Re : Whining PSU

[Guest]

scotoma ([email protected]) wrote in

what makes the capacitors on a PSU whine?

The whining PSU is a new Mercury 300W powering a
Gigabyte KT400 mobo, 40GB Samsung ATA100, AthlonXP
3000 "in a box" with supplied fan, 512MB DDR333,
nVidia Riva TNT2 M64 (lol), 1x case fan, 52x CDRom,
onboard sound, usb2 x 4. OS is windows XP Pro.

I've tried changing the new 300W PSU with an old and
known to be quiet same brand Mercury 300W PSU which used
to silently power a Chaintech Apogee mobo, AthlonXP 1800,
dual fan, 512MB DDR266, nVidia FX5200 128MB, 1x case fan,
200GB Maxtor ATA133, Pioneer 107, Plextor 48a, Yamaha SW1000XG
sound card, SBLive 5.1 sound card and 6x usb.
Surely the AthlonXP 1800 system (300W PSU) was drawing more
power than the AthlonXP 3000 system yet the PSU's are
squeeling like **** on the AthlonXP 3000 system????

Capacitors rarely whine noticeably, but bad capacitors can make coils,
including transformers, whine by causing the voltage regulation to
overcompensate. The power supply generates high frequency AC whose
duty cycle (amount of time it's on verses off) is varied to vary the
output voltage, but when capacitors are bad or the power supply is
simply overloaded, the regulation makes the duty cycle go too high,
causing very narrow pulses to be fed to the transformer and cause it
to whine. Since these are Mercury brand supplies, probably by Deer,
one of the worst manufacturers, and are rated for only 300W and you
have an XP3000+, I suspect they were simply overloaded, especially
their +5V and +3.3V outputs (the KT400 uses +5V to run the CPU). You
can't automatically assume that an XP1800+ system with an FX5200 draws
more power than an XP3000+ with a TNT2 M64, and most 300W power
supplies can't run an XP3000+. Think of getting a supply made by
Fortron-Source, such as those sold by www.directron.com or
www.newegg.com under several different brands. They're very good but
unusually inexpensive, and a 350W model should be able to run almost
anything.

 

[scotoma]

scotoma ([email protected]) wrote in



Capacitors rarely whine noticeably, but bad capacitors can make coils,
including transformers, whine by causing the voltage regulation to
overcompensate. The power supply generates high frequency AC whose
duty cycle (amount of time it's on verses off) is varied to vary the
output voltage, but when capacitors are bad or the power supply is
simply overloaded, the regulation makes the duty cycle go too high,
causing very narrow pulses to be fed to the transformer and cause it
to whine. Since these are Mercury brand supplies, probably by Deer,
one of the worst manufacturers, and are rated for only 300W and you
have an XP3000+, I suspect they were simply overloaded, especially
their +5V and +3.3V outputs (the KT400 uses +5V to run the CPU). You
can't automatically assume that an XP1800+ system with an FX5200 draws
more power than an XP3000+ with a TNT2 M64, and most 300W power
supplies can't run an XP3000+. Think of getting a supply made by
Fortron-Source, such as those sold by www.directron.com or
www.newegg.com under several different brands. They're very good but
unusually inexpensive, and a 350W model should be able to run almost
anything.

Hi do_not_spam_me

Thank you very much for your reply. )
I'll replace the PSU as suggested.
Thanks again

Scotoma

 

[Mantric]

Hi do_not_spam_me

Thank you very much for your reply. )
I'll replace the PSU as suggested.
Thanks again

Scotoma

With any cheapo psu there is the risk that it can't supply the current
without a struggle, bad regulation or a crap switching circuit can cause
the switching frequency to drift off once the psu is under load. Poorly
wound inductors can also cause noise, in either case, I would change the
psu for a better make.

 

[Mike Kay]

With any cheapo psu there is the risk that it can't supply the current
without a struggle, bad regulation or a crap switching circuit can cause
the switching frequency to drift off once the psu is under load. Poorly
wound inductors can also cause noise, in either case, I would change the
psu for a better make.

I must admit i dont understand this whole crap PSU thing.

surely any product sold should work as advertised otherwise it shouldnt
be on the market.

how can a 400watt psu be sold, when it isnt actually 400watt, isnt that
illegal? and yet it just seems to be accepted as ok in this instance.

