Switching to linear...

Docfoster

This discussion was created from comments split from: latest mac mini tested on the Hi Fi world site.

Docfoster

You're an expert Si, so I'd be interested to know of your thoughts on any audible or measurable differences (or not) between switching power supplies and linear ones when it comes to computer audio.

sq225917

I wouldn't say expert, far from it, but I know how to use a scope and what to look for.

You have to take each case individually.

1. Sound card outputting bits over electrical spdif. I'd look at the eye pattern and I'd check that the files are bit perfect under all conditions. I'd be interested in noise coupling as well via the signal and earth of the 75ohm spdif cable. If these look the same and measure the same between smps and a linear one then I wouldn't expect to hear differences, nor would I sit down and search for them. The assumption is that jitter is adequately dealt with.

2. Soundcard outputting optical, I'd do a bit perfect test and leave it at that, jitter is a dac issue in this case.

3. If you were taking analogue outs from a soundcard then it's more difficult, I'd have to rely on some form of null test as I'm not able to do any distortion analysis. I'd probably look inside the pc and scope out the power lines for the dac chip and any active devices in the output stage, and real variance would likely be audible it depends on the power supply rejection ratio of the sound card.

4. Streaming, any tcp/ip asynchronous protocol. Assuming the dac is competently designed jitter is a none issue here. Really all you are interested in is coupling noise via ethernet, scope the signal and ground lines, though honestly ethernet is galvanically decoupled at both ends, its part of the spec, so it would have have to be a serious POS design to miss this.

I'm limited in that my scope doesn't have huge bandwidth and you really need to be looking high up to find some of these effects. The one thing not to discount however is that most of us have hifi that wasn't necessarily designed to deal with smps line noise- it simply didn't exist 20+ years ago so resisting it wasn't part of the specification of most bits of hifi at the time. The older the hifi the more likely to be polluted by having a smps on the mains. That said the standards for commercial devices these days are pretty high and any CE approved smps has to be pretty damn clean. Of course some are MUCH worse than others.

All things being equal I'll take a full wave rectified, snubbed and choke regulated power supply over anything else where possible- it's not always possible of course.

Alan

I would have though that a computer has so many power supplies and regulators inside it (probably all of the switching variety), that the original power supply makes little difference? With so much potential for power supply pollution on the output (be it USB or S/PDIF), perhaps a reasonable argument for a really stripped out, over simplified and over specified music computer could be made.

However, if a bog standard, two-a-penny, mass-produced-to-a-price box of capacitors and chips from Apple has nothing problematic on it's output in spite of crummy internal power supplies and a decidedly un-audiophile main PS, the assumption has to be that Apple engineered their computer to deal effectively with these power supply 'issues'.

Please note I am not saying anything definitive here, I lack the technical nous to understand all the issues at play - but so, it seems, do people designing their way out of such 'problems'. It was postulated after all, on AudioChews sometime ago that a MACMini puts out 'a load' * of noise on it's output.

I am also not saying that such carefully designed equipment doesn't sound better either as I have not heard it. It could be an interesting comparison.


* Technical 'rounding off' by myself as I can't remember the figure, nor be bothered to check.

Docfoster

I wouldn't say expert, far from it,...

@-)

Your post suggests that relatively speaking, here, you are.

It's interesting to read your breaking down of the different stages (thanks), which I think I follow, in concept if not in detail.

I suppose I asked because in my chats with Jason re. his NVA TFS he spoke about his confidence in the ability of a good linear PSU to improve the performance of his product over the bog standard switching supply that comes with the mother board he uses.
I don't know if the TFS PSU has "full wave rectified, snubbed and choke regulated". I do think it has separate transformers for the ASUS mother board and the Xonar DAC (certainly it has 2 separate DC feeds that go into the TFS). Jason did give me the bog standard switching power supply for the TFS, so I suppose I should make the effort to compare it with the one he makes himself. Would you predict differences in measurements? I don't own a scope - What differences, or not, would you predict that I would hear?

sq225917

Try them both and see what you can hear, then do it blind for good measure. If my dac wasn't immune to the differences between two adequate power supplies powering a computer I'd want a better dac, not waste money on chasing the minutia of PS design within the source. A good dac should buffer data to remove jitter and be adequate designed to reject signal source pollution.

