Parts quality and measurable performance

[ginetto61]

Hi !
i have a very undefined question about design and construction of power supplies.
I know it can be a weird question.
I wonder how much of an impact on a cheap power supply performance can have replacing original parts with higher quality parts.
I am referring to measurable performance of course.
Could ripple go down for instance ? or instead it is completely dependent on the power supply design ?
This could very be my first and next diy project.
Just to start easy with something
Also because I could measure the outcome.
Thanks a lot indeed.
Kind regards, gino

 

[DF96]

Depends what you mean by "higher quality".

In a simple PSU ripple is almost entirely set by circuit design, especially capacitor value. ESR has a small effect too; buying very low ESR caps (which could be consider 'higher quality') will reduce ripple, but much cheaper to buy higher value caps instead or slightly alter the circuit. No point in buying low ESR caps, then having a poor grounding design!

In general, you can assume that ordinary decent commercial quality components are fine for audio. If you need special characteristics then this should emerge from the circuit design, not be pasted on afterwards as some sort of 'designer-approved' tweak.

 

[ginetto61]

Depends what you mean by "higher quality".

hi ! thanks a lot for the always kind and very valuable advice.
I try to elaborate better.
I am attaching a picture that I think can explain what I mean
Sorry for the bad quality
I have those peaks at lower Hz.
I was wondering if better caps or diodes, the passive parts, can help in smoothing those peaks out.
At first I would like to leave regulators untouched (I mean to keep if possible the original design. Changing it woul imply a complete rework of the unit).
Only intervening on passive components replacing them with others with better specs/different value.

In a simple PSU ripple is almost entirely set by circuit design, especially capacitor value. ESR has a small effect too; buying very low ESR caps (which could be consider 'higher quality') will reduce ripple, but much cheaper to buy higher value caps instead or slightly alter the circuit. No point in buying low ESR caps, then having a poor grounding design!
In general, you can assume that ordinary decent commercial quality components are fine for audio. If you need special characteristics then this should emerge from the circuit design, not be pasted on afterwards as some sort of 'designer-approved' tweak

Thank you very much indeed. This is indeed what I needed to know.
If I will increase uF and lower ESR (usually as you say a characteristic of higher grade parts) I should get some kind of benefit ripple wise ?
Yes ripple in the lower Hz is the main problem.
I have to say that the psu is brand new and with very few hours of work.
I will take it apart and see the parts used and list them.
Maybe also diodes could be replaced with something better I do not know.
Reading around there is a big trend of modders. Experts who take a commercial unit and with strategic interventions elevate their performance.
Thanks a lot sincerely again.
Kind regards, gino

P.S. I have decided to go with a linear power supply after seeing the internal of some cheap smps ... very disappointing.
I have to say that smps so called genuine from computer brands look much better built but still cheap. And usually ripple is about 1% of the Vout.
Instead for linear psu is about 20-50 times lower.

 

[DF96]

You have strong peaks at 50Hz and odd multiples of 50Hz. The peak at 100Hz is smaller. This is the opposite of what you should expect from a full wave PSU. A commercial PSU doing this might be considered faulty. A DIY PSU would be a design or build error.

Be aware that most 'modders' have no idea what they are doing. They have just learnt to solder and so do what they can: swap components.

 

[ginetto61]

You have strong peaks at 50Hz and odd multiples of 50Hz. The peak at 100Hz is smaller. This is the opposite of what you should expect from a full wave PSU.
A commercial PSU doing this might be considered faulty.
A DIY PSU would be a design or build error.
Be aware that most 'modders' have no idea what they are doing. They have just learnt to solder and so do what they can: swap components

Thanks a lot sincerely again. I will peep inside and take some pictures.
I have to understand that there is this design issue you mention.
I will take it apart completely.
I have seen already at Ikea a nice wood board that could work as a test bench.
I just need some 12V/1A ... but possibly clean
I will revert with pictures.
Thanks a lot again, gino

 

[marce]

You generalise again regarding power supplies, both linear and SMPS can be rubbish or they can be good...
A bad design and layout can only be cured by throwing it in the bin....
Also some supplies may have built in filtering again you could add this to an existing supply and improve its performance. I once steared someone in the direction of Picor/Vicor active filters and the like, for a fraction of the cost of an "audiophile PSU" he built a far quieter better specked PSU.
A y axis grid is always useful on a graph...

 

[ginetto61]

You generalise again regarding power supplies, both linear and SMPS can be rubbish or they can be good...
A bad design and layout can only be cured by throwing it in the bin....

