PDA

View Full Version : Off The Shelf Power Supply



fuzzyT
2005-05-23, 08:14
OK. All this talk of PS audio improvement has left me wanting to give
it a try. I do DIY to some extent, but I never have as much time to do
so as I would like.

This leads me to think about looking for an off-the-shelf linear 5VDC PSU.

I'm sure there are many out there, but I'm not sure what the
differentiators are. Would anyone care to chip in with advice on specs
or, even better, a particular unit which works well and has the desired
effect?

Here's one unit that looked like it might fit the bill:
SolaHeviDuty Model#: 83-05-230-3
<http://www.sola-hevi-duty.com/products/powersupplies/pdfs/LinearOEM.pdf>

Mouser has this unit listed at 64USD. Looks like it could be boxed up
with an IEC line voltage input and placed close to the SB. Thoughts?

--rt

Mike Hanson
2005-05-24, 04:41
The output specs don't look very promising, although the only way you would know for sure is to try it.

-=> Mike Hanson <=-

Andrew L. Weekes
2005-05-24, 16:14
Mouser has this unit listed at 64USD. Looks like it could be boxed up
with an IEC line voltage input and placed close to the SB. Thoughts?

The specs don't really tell you much that's useful for audio, but I would expect this to be very much quieter than the supplied SB2 PSU.

The downside of this one, ignoring the basic regulation specs, is by placing it close to the squeezebox you are placing an EI core transformer, with a significant stray magnetic field, close to the SB, which is not generally a good idea, all else being optimal.

I don't think you'd find an off-the-shelf solution that would allow these elements to be split though.

Based on the noisy behaviour of the SB2's supply, I would still expect a modest gain from it's use, even if more remote from the Squeezebox itself, which might be worth a try.

Andy.

JJZolx
2005-05-24, 16:58
The downside of this one, ignoring the basic regulation specs, is by placing it close to the squeezebox you are placing an EI core transformer, with a significant stray magnetic field, close to the SB, which is not generally a good idea, all else being optimal.

Excuse my ignorance, but isn't it commonplace to put power supplies of similar design within the confines of (even very high end) single box players with Transport/DAC as well as high end DACs? I know separate PS chassis are sometimes employed, in preamps as well, but it's fairly uncommon.
________
Kawasaki ZR750K (http://www.cyclechaos.com/wiki/Kawasaki_ZR750K)

Patrick Dixon
2005-05-25, 00:49
I think they tend to use toridal transformers, which have a different magnetic field. Troidal transformers have a field that radiates above and below, which is why most high end audio manufacturers recommend stacking their equipment with gaps between.

fuzzyT
2005-05-25, 08:03
Mike Hanson wrote:
> The output specs don't look very promising, although the only way you
> would know for sure is to try it.

OK. Which specs should I be paying attention to, and what values would
fall into the promising range.

Thanks,

--rt

Mike Hanson
2005-05-25, 08:42
The problem with basic specs is that they don't address all of the issues. For example, as Andrew mentioned in another thread, things like THD (total harmonic distortion) may look good, but the type of THD that is present may be particularly detrimental. Therefore, the only real test is to listen to it and see if it helps. Unfortunately, this type of experimentation can get rather costly. That's why people often trust a manufacturer's claims, because it takes the guesswork out of the equation.

-=> Mike Hanson <=-

fuzzyT
2005-05-25, 09:42
Mike Hanson wrote:
> The problem with basic specs is that they don't address all of the
> issues.

Fair enough.

My goal is just to identify a good-enough, easy-enough and cheap-enough
alternative to the DIY route. It makes sense to me that a PSU could
have positive sonic effects. If I knew I could spend $50-$100USD on a
replacement PSU and get a definite bump in sonic quality then I'd do it
in a heartbeat.

Ideally, this would take the form of an SB community tried and verified,
fully assembled off-the-shelf unit. For economics sake, it would be
great if this solution took the form of an easily available mass
produced item. Alternatively if a known-good kit were available, that'd
fill the bill as well.

I haven't seen anyone ring in with a specific recommendation, so I'm
just trying to drive the conversation in that direction a little bit.
Or to understand what characteristics are desirable so that I might make
a more informed choice if taking a shot at one of the available options.

So far, the characteristics mentioned seem to read like so:

1. Stable voltage under varying load demands
2. Good amount of current overhead
3. Low noise (from inputs or generated)
4. Low stray magnetics

I suspect I may simply have to wait for the DIY electronics design
hobbyists to sort it all out.

Thanks,

--rt

Andrew L. Weekes
2005-05-25, 11:38
The problem with PSU design for audio, in many cases, is people view it from a DC perspective only.