Putting it in a box with noisy smps, rfi emiters/receivers doesn't strike me as a good decision. I'd love to hear some of the feted servers vs my dac and shonky old macbook just to see if there is any mileage in them.

Docfoster

Thanks Si,
I'll give it a go over the Xmas break! :-)

uglymusic

Don't forget to order a pizza cutter in plenty of time :-)

Brain_Dead

I found SMPSU on linear amps did perform badly, the worst being Flyback the best was a full resonant mode with its clock frequency locked to the clock in the TriPath chip I used.(Class T)

The nice thing about full bridge resonant is you can run a 1KW psu at 1MHz way past audio.

I also found standard 3 post linear design where OK on slow power amp but push-pull linear power supply was much much better.

Check out http://www.unitrode.com/products/apps_prt.htm

and

http://sound.westhost.com/projects.htm

for more info.

Alan

What exactly is a push-pull design please? (bearing in mind I am a total technological dunce)

I have seen that expression used to describe a certain type of amplifier but never really knew what it it meant. I assume it has something to do with A/C?

Brain_Dead

Basically it is a big DC Amp but instead of a signal i.e. music it is a DC level, the nice thing they recover quickly from under and over shots when used on a dynamic load i.e. a large audio amp with a inductive and capacitive load like a speaker. If made correctly they also reduce hum and ripple very well and survive rapid  input voltage changes. But they do get warm, well very warm. You can make them run in A - D easily .

Try one out!

Alan

Would love to! So I guess this is one of those amps that uses juice all the time, not just when playing music?

Brain_Dead

Not always.

Brain_Dead

Just use a PP9 no transformer no nasty SMPSU and no other comments!

Alan

A PP9 is a 'bactery' - innit? You can use one in an amplifier?

Brain_Dead

PP9 is Plutonium Powered number nine, you never needing lights is a added bonus, also kills all none gems.

The problem with 'bactery' they produce noise as the chemical destroys the electrodes and on a MC pre amp that can be a problem.

There was a joke design a long time ago in Elecktor using carbon high value resistor and diodes to produce power from the noise produced. 

But using a Peltier Diode Cells and liquid Helium is better, try it Alan, it will also keep the beer cool.

Tesla also made a Earth Battery using Cu and Zn or Fe in clay soil that works well to. I did it with a old circuit board and corrugated iron sheets.

Have a Jolly New Year Alan, I am home alone again as wife works. But the music is loud and mine.

Alan

Cool - you can power the hifi from a green power plant in the garden - the future is here! - or was, however many years ago that happened to be.

I hope you feel good again Col? Played lots of Les Dawson over the hols? I also had some quiet time on Monday without family, and played lots of (loud) music whilst decorating the lounge. I played a lot of Art Pepper and then Oli Brown, a young rocker in the Hendrix mold. Your cables helped make it a most enjoyable session indeed, made me want to paint something else.

Brain_Dead

I had better have the cables back if it makes you want to work you should relax.

The earth bats are running lights in my shed.

The Peltier diodes do and are still be used by a couple of Uni's and on the Space Station on the back of the  solar panels.

)

Look out for the review on the new power cord in the new year. And the other electronic stuff we are doing, it going to be fun year for me making new things, LDR Pot done just waiting for PCB,s

Docfoster

Re. The linear v switching psus on the TFS I hope to get some listening time tomorrow...!

Docfoster

I had better have the cables back if it makes you want to work you should relax.

The earth bats are running lights in my shed.

The Peltier diodes do and are still be used by a couple of Uni's and on the Space Station on the back of the  solar panels.

)

Look out for the review on the new power cord in the new year. And the other electronic stuff we are doing, it going to be fun year for me making new things, LDR Pot done just waiting for PCB,s

Sounds exciting! :-)

Docfoster

Had a chance to try this today.
Changing psus was a faf, but yielded interesting preluminary results.
I did the comparisons sighted, the pluggings and unpluggings were too fiddly and risky to entrust to any member of my family. I'd like to repeat blind in future with a more trusty pair of hands to help.
In short the NVA's psu seemed significantly better. The switching Asus psu sounded less sparkly, vibrant and real. The NVA psu just cut a sharp edge around all the sounds - better dynamics and imaging. Just more "NVA" really.
The difference between the 2 psus was more than I had consciously expected, given that it was only the mother board's psu that was being changed (the sound card was powered by the NVA psu throughout). But what do I know about computer audio.
I feel more listening investigation is required, but so far, so good. :-)