 

[Chris Ryan]

I must admit i dont understand this whole crap PSU thing.

surely any product sold should work as advertised otherwise it shouldnt
be on the market.

how can a 400watt psu be sold, when it isnt actually 400watt, isnt that
illegal? and yet it just seems to be accepted as ok in this instance.

But its not just as simple as just being 400W. Different power rails will
have different power ratings and so depending on the loads placed on each
rail by a motherboard then the total power is going to change. And this is
going to change depending on your setup ie. more fans or hdds etc.

The only half decent analogy i can think of is when you buy a car that does
x miles/gallon. It doesnt mean you will actually get that efficiency all the
time. It means that is the best feasible value in the correct conditions.
Manufacturers are always fiddling with numbers to make their product appear
as great as possible

 

[Mike Kay]

But its not just as simple as just being 400W. Different power rails will
have different power ratings and so depending on the loads placed on each
rail by a motherboard then the total power is going to change. And this is
going to change depending on your setup ie. more fans or hdds etc.

The only half decent analogy i can think of is when you buy a car that does
x miles/gallon. It doesnt mean you will actually get that efficiency all the
time. It means that is the best feasible value in the correct conditions.
Manufacturers are always fiddling with numbers to make their product appear
as great as possible

a fair point, however continuning that analogy, you may accept that the
stats may be slightly off, however the bottom line is that the car WILL
still travel in a safe & reasonably reliable manner. surely if a psu is
unable to supply the correct voltage then it is unfit for its purpose

 

[kony]

I must admit i dont understand this whole crap PSU thing.

surely any product sold should work as advertised otherwise it shouldnt
be on the market.

Correct, it shouldn't be on the market.
But, it is.

how can a 400watt psu be sold, when it isnt actually 400watt, isnt that
illegal? and yet it just seems to be accepted as ok in this instance.

It can be sold because people buy it.
It *continues* to be sold because of apathy, who do you know that has
reported their overrated crap power supply to any governing agency or
started a lawsuit over one?

 

[David Besack]

I must admit i dont understand this whole crap PSU thing.

surely any product sold should work as advertised otherwise it shouldnt
be on the market.

From what I understand, if it says 400W, it *can* deliver 400W, but cheap
ones crap out after a couple of months if you operate them at near-capacity.

Think of computer speakers. Crappy ones, even if they are rated at 12W,
will blow out if you turn them all the way up and listen to them like that
day after day. Pretty soon they just can't take it. But if you keep them
about 1/2 - 2/3 most of the time, they'll probably last a while.

Same for power supplies. The crap ones can give you 400W, but if you run
every last ounce of power through them 24/7, they are likely to break. Then
again, if you have 1 HDD and a mediocre GPU, you're probably okay with a
cheap one.

When I was looking for a case/PSU combo, I read a lot of reviews, and the
most complaints I ever saw was about PSU problems.

 

[Mac Cool]

surely any product sold should work as advertised otherwise it shouldnt
be on the market.

The specs on electronics are usually overblown. (e.g. horsepower of a
tablesaw, watts per channel of a car stereo, life of a coppertop). Quality
brands usually have ratings that are closer to reality, in a few rare
instances, they best brands may even have realistic ratings. That isn't
saying that they are all liars, they are extreme optimists.

 

[Mr Bill Payer]

With any cheapo psu there is the risk that it can't supply the current
without a struggle, bad regulation or a crap switching circuit can cause
the switching frequency to drift off once the psu is under load. Poorly
wound inductors can also cause noise, in either case, I would change the
psu for a better make.

Hi Mantric

Cheers for your reply.

I put a 400W Mercury PSU taken from my "queit" Chaintech Apogee, AthlonXP
1800+ system into the Gigabyte KT400, AthlonXP 3000+ system to see if it was
any quieter. It made no difference, still squeelin' like ****. I changed the
video card, disconnected the USB and removed the modem. Made no difference.
I then put the 300W PSU back in the Gigabyte/AthlonXP 3000+ system to see if
it was louder. Still the same. I also stopping all the fans, still the same.
Got on my hands and knees with ear pointing/pressed against the case/psu end
to localize the source of the noise (lol). I'm now not sure if it is the PSU
or maybe I've cooked my brain!

