Suggestions for PSU improvements

[ChaoZ]

Guys,

Is there any way whereby we can improve the PSU's performance?

I dont think we can do much regarding the pcb design.

We can replace almost everything which are not SMT.

Other the the usual caps replacements with higher tolerence/capacitance etc., what can we do to further enhance its performance?

Some stuffs which i thought of :-
- better noise filtration (emi, spikes, hum etc)
- voltage calibration to every power rails
- additional circuitary to fan control?
- additional protection circuit (varistors etc)
- additional PFC unit (am i dreaming?)

What do you think?

[Super Nade]

Re: Suggestions for PSU improvements

You can replace the complete secondary side Pi Filter if you choose. Perhaps somebody here who has more knowledge of SMPS magnetics than I have can point you to the right vendor and suggest a proper core material. Note that you can't randomly mix and match the "best of the best" There are elaborate calculations involved when you try replacing magnetics.

[severach]

Re: Suggestions for PSU improvements

>voltage calibration to every power rails

Wouldn't that require 4 primary drivers and 4 coils. One set for each 3.3, 5, and 12, and another tiny set to produce -12. That would be one big power supply.

[999999999]

Re: Suggestions for PSU improvements

You're better off buying a better PSU than trying to replace everything in a budget PSU.

While it may use crap capacitors, and installing better caps may prolong it's life, often there is not enough room for substantial capacitance increase. Replacing magnetics would be more useful if you could also upgrade the transformer to a larger size, but in doing so it would also then mean you need further cap upgrades to strike a balance.

If it doesn't have adequate AC noise filtration yet it is another sign it's a junk PSU not worth upgrading, unless you happened to have the spare parts cannibalized from something else so there was no cost and minimal time spent.

Anothe improvement is often to replace a low quality fan with one more reliable.

In general, beyond caps if there are a lot of other areas you don't like in a PSU, it's better to start out with one professionally engineered better, instead of spending hours changing and testing something starting out worth less than the time and parts to upgrade it.

[linuxguru]

Re: Suggestions for PSU improvements

Yup, start with a well-designed PSU and upgrade it if necessary. A good starting point is one of the 250W HiPros that caused so many problems to Dell - there are a lot of dirt-cheap ones with bad caps floating around, it works great with just a recap, and many of the caps have 12.5mm footprints.

[gonzo0815]

Re: Suggestions for PSU improvements

The most interesting upgrade would be a synchron rectification circuit with a mosfet transistor.
This would increase efficiency and would reduce the heat generation significant.
But it isn`t that easy to do, but sure it is doable.
Fro further information i think this could be interesting:

* Carsten, "High Power SMPS Require Intrinsic Reliability", PCI '81 Proceedings Conference, Sep. 14-17.
* Kagan, et al., "Improving Power Supply Efficiency With MOSFET Synchronous Rectifiers", Proceeding of Powercon 9, Ninth International Solid-State Power Electronics conference, pp. 1-5, Jul. 13-15, 1982.
* Jitaru, "Constant Frequency, Forward Converter With Resonant Transition", High Frequency Power Conversion 1991 Conference, Proceedings 1, pp. 282-292, Jun. 1991.
* Murakami, et al., "A Simple and Efficient Synchronous Rectifier for Forward DC-DC Converters", Apec '93, Eighth Annual Applied Power Electronics Conference and Exposition, pp. 463-468, Mar. 7-11, 1993.
* Acker, et al., "Current-Controlled Synchronous Rectification", Ninth Annual Applied Power Electronics Conference and Exposition, vol. 1, pp. 185-191, Feb. 13-17, 1994.
* Jitaru, et al., "High Efficiency DC-DC Converter", IEEE, Apec '94, ninth Annual Applied Power Electronics Conference and Exposition, vol. 2, pp. 638-644, Feb. 13-17, 1994.
* Alexander et al., "MOSFET's Move in On Low Voltage Rectification", Applications Handbook, Siliconix Technology Article, Siliconix Inc., pp. 569-580, 1984; No Month.
* Tarter, "Principles of Solid-State Power Conversion", pp. 544-547, 1985, No Month.
* Tabisz, et al., "A MOSFET Resonant Synchronous Rectifier For High Frequency DC/DC Converters", PESC '90 Record vol. II, 21 Annual IEEE Power Electronics Specialist Conference, pp. 769-779, 1990, No Month.
* Carsten, et al., "Design Tricks And Tribulations at High Conversion Frequencies", HFPC Apr. 1987 Proceedings, pp. 139-152.
* Blanc, "Practical Application Of MOSFET Synchronous Rectifiers", Intelec '91, pp. 495-501, (Nov. 1991).