Brain_Dead

Had a chance to try this today.
Changing psus was a faf, but yielded interesting preluminary results.
I did the comparisons sighted, the pluggings and unpluggings were too fiddly and risky to entrust to any member of my family. I'd like to repeat blind in future with a more trusty pair of hands to help.
In short the NVA's psu seemed significantly better. The switching Asus psu sounded less sparkly, vibrant and real. The NVA psu just cut a sharp edge around all the sounds - better dynamics and imaging. Just more "NVA" really.
The difference between the 2 psus was more than I had consciously expected, given that it was only the mother board's psu that was being changed (the sound card was powered by the NVA psu throughout). But what do I know about computer audio.
I feel more listening investigation is required, but so far, so good. :-)

Is the NVA a regulated linear PSU? Often the SMPSU are made cheap and are flyback type, they do tend to be slow on dynamic loads, and they love constant loads. Could this be due to the ASUS reacting far to slow?

Docfoster

Tbh Colin, I don't know.
I'll try to find out.

Alan

Look out for the review on the new power cord in the new year. And the other electronic stuff we are doing, it going to be fun year for me making new things, LDR Pot done just waiting for PCB,s

That's quite exciting Col, comparing it to my Lightspeed will be very interesting.

The Lightspeed has its own power supply, so the Linear unit I used with my old EVA is surplus to requirements now. It really did make a difference before, but isn't compatible with the lightspeed. I was wondering if I should sell it, or will it be OK with what you are making Col? Its a regulated 12v supply, detailed here.

Alan

I think branching out beyond cabling will work very well for you Col, you have a lot to offer the industry.



It must be contagious, as John Westlake has finished his tenure working
for IAG/Audiolab in China and is setting up independently as LakeWest -
continuing what he did with Audiolab and also making some amps (which he
reckons will be quite unique).



Are we seeing a reconnaissance of the British HiFi industry? (maybe this post should be a thread on its own?)

Brain_Dead

I use linear PSU in all the linear product I make unless there is constant load like a massive A Class then I use a Resonate PSU design that will produce about 1KW just so I can carry it. The 20W we have just finished is about 25KG and that is enough at this time for me, and the Phono amp is only 3KG it going for review this week.

I hope they like its.

A mixture or hybrid is also very good, but I love big really  big caps with large transformers.

I can do a nice PCB with +7 to 18V @ a max of 10A if you like but you will have to make it, getting busy here.

Linear to a Buck Converter is also very good try the Simple Switcher from National or the 10A Linear Technology beasties I used them in a Naim project very good.LTM4600

uglymusic

I think branching out beyond cabling will work very well for you Col, you have a lot to offer the industry.



It must be contagious, as John Westlake has finished his tenure working
for IAG/Audiolab in China and is setting up independently as LakeWest -
continuing what he did with Audiolab and also making some amps (which he
reckons will be quite unique).



Are we seeing a reconnaissance of the British HiFi industry? (maybe this post should be a thread on its own?)

I agree, but I'm not sure how many of the posts should be unravelled from this thread.

Brain_Dead

At this moment I may just give it all up and hide.

Docfoster

Col,
Forgot to post earlier. The NVA TFS PSU is regulated.


EDIT: Now I'm not sure if it is or if it isn't. Richard Dunn (of NVA) has asked me to ask you to ask him directly.

ItemAudio

I've never known anyone swap a single switched rail for a linear one on a transport that didn't hear an immediate and obvious improvement. In an ideal world, in the ideal computer transport, there would be no primary generation of 12V, 5V or 3.3V rails with anything less than a sub-1mV ripple linear or equivalent supply.

True, there's a lot of on-board regulation it's difficult to bypass, but the closer you solve the problem to source, the better the result. We've never sold a computer with a primary switching supply: the closest we come is a recommendation to use a yellow-series Pico PSU with a 12V bench supply (Hi Figlet!).

The problem is that it's traditionally expensive and bulky to have three large linear supplies attached to your PC. But it's very definitely worth it.