I'll get a quality power supply to make sure that's the problem. If it still
whines then the mobo's getting swapped to check it's not overloading the
psu's.

Can motherboards emit high frequency noise?

Thanks again.

Scotoma

 

[Tony Houghton]

In <[email protected]>,

I'll get a quality power supply to make sure that's the problem. If it still
whines then the mobo's getting swapped to check it's not overloading the
psu's.

Can motherboards emit high frequency noise?

Yes, the coils can whine, and also some of those tiny srface mounted
components. If it's making the same noise with 2 PSUs it's much more
likely to be coming from the motherboard IMO. I've got an Asus which has
been doing it under load for over a year with no other ill effects
except that it's annoying. I had a noisy Intel cooler on it at first so
by the time I noticed the other noise I'd got rid of the board and CPU
it had replaced so it was too much hassle to RMA it :-(.

 

[ric]

I must admit i dont understand this whole crap PSU thing.

surely any product sold should work as advertised otherwise it shouldnt
be on the market.

how can a 400watt psu be sold, when it isnt actually 400watt, isnt that
illegal? and yet it just seems to be accepted as ok in this instance.

Specmanship. Remember the problem with hi-fi amp power ratings before
standards were reached? [Or are you too young to remember?]

Well, the similar situation exists with PSUs. Is a PSU to be called
400 watts if all of its individual outputs add up to 400 watts? Yes?
Even if they can't all reach full rated load simultaneously? Even if
they can only do so momentarily? Even if they can do so only at 25
degrees Celsius internal ambient? Even if the PSU has an MTBF of only
10,000 hours at 400w, but 50,000 hours at 300w?

There are NO standards, other than what UL, etc., insists on if the PSU
is to be UL listed. And that ain't much.

 

[Paul Hopwood]

a fair point, however continuning that analogy, you may accept that the
stats may be slightly off, however the bottom line is that the car WILL
still travel in a safe & reasonably reliable manner. surely if a psu is
unable to supply the correct voltage then it is unfit for its purpose

The point is that total power output is almost academic; what matters
is the power on each of the 5V and 12V rails. Seemingly similar
supplies, with the same total output rating, will vary hugely.

As we seem to be using car analogies, it's similar to measuring the
performance of a car based on the size of it's tyres. It may give
some indication (faster more powerful cars tend to have wider low
profile tyres) but there's nothing to stop anyone sticking huge fatboy
racer tyres on a 1.1l shopping wagon. Indeed, if tyre size became the
de facto standard for measuring performance it would be pretty much
certain that's exactly what every manufacturer *would*. It is
precisely what has happened to many computer components, including
PSUs, where an almost irrelevant method of measuring comparative
performance has been adopted.

What's the quality of the design, construction and components used
will greatly affect the reliability of the unit. A low quality PSU is
far more likely to surge or dip under load, or fail completely in a
way which can do permanent damage to the delicate electronic
components it is supplying.

You might arguably take the risk with a budget build costing a few
hundred quid, where a £60+ PSU would add significantly to the cost of
the machine, but it seems ludicrous to me that people are equally
inclined to risk reliability or even damage to components worth many
hundreds of pounds for the sake of saving £20 or so against a total
cost of a grand or more. The amount you spend on a power supply
should be a proportion of the total machine cost; I'd consider 5% or
even approaching 10% of the total budget to be reasonable.

Coming back to our car scenario; you probably wouldn't bother fitting
a clapped out Fiesta with Ferrodo air cooled brakes and Continental
ContiSportContact 2 tyres but sticking a cheap PSU into a high spec
machine is something akin to driving a Porsche with cracked
second-hand brake disks and remould tyres!

--

 

[Scotoma]

In <[email protected]>,


Yes, the coils can whine, and also some of those tiny srface mounted
components. If it's making the same noise with 2 PSUs it's much more
likely to be coming from the motherboard IMO. I've got an Asus which has
been doing it under load for over a year with no other ill effects
except that it's annoying. I had a noisy Intel cooler on it at first so
by the time I noticed the other noise I'd got rid of the board and CPU
it had replaced so it was too much hassle to RMA it :-(.

addresses.

