out that the clock, used in the Mdac for timing the Sabre DAC chip, is likely to be clocking at frequencies in excess of 80 MHz. Anthony is quite right, capacitor banks have their limitations and regulators have to be designed with exceptionally wide bandwidth to have any effect at these signal processing frequencies, and they must be damn fast to allow stable signal processing whilst keeping supply reaction low. This is why I use 5,000 volts per microsecond regulator error amplifiers, with a 1 GHz gain bandwidth product, for the regulator kits I make for the Buffalo DAC, Eastern Electric Minimax DAC etc. which use the Sabre Dac chip, to keep regulation and supply line rejection going to the highest frequencies I can. My regulator kit for the Buffalo Dac, etc is internationally regarded as the best regulation available for the Sabre Dac chip.

I do not have information about the specification of the Mdac external supply but if anyone can provide the specification It is easy enough to design a custom high performance supply to replace it. This will prove the point one way or the other.

Regards
Paul

[sq225917]

Here the internal psu board schematic

http://i50.tinypic.com/2wn2yxt.jpg

and the pin out for the external trasformer


http://i50.tinypic.com/dxmhyp.jpg

[Marco]

Hi Paul and Anthony,

Many thanks for your input. Anthony:


On another note, some people think that having banks of huge value capacitors is all that is needed for a good stiff power supply, unfortunately capacitors do not have a linear impedance with frequency, hence the need for well designed active regulators that have ultra low impedance way beyond the audio frequency bandwidth.


Interesting... Basically, it goes back to the tried and tested practice of keeping circuits as simple as possible and reducing the component count, wherever one can, in order to minimise unwanted interactions, thus potentially maximising the sonic effectiveness of the circuit. The best component, after all, is NO component. However, if one must be used, then use the best possible option for that specific application.

Paul, even though I'm not a technical person, I understand what you mean by this:


Firstly, quoting noise levels of –120 dB to –150 dB means little without qualifying the measurement spectrum and operating conditions for both the regulation and the load.


Which is why that judging equipment solely by 'the numbers' (measurements), without looking at the bigger picture, as it were, doesn't tell the FULL story! Simon, are you paying attention? ;)

This is why it's completely pointless buying (or judging) equipment, solely on technical specifications, unless said specifications fully explain their relevance to the design(s) in question - and you *really* understand what the information given means. The fact is, the vast majority of end users of products will not have the required technical knowledge to make proper sense of the 'numbers'.

Therefore, that's why anyone without the level of technical knowledge, of the likes of Anthony or Paul, should ignore the spec, or at least don't treat it as an 'absolute', and always use their ears as the final arbiter, or otherwise be prepared to arrive at erroneous conclusions, based on only being in possession of part of the story. As they say, a little knowledge is dangerous...!


The cascaded low noise regulation combined with the SR7 used on your SL1210 easily achieves supply line rejection exceeding this well past 100 KHz, and does not run out of regulation capability until it gets well past 100 MHz. Upgrading the mains transformer is clearly audible even with this power supply as you are well aware.


Indeed. Even before your regulator mods were fitted to my T/T, upgrading from the SR5, to the SR7, with its massively bigger mains transformer, caused a notable improvement in sound quality, which proves that, with linear PSUs, one can never have too big a transformer (or transformers), providing of course that they are also of the requisite quality. That last bit is very important, as in that respect, not all transformers are equal.

In my experience, the better the mains transformer is to start with in a PSU, the better the end result will be, regardless of how good the regs are.

This would therefore suggest that even though, superficially, the M-DAC, due to its specific design, should not benefit from being fed by a bigger, better quality mains transformer, that it would not necessarily be the case if it were put into practice...


I do not have information about the specification of the Mdac external supply but if anyone can provide the specification It is easy enough to design a custom high performance supply to replace it. This will prove the point one way or the other.


Well, now that Simon has provided the circuit diagram, why not give it a go? In my opinion, it would be a valuable addition to your product range and would also prove to people that the PSU arrangement inside (and outside) of the M-DAC, good though it may be, is not yet the definitive article :)

Marco.

[realysm42]

Not long until I receive this to demo now; expect pictures and full write up...

[bobbasrah]

Not long until I receive this to demo now; expect pictures and full write up...

Excellent Martin, some facts are much appreciated...:D

[realysm42]

Excellent Martin, some facts are much appreciated...:D

I'm building suspense :eyebrows:

I'll say no more until I've actually received and reviewed it :)

[MCRU]

Well I may not want to let it go as I am "burning" it in for you currently :ner:

I have these in the house currently:-

M-DAC
R-DAC
Bushmaster
Furutech Esprit
Teac just arrived (selling off 1/2 price at Richersounds)

Luckily I have enough PSU's to suit them all and the differences are quite amazing when all are played through the same source (SB Touch with enhanced digital output), I won't say which is best or worst as I am still evaluating and you can be the judge on the effect of the M-DAC PSU.