Apart from that, i think a good high quality fan, good caps, and may be an improved line over voltage protection with either some SIL or bigger varistor is all what you need (but mostly at low power PSU`s, most ATX PSU`s do already have some varistors).

[gonzo0815]

Re: Suggestions for PSU improvements

It seems not that hard to get it working with this new chip :

[PeteS in CA]

Re: Suggestions for PSU improvements

In terms of what would be practical if you aren't a P/S engineer, probably the best you can do would be to:

* replace junky O/P caps with same or similar value Nichicon PW, UCC LXZ, Panasonic FC or Rubycon ZL series caps;

* replace the case fan with a high airflow, high-quality, ball-bearing fan (yes, this increases noise, but heat kills electronics);

* replace the large I/P lytics with brand-name (Nichicon, Panasonic, Rubycon or UCC), higher value parts (this should reduce line frequency ripple at the O/P and improve line drop-out tolerance).

If you're a bit technoid and are patient and careful, you can improve the O/P inductor some. Typically the core is a powdered iron core, which can deteriorate if it is run at high temp. If the core is Micrometals -26 material, rewinding the inductor (same number of turns, same arrangement, same wire gauge) on a Micrometals -56 material core will result in lower core losses (efficiency improvement, less heat, surrounding parts should be cooler, inductor core should last longer). Micrometals also has 200C rated cores, which should improve core life, though I'm not sure how the core losses compare to -56 material.

[Per Hansson]

Re: Suggestions for PSU improvements

PeteS in CA; Is it safe to replace the I/P cap(s) with larger variants?

I mean for most standard PSU's it wont affect the circuit in any negative way?

So this way we get better protection from line swags/brownouts?
Like doing something crazy like going from 330uF to 1000uF if we can find one that fits and retains the ESR values etc it is safe to do?

[gonzo0815]

Re: Suggestions for PSU improvements

What are the symptoms for an fried o/p inductor core?

Increasing the i/P caps should not be a problem, as long as inrush current considerations and the rectifier can handle this.

From my experience, it does not hurt, but the size does not allow to increase it too much.

[PeteS in CA]

Re: Suggestions for PSU improvements

PeteS in CA; Is it safe to replace the I/P cap(s) with larger variants?

Basically, yes. If it is a larger can (usually taller, since available PCB space will often prevent increases in diameter) you have to consider changes in airflow pattern, i.e. what is "downwind" from the caps.

What are the symptoms for an fried o/p inductor core?

I haven't seen it that I know of, but the core loses its magnetic properties. There may be some visible heat damage, but the O/P ripple will increase considerably, since the "core" is basically air. Here is Micrometals's article on the phenomenon, though it looks more at the phenomenon than at the consequences.

[Elitist]