The specs for a PSU usually cover basic features like DC voltage stability, line / ripple rejection (i.e. how much noise on the input, is passed to the output), load regulation (i.e. how much the PSU output changes in response to a load change at the o/p). These are dead easy parameters to achieve and any basic electronics text will give you the fundaments you need.

Noise is often quoted and occasionally output impedance.

The problem, like many data sheets, is you have to scratch the surface to get to the important stuff. The parameters displayed in the headlines are the best case figures, usually at *DC*. At *AC* things can degrade VERY rapidly.

The thing is that for audio, the DC spec is of little relevance in most cases, beyond some basic specs. The thing you need to examine is the effect of the AC load on the PSU, i.e. it's response to the audio signals and other nastier (i.e. digital) stuff that can be impressed upon it.

You have the effect of the raw (unregulated) supply on the regulator, the load circuit on the regualtor, the AC signals impressed upon that PSU on the audio circuits etc. etc.

You won't find those specs in *any* PSU you can buy though, as far as I'm aware and even a manufacturers data sheet, for a regulator for example, won't give you the full picture, even if you read to the greater AC detail that is sometimes presented therein.

Couple this with the fact that the external components (like decoupling caps etc.) have a very fundamental effect on the operation of the circuit, especially, if like most regulators, the circuit uses feedback. You really have to do the donkey work yourself or find someone who already has.

I've no doubt at all know that there are significant benefits to be had from a more optimal external PSU, even if it isn't the 'perfect' solution, in an engineering sense. At least it's non-invasive, which is good from a warrantly perspective.

An interesting exercise is to actually listen to the AC content of the power rails - this can be very enlightening, sometimes you'll actually hear music on them, sometimes it's clean, sometimes it's distorted, what you hear might give you some clues about what's going on.

I use a simple instrumentation amp, with a true differential, AC-coupled input, to extract the AC content of the rails, this particular meaurement system can resolve signals down to -160dBv with an FFT analyser attached.

Feed it into an amp, and you can hear a lot of interesting stuff on a pair of headphones!

If you want to get numbers out, you need to put defined stimuli in, which isn't that easy, once you get to line rejection figures of 160dB+ and output impedances measured in microhms, it gets a challenge!

It's not easy stuff, you have to think of the PSU not as a DC circuit, but an AC one, like an audio amplifier.

Regards,

Andy.

fuzzyT
2005-05-25, 13:38
Thanks Andy, excellent post.

I love that you are blending rationalist with intuitionist approaches.
_Listening_ to the noise in the system! And then building on your
mental model of how the system is operating based on what you hear.
Beautiful.

It would appear that you are much more skilled and well equipped for the
'donkey work' than I. Happy to have you on board. Let us know how we
can help. Reviewers ears available.

Knowing very well that 'perfect' is unobtainable in any case I am simply
looking for 'noticably better'. And that in a subjective sense. Seems
a sensible goal.

I've no idea what your intentions are w/r/t an SB2 PSU project, but I do
hope that whatever you learn and implement will be shared in some way.

--rt

Andrew L. Weekes
2005-05-25, 13:57
Hi there,

Thanks for the positive response, I'd be glad to help in any manner possible, and am always ken for feedback, be it good or bad!

I've just built a squeezebox supply for someone, the details on some elements of it I can't reveal publicly yet, but can in time and it will be open and available for anyone.

In the meantime, the principle is to create a reasonably quiet mains supply, with a seperately boxed regulator unit, that sits close to the squeezebox itself.

The basic idea (if the image attaches) is shown here - the parts are all pretty much off the shelf. This was a proof of principle, which I'd like to develop into a simple, minimal wiring, PCB-based approach that anyone can build, safely, as I suspect this will be of interest to others not just me.

Andy.

Andrew L. Weekes
2005-05-25, 14:01
This bit would be the part I'd make into a single PCB - this is manually wired at present, which takes time and doesn't look so pretty ;)

fuzzyT
2005-05-25, 14:02
Andrew L.Weekes wrote:
> Would be a derivative of this unit, on a neat PCB...

Sweet.

These look entirely buildable, even by a ham-handed solderer such as myself.

--rt

Andrew L. Weekes
2005-05-25, 14:04
Would be a derivative of this unit, on a neat PCB.

GregD
2005-05-25, 14:31
Certainly be interested in one when you have them available.

Yannzola
2005-06-28, 17:03
Add me to the list as well... depending on cost, of course.

y.

corwin
2005-06-29, 04:00
As would I.

Aylwin
2005-06-30, 04:04
Add me to the list too!