[bobbasrah]

I'm building suspense :eyebrows:

I'll say no more until I've actually received and reviewed it :)

Not suspense necessarily, rather more a large pile of....conjecture....;)

Certainly sounds quite intriguing what David's supplied unit will bring to the party given the existing performance... :)

[realysm42]

Well I've not said anythign regarding it's performance yet, so I'm not sure about conjecture; I'm excited to see what it's capable of though, that's for sure!

[bobbasrah]

Well I've not said anythign regarding it's performance yet, so I'm not sure about conjecture; I'm excited to see what it's capable of though, that's for sure!

Nothing you said at all mate.... ;)

[realysm42]

Sure.

[Reid Malenfant]

Here the internal psu board schematic

http://i50.tinypic.com/2wn2yxt.jpg

I'd hardly call that or should I say those - precision regulators :eyebrows:

Using a P channel device in a positive regulator & an N channel device in a negative regulator means that they will be slow in comparison to a regulator that acts like a source follower.

So they already have compromised regulation at higher frequencies & that's before going into any more detail.

That can very easily be bettered :)

[sq225917]

Those are just the pre-regs, there's a couple more local regs on the main pcb.

http://farm8.staticflickr.com/7024/6800655209_2d3d1c1068_b.jpg

The schematic you're looking at if for the small pcb along the RHS edge.

[Reid Malenfant]

Ah, so I see :) Yes, they'd help :eyebrows:

[sq225917]

I was speaking to John about it over the weekend and he's got a few more things in mind.

He's moved the kelvin sense point for one of the regs, and he's implemented a slightly revised design around some higher bandwidth op-amps with lower noise that he reckons pep things up a bit as well. There's a couple of further mods to two of the regs on the main pcb, different decoupling scheme and local gnd path. He's going to offer them as a package for people who currently have Mdacs.

He's going to ship a modded pcb over in a the next week or so and I'll do some back to backs. With the mdac removed from it's casework it should only take 20 seconds or so to swap main boards over which isn't going to be much worse than swapping IC's over between two dacs. Should be decent enough for some AB stuff.

Be interesting to see how much closer it comes to the Weiss.

[MCRU]

Encourage punters to open up their DAC and do some soldering? Is that wise? Can't see it working myself. I like the DAC as it is, after all it's what you feed the DAC that really counts.

[sq225917]

I'm not encouraging anyone to solder anything. I'm simply stating that John has some mods. He's actually performing them free of charge for a few lucky customers, though I assume he'll start charging at some point.

Yeh its what you feed it that counts- the music.

[realysm42]

Got this in my system now, review coming very shortly, once I've stopped listening to tunes :)

[realysm42]

So, here it is:

http://i135.photobucket.com/albums/q153/realysm42/Image1256-1.jpg

I plugged this in yesterday, after twenty seconds I came to the conclusion it's never going back!

Build quality is good. Nice solid case, four rubber feet. The main thing that used to concern me badly with the M-dac was the power connector, from the unit to the dac itself, flismy and unconvincing fit. I was never happy about it (could ruin the whole thing and I worried you can break it by looking at it too hard). This has been remedied by a much chunkier plug that fits convincingly now:

http://i135.photobucket.com/albums/q153/realysm42/Image1259.jpg

Excuse the horrible quality photo, M-dac owners however will appreciate how much better this thing looks. It's a firm and convincing fit, a big weight off of my mind.

You might have read that I got some new isolation platforms not so long ago, they gave me a soundstage like I've never heard before. Adding this to the chain immediately improved this substantially!

All of a sudden my M-dac has control over the entire sound spectrum like never before.

Bass is deeper, but not overblown (drum and bass never sounded so damn good). It hits with such slam now! I didn't realise I was missing this until I heard the Lampizator a few days before, it hurt, losing that again... I feel happy to have it back and know it's going nowhere :D

Vocals and mids are to die for, they've always been something I've loved about my system, I worry that it can be taken too far, in this case, clarity has improved without sounding clinical, everything is so musical!

The top end is clean. Some music I have has analogue flaws added to the sound. Hisses crackles and pops sound so natural and organic. Cymbals decay properly, with the most delicate shimmer I didn't know my system could deliver.

I've not got a bad thing to say about this upgrade, every aspect of the sound has improved dramatically. I've bought a bespoke granite platform for it and some RDC cones to ensure it's properly isolated. David provided a no.27 cable to test it with. Great build quality, that's not going back either.