Re: Suggestions for PSU improvements

* replace the case fan with a high airflow, high-quality, ball-bearing fan

I am no PSU engineer, Pete, and used to believe this strategy to be best. However, I was rudely disavowed about this some years ago.
Although a sleeve bearing fan has the advantage of quieter operation, they have become associated with shorter lifetimes due to rapid wear.
However, if you are advocating use of high quality fans, an excellent option, you might be pleasantly surprised by the MTBF of Papst' sleeved units. The rated life of these can usually be doubled, at least, by regular maintenance (simply cleaning and relubrication) - what else would a good engineer expect? However, if relube includes reassembly with a smear of MoS2-bearing grease to provide a measure of dry running protection, these fan units will exceed the lifetime of any other components they are associated with.
Furthermore, even a cursory acquaintance with basic fluid dynamics would validate Pete's proposal to replace the PSU fan with a higher throughput unit. I've been doing this for a decade or more with outstanding results. According to AMD's published datasheets, as well as aerodynamics, I also remove all other fans from the case, clear the airflow path of as many obstructions as possible and mount the PC case on a block of wood with a slight overhang at the front to improve the effectiveness of the front intake vent. I suppose, for perfection, this vent would be fed via trunking from external ambient in N Europe, Canada and Russia, maybe from the a/c output or from inside the fridge in more S. nations?

[999999999]

Re: Suggestions for PSU improvements

Papst fans are a great option for front or rear chassis fans, but their sleeve bearing models are unsuited for PSU use.

The problem with even high-quality sleeve bearing fans is the specific implementation in a PSU. Sleeve bearing fans are contraindicated in positions that are non-vertical, or in higher heat areas such as the exhaust on a PSU rather than intake. This effectively removes them from consideration on a std. design PSU if long life is important.

Also, I have no idea where Elitist gets the idea to remove all other fans from the case, as both AMD and Intel clearly document in their design guides to have a rear chassis exhaust fan, and with BTX, a front chassis intake fan.

There's no need to change the time-tested and proven design/results, only to use high quality parts instead of lowest-cost junk.

[Elitist]

Re: Suggestions for PSU improvements

All my fans run in the vertical plane. The actual thermal excursion at the fan bearing, esp. of a hi-rate fan, is exceedingly small. Moreover, a well-lubricated sleeve contributes less frictional loss, heat thermal dissipation, than a bearing assembly. One reason it is quieter.
As for fan specification, if one reads the full AMD document, it is clear that the secondary fan located adjacent the cpu, is an option/suggestion - not necessarily a recommendation. They clearly and correctly state that no intake fan should be fitted. Even the briefest reference to fluid flow dynamics provides the optimal solution - and it's always going to be K.I.S.S.
As for Intel/BX - I avoid them. That company never did anything right.

[999999999]

Re: Suggestions for PSU improvements

ALL major fan manufactuers concede (if they bother to provide details at all), that vertical orientation is crucial for sleeve bearing fans.

A well lubricated sleeve is not a permanent state, it does not matter if it has a minor % less frictional loss, rather what matter is if it remains well lubricated or whether the maintenace interval to keep it "well lubricated" is reasonable. For most uses, it is not reasonable unless vertically mounted.

If one reads AMD recommendations, it is clear this is the suggested config. to have a rear exhaust fan and that if there isn't one, you are entirely abandoning their recommendation and it is no longer in any way adhering to their (or Intel's) specs. Indeed a system can achieve acceptible cooling otherwise, but it would be completely wrong to make any mention of AMD or Intel if not clearly detailing what they spec'd, which was to have the rear exhaust fan in addition to PSU exhaust.

Intel and AMD do not clearly state no intake fan. They state a minimal first step before anything further is done. They state a route, of airflow, within which if the intake fan is not counter-productive, it can be implemented with success if deemed necessary.

[999999999]

Re: Suggestions for PSU improvements

In short, it would be ignorant to use a sleeve bearing fan in a PSU unless the only important parameter was lowest noise. In some systems, it is indeed the most important, and some will accept having to relube a fan to achieve this, but functionally speaking, there is no place in a PSU for a sleeve bearing fan unless one accepts a large burden to reduce noise a minor amount (since a large, low RPM ball-bearing fan does not have a large increase in noise, only at high RPM or a very poor/defective ball-bearing fan will have significant noise increase).

More to the point... The high quality dual ball bearing fans I've installed in PSU to replace junk fans, have lastest over 10 years, with no maintenace. and low noise. TO me this is ideal. If you have some other ideal, I wonder what it might be since the noise from these fans was due to turbulence if anything, not the bearing type.