Hi Tony Houghton

Thanks for your reply
I bought the Gigabyte mobo (£40uk) localy so I'll be giving the store a
visit quite soon.
Cheers again

Scotoma

 

[Mike Kay]

The point is that total power output is almost academic; what matters
is the power on each of the 5V and 12V rails. Seemingly similar
supplies, with the same total output rating, will vary hugely.

As we seem to be using car analogies, it's similar to measuring the
performance of a car based on the size of it's tyres. It may give
some indication (faster more powerful cars tend to have wider low
profile tyres) but there's nothing to stop anyone sticking huge fatboy
racer tyres on a 1.1l shopping wagon. Indeed, if tyre size became the
de facto standard for measuring performance it would be pretty much
certain that's exactly what every manufacturer *would*. It is
precisely what has happened to many computer components, including
PSUs, where an almost irrelevant method of measuring comparative
performance has been adopted.

What's the quality of the design, construction and components used
will greatly affect the reliability of the unit. A low quality PSU is
far more likely to surge or dip under load, or fail completely in a
way which can do permanent damage to the delicate electronic
components it is supplying.

You might arguably take the risk with a budget build costing a few
hundred quid, where a £60+ PSU would add significantly to the cost of
the machine, but it seems ludicrous to me that people are equally
inclined to risk reliability or even damage to components worth many
hundreds of pounds for the sake of saving £20 or so against a total
cost of a grand or more. The amount you spend on a power supply
should be a proportion of the total machine cost; I'd consider 5% or
even approaching 10% of the total budget to be reasonable.

Coming back to our car scenario; you probably wouldn't bother fitting
a clapped out Fiesta with Ferrodo air cooled brakes and Continental
ContiSportContact 2 tyres but sticking a cheap PSU into a high spec
machine is something akin to driving a Porsche with cracked
second-hand brake disks and remould tyres!

good point well made i guess!.

 

[Stuffed]

I'll get a quality power supply to make sure that's the problem. If it still
whines then the mobo's getting swapped to check it's not overloading the
psu's.

Can motherboards emit high frequency noise?

Yes. And then they squeak and chirrup, accompanied by whistles. And then
they die.

That's what happened to me anyway, as my caps were faulty (and one was the
wrong way round!). Got it fixed for the sake of it, it's now lovely and
quiet

 

[w_tom]

You are making many assumptions. For example, if a power supply is
rated to handle an input of as much as 400 watts - input; then will it
output 400 watts output? Not necessarily. That 400 watts does not
even claim it will consume 400 watts input. They have said only
enough for the technically naive to make assumptions. This is what
those discount power supplies do.

Furthermore, who is going to sue them? These supplies arrive with
all kinds of unknonwn trade labels. Who actually made them? And if
you sue, what happens to that little company that only resold a few
power supplies? In short, there is no downside to fudging labels when
'the expert' buys only on price.

In the meantime, many power supplies don't even claim to meet
standards that were defacto even 30 years ago. For example, an
abbreviated list of specs that is provided by every acceptable power
supply is demonstrated:
Specification compliance: ATX 2.03 & ATX12V v1.1
Acoustics noise 25.8dBA typical at 70w, 30cm
Short circuit protection on all outputs
Over voltage protection
Over power protection
100% hi-pot test
100% burn in, high temperature cycled on/off
PFC harmonics compliance: EN61000-3-2 + A1 + A2
EMI/RFI compliance: CE, CISPR22 & FCC part 15 class B
Safety compliance: VDE, TUV, D, N, S, Fi, UL, C-UL & CB
Hold up time, full load: 16ms. typical
Efficiency; 100-120VAC and full range: >65%
Dielectric withstand, input to frame/ground: 1800VAC, 1sec.
Dielectric withstand, input to output: 1800VAC, 1sec.
Ripple/noise: 1%
MTBF, full load @ 25°C amb.: >100k hrs

Did your supply make any of those claims? If not, then a failure
that damages the adjacent computer is acceptable. Even 30 years ago,
a failing power supply could not damage the load - ie computer, disk
drive, RAM, etc. Overvoltage