Thanks for letting me test this out for you David; I wholeheartedly recommend this unit to anyone owning an M-dac. My brother does too, he's extremely jealous of me now :D

[sq225917]

Is it an AC supply ?

[realysm42]

No idea, sir!

[lurcher]

No, +15v, -15v, nothing magic, just very very quiet DC with excess capacity.

[sq225917]

Very nice.

[realysm42]

Simon, you got the Soveriegn upgrade in your M-dac?

It's like speaking to a f*cking brick wall on that other forum, I either get ignored or told to trawl entire threads to find things...

[steviej233]

I am extremly jealous now! :eyebrows: The M-Dac is a fantastic little unit and if i had a spare £600 i would buy one in a heart beat, but knowing what this new PSU does for it i wouldn't feel right about not adding it in to the equation.

Martin's system has come along leaps and bounds and it has been a real eye opener as to what decent cables and PSU's can achieve. :)

[Mr Perceptive]

Although the MDAC is very popular, the addition of the PSU takes it into another ballpark with other competing DACS. At the MDAC+PSU level, one should audition the Wired4Sound DAC2 and the Leema Elements DAC as well. All IMHO of course!

[realysm42]

Fair enough, I won't be auditioning anything for some time, personally :)

[sq225917]

Martin, i haven't had any mods done to it. I sent John 20 op-amps last week for him to play with upgrading other people's units, I'm in no rush- what with having the Weiss.

[realysm42]

Cool, were they all the same or a selection to help him make a choice on the best one for the unit?

[Sovereign]

Simon, you got the Soveriegn upgrade in your M-dac?

It's like speaking to a f*cking brick wall on that other forum, I either get ignored or told to trawl entire threads to find things...

I'm not sure what John did to my DAC, I have it on an email somewhere, I think he upgraded the power supply around the chip or something.
It definitely sounds more involving, clear and dynamic.

[realysm42]

You're the guy the upgrade is named after?

I've tried repeatedly to speak to JW himself and ask other people about what needs to be done to make it happen or if it's even an option to wider audiences; it's like I'm not there lol!

[sq225917]

I'm not sure if it was the AVCC or the VDD voltage rail that he modded, I think it was the 1.2v rail for the VDD. Either way he's implemented a better regulator and moved the Kelvin sense point for the regulator so that it isn't polluted by the gnd plane noise.

The op-amps I sent were all the same, so that eh could mod a few test units for people. The mod isn't something you could do for yourself, not on a 4 layer smt board. I wouldn't try to do it myself and I've built loads of smt stuff.

[Sovereign]

You're the guy the upgrade is named after?

I've tried repeatedly to speak to JW himself and ask other people about what needs to be done to make it happen or if it's even an option to wider audiences; it's like I'm not there lol!

I'll see if i can dig out the email from John for you. I haven't been back to the thread since I made my post of my findings of what John did to my DAC, I find the MDAC threads a wee bit tedious.
I'll see if I can get the email up here tomorrow for you.

[Sovereign]

Cant seem to find what John W did to my DAC on email, he must have let me know what he did on one of the MAMOTH threads, sorry....good luck!

[realysm42]

Cheers for looking dude!

[Mileend]

This looks a great buy for MDAC owners and I think would be hard to DIY build for less money given the 3 rail requirement. Is it a 2 reg design as for Nick's other power supplies ? Also I beleive John Westlake will be making a new PCB board upgrade for the MDAC to give dual mono outputs which means 2 more power rails. Can the unit be upgraded to 5 rails ?
Thanks Mervyn

[realysm42]

Interesting questions, Mervyn. Anyone know?!

[lurcher]

This looks a great buy for MDAC owners and I think would be hard to DIY build for less money given the 3 rail requirement. Is it a 2 reg design as for Nick's other power supplies ? Also I beleive John Westlake will be making a new PCB board upgrade for the MDAC to give dual mono outputs which means 2 more power rails. Can the unit be upgraded to 5 rails ?
Thanks Mervyn

Yep, Its the same topology as the two box regs, dual distinct regulators, just in the one box this time, with the addition of -ve versions. And yes, there is no reason why additional second regulator stages could not be added giving as many distinct rails as needed. As you say, not that much point with the current power board in the mdac, but who knows.

The point of it (in this case) is not so much about regulation, as the dac itself has that covered, but its main goal is to remove the injection of noise into the main box that the standard power supply does.

[Mileend]

Many thanks for that Nick. Yes John Westlake says the on board regs are very low noise and his latest tweaks seem to improve this further. The standard supplied adaptor is AC so any dc replacement should theoretically proviide some sonic improvement.