[starfury1]

Re: Suggestions for PSU improvements

well I guess there is always room for some improvement

I think as has been more or less out lined above,
"don't try and turn a row boat into a battleship"

how far you can go is limited to your technically abilities and is it really worth it anyway I suppose

Caps I think are worth the time and trouble and a decent fan with a basically good design.
Possibly depending, change some components to beefer components.

Add ins such as fan control or extra circuit protection, if you can do it without compromising the circuit wouldn't hurt

I do remember OW did a article posted over a Jonny G's on a recap of a couple of psu's which showed worse results after recap with quality caps and I guess given a possible redesign of the output section could have been improved.

I think it was really neither here nor there but the big plus is the supplies should have a lot longer service life with the better quality caps used.
(the objective I suppose is to get them to last, not pop caps every year or so)

What I am really driving at is component selection are some what dependent on each others characteristics,so you have to be a little cautious.
(this is were really knowing what your doing helps...me still on training wheels in this regard)

Anyway while looking for something else.... found the below, and it reminded me of a post round here on a DVD PSU that worked fine with GP caps but wouldn't work properly with low ESR....
(can't recall the full story)

When better may be worse
ESR is usually something to be minimized in a capacitor. However, where the original design depended (probably by accident) on a certain ESR, this may not always be the case:

(From: Lee Dunbar (dunbar@unitrode.com).)

Substitutions of low ESR caps into circuits which had lousy caps is not always the good idea that it appears to be.... Caution is advised, as low ESR caps will not limit surge currents.

The circuits' series impedance drops (compare substituted capacitors ESR when new with the original capacitor's ESR was when it was a new capacitor), which, in turn, lets the surge magnitude rise, the higher currents destroy can semiconductors and other components.

I guess what the industry needs is a good capacitor cross reference guide for aluminum electrolytics!

"the higher currents can destroy semiconductors and other components."
(think thats how it was meant to read )
link here

me, I think changing them is a good idea regardless of the slight degradation you might get but its a point to bear in mind if things start popping in the psu I suppose (not that I remember reading of any round here doing that)

To your question Per, yeah Id wack in bigger I/P caps
but within reason or otherwise you may really have to look at beefing up a lot of the front end to handle it.
(you still possibly would have to to seek the result your after anyway)

With the older 120/240V switch type supplies the capacitive value is being wasted anyway as they are in series on 240V, although the heat side I guess might be an advantage...having it spread over 2 caps

Redesigning for 240V only operation might be an idea
You would have to put in proper safety features, whatever that might involve and the electrical legalities.... well I am not going there...cause I wouldn't have a clue.

I think if you are getting to the point were you wanting to redesign one your probably better off just buying a quality one to do the job

but there is room for some improvements and that I supposed depends on what you got to start with

Anyway just my thoughts on it

Cheers

[PCBONEZ]

Re: Suggestions for PSU improvements

I agree with starfury1 and some of the others.
Just buy a better one.

If your wiring has noisy or unstable AC to start with you may see some improvement by simply plugging it into a UPS that regulates the AC.

[starfury1]

Re: Suggestions for PSU improvements

just on fans the less (electrically speaking) complexity the better to my way of thinking
A dead fan is a cooked psu (mostly) so the less points of possible failure with it the better.

On the mechanical side, well you guys know better then I, you service them on a day in day out basis
but a little bit of maintenance can go a long way, problem is a lot of fans unfortunately dont get it so one that
longest lasting without any maintenance I guess is the go.
(If such a beast exists)

but thats just my opinion

[Elitist]

Re: Suggestions for PSU improvements

Mostly, a sleeve bearing that is wearing becomes noisy. A ball-bearing that is wearing is more likely to seize, resulting in expensive damage. Regular lubrication is a panacea in both cases. Using an MoS2 assembly compound is a precaution well worth the effort.

I'll ignore the implication that I am ignorant from anyone who clearly hasn't even opened the pages of an aeronautical design tome, but an apology would be appropriate. Bad manners tells its own story.