Hi ! very true. I will open the box and see what is inside.
Nothing can cure a wrong design. But I assume that this unit is not wrong completely.
From what I have read transformer, caps and diodes selection/quality is very important in the overall performance of a power supply.
Some parts have an almost mythical status ... like Sikorel caps and Schottky and so on. And expensive.
But I have to see inside first.

Also some supplies may have built in filtering again you could add this to an existing supply and improve its performance.
I once steared someone in the direction of Picor/Vicor active filters and the like, for a fraction of the cost of an "audiophile PSU" he built a far quieter better specked PSU

Thanks a lot for the advice and it is indeed a very good solution.
Now the unit outputs 12.2 V. My need is 11-12V.
I have to check if there is a trimmer to increase a little the Vout of some Volt.
Then I could think of an additional regulation stage to suppress some of this noise.
To find a nice 12V/1A power solution is my present main goal.

A y axis grid is always useful on a graph ...

Yes you are right. But that could trigger some other discussion.
My main target is to smooth out and lower the low Hz noise at the 500 Hz level. To flatten out the curve.
Then I will be done.
I have to look for any internal trigger. I guess that in order to get 12V (but also 11V should be enough) a regulator needs at least 2-3V above that at the input ?
I like the idea immensely anyway. I first rough regulation and a finer one close to the circuit. Yes. It sounds very good indeed.
But first a look inside.
Thanks a lot again for the very interesting advice.
Kind regards, gino

 

[DF96]

From what I have read transformer, caps and diodes selection/quality is very important in the overall performance of a power supply.
Some parts have an almost mythical status ... like Sikorel caps and Schottky and so on. And expensive.

You may have been reading the wrong things. Right technology inside, yes; right brand name outside, not needed.

 

[ginetto61]

You may have been reading the wrong things.
Right technology inside, yes; right brand name outside, not needed.

Hi ! about the famous Siemens Sikorel i have read praises here also.
Now they are branded Epcos

http://www.diyaudio.com/forums/parts/6886-siemens-sikorel-caps-2.html

someone consider them the best caps for audio power supplies.
However i agree that there is nothing worse than a bad design and no parts replacement can make it right.
Kind regards, gino

 

[Andersonix]

To find a nice 12V/1A power solution is my present main goal.

How about 12V, 0.4A? ~$10.
Murata 7812SR:
http://power.murata.com/data/power/dms-78xxsr.pdf

 

[ginetto61]

How about 12V, 0.4A? ~$10.
Murata 7812SR:
http://power.murata.com/data/power/dms-78xxsr.pdf

Hi ! thanks a lot for the kind and interesting advice.
I am afraid is not enough.
My current "project" is to use a good usb soundcard that I have already to test different 12V power solution in terms of noise mainly.
I am so naïve to think that the lower the noise the better the sound
However this unit draws about 450 mA. But I have another one that draws about 800 mA. I would like to find a supply that can be universal in case I change interface.
From this the idea to look for 12V/1A power supplies.
I love the convenience of smps, but I understand that because of their switching mode they have high Hz noise.
At present I am playing with a mixed solution (mostly for fun) with 15VDC smps followed by different linear regulator to step down to 12VDC.
It is all not serious because, very very sadly, I cannot afford a very nice spectrum analyzer with which checking for wideband noise.
I love immensely lab instruments. They are like a cockpit in a spaceship.
A spectrum analyzer could show the noise spectra of different power solutions.
So for now I have to rely on a software called Arta with which I observe the noise floor of the soundcards up to 96kHz
Honestly I have to say that for a same soundcard the noise floor spectrum does non change too much changing power supply.
So maybe I am just worrying too much.
In the sense that the limiting factor can actually be the soundcard and not the power supply.
So I better performance are need better to look to a better soundcard
Thanks a lot again, gino

 

[Mooly]

I wonder how much of an impact on a cheap power supply performance can have replacing original parts with higher quality parts.

Out of the box and the answer is probably none (or very little).

Come back to it after 5 or 10 yrs of constant operation and compare it to an identical one that you fitted 'top quality' parts to and you will quite possibly get your answer

That 5 to 10 yrs may drop to 1 to 2 yrs for some poor examples.

 

[ginetto61]

Out of the box and the answer is probably none (or very little).
Come back to it after 5 or 10 yrs of constant operation and compare it to an identical one that you fitted 'top quality' parts to and you will quite possibly get your answer
That 5 to 10 yrs may drop to 1 to 2 yrs for some poor examples

Good Morning !
thanks a lot for the always kind and helpful advice.
Let's say that maybe it is more likely that reliability will benefit from using better parts.
I have no more doubts on the issue.
This is a field where people spend usually huge many.
I have seen caps priced like jewelry
Caps are very intriguing parts to me. Energy reservoirs.