Dave D
2005-07-03, 07:13
Hi Andy,

An interesting thread. It's implying that the SB2's power supply can be improved (as can anything, but where do we stop? ;)

I have a quick question:

Have you employed your empirical method (AC-coupling the PSU output, amplifying, and listening to the AC noise with headphones) to the SB2's own power supply (connected to the SB2), vs. your homebrew power supply (also connected to the SB2)? (Under the same conditions, with and without music, etc.)

- Dave




I've just built a squeezebox supply for someone, the details on some elements of it I can't reveal publicly yet, but can in time and it will be open and available for anyone.

Andy.

Dave D
2005-07-03, 07:17
And, of course, can you hear a difference in the output of the SB2? (Could you pass a blind test? ;)


Hi Andy,

An interesting thread. It's implying that the SB2's power supply can be improved (as can anything, but where do we stop? ;)

I have a quick question:

Have you employed your empirical method (AC-coupling the PSU output, amplifying, and listening to the AC noise with headphones) to the SB2's own power supply (connected to the SB2), vs. your homebrew power supply (also connected to the SB2)? (Under the same conditions, with and without music, etc.)

- Dave

rwh
2005-07-04, 18:45
Wanted to register my interest in your PCB kit for a DIYer, Andy. Thx!

Andrew L. Weekes
2005-07-05, 15:05
And, of course, can you hear a difference in the output of the SB2? (Could you pass a blind test? ;)

I'm confident I could, although I haven't tried a blind test. To me the SB" with the standard PSU is edgy and tiring after a while, whilst the linear PSU makes it far more listenable, but obviously lacking in comprison the the CD player.

I've just pulled a pair of super reg's from something, that I can easily convert to 13V or so, to try isolating the audio supplies further, as Sean has mentioned in a number of places. Tha aim being to feed in an external 13V to replace the internal mutliplier-derived supply from the display, although this isn't that noisy as it's quite easy to filter, as it contains little HF energy like many capacitor-only voltage mutlipliers.

This is the first line of attack, but I have reservations that the fundamental limitations lie in the clocks at present - the oscillators run from a switcher, not a linear PSU and the two Pierce reference oscillators are sharing a single chip, which further exacerbates the jitter problems, based on experience.

I really need to re-box it to make space and gain access to try and address these areas in order to evaluate the differences, I have a nice 1U 19" case to do this, just need to persaude the guys at work to do some CNC work for me ;)

I've also re-configured the output stage from a single-pole filter to a 2-pole Bessel, to try and help reduce the supersonic stuff at the output further, but I'd like to see more suppression - running a sweep on the SB2 you can see the aliasing products from the supersonic output folding back on the analyser - a Bessel was chosen as it has the best transient performance, but this is at the expense of ultimate attenuation, I'd need more poles ideally, which is impractical as-is.

TI recommend a 2-pole Butterworth, but I'm far from convinced this is adequate, given the performance of the DAC's own internal analogue filter.

Andy.

Robin Bowes
2005-07-05, 16:03
Andrew L.Weekes wrote:
> I've also re-configured the output stage from a single-pole filter to a
> 2-pole Bessel, to try and help reduce the supersonic stuff at the output
> further, but I'd like to see more suppression - running a sweep on the
> SB2 you can see the aliasing products from the supersonic output
> folding back on the analyser - a Bessel was chosen as it has the best
> transient performance, but this is at the expense of ultimate
> attenuation, I'd need more poles ideally, which is impractical as-is.
>
> TI recommend a 2-pole Butterworth, but I'm far from convinced this is
> adequate, given the performance of the DAC's own internal analogue
> filter.

Andy,

I'm interested in your work in this area.

I have a a couple of ART DI/O DACs, one of which I have been planning to
modify using several of your super-regs.

As part of the modification, I'm essentially going to throw away
everything but the digital boards and so need to build an output stage,
including the filter.

Have you got any references/links for output stage design?

Thanks,

R.

Dave D
2005-07-06, 05:24
Just curious: how low in frequency do you see these components and how far down (magnitude) from the fundamental (if you have that handy)? - Dave



- running a sweep on the SB2 you can see the aliasing products from the supersonic output folding back on the analyser

Andrew L. Weekes
2005-07-06, 14:33
Just curious: how low in frequency do you see these components and how far down (magnitude) from the fundamental (if you have that handy)?

I'll re-run the test when I get a moment, but from memory the products I was seeing were around -65dB just above 20k.

You can tell they are aliasing products (produced by the analyser, not the SB2 directly) since in response to a rising sweep, the resultant signals drop in frequency, which is classic aliasing behaviour. Thay can't be images, sinc ethe DAC is oversampling.

This is just with an audio band (20Hz to 20kHz) sweep too, I'll produce some wider band test files when I get a moment.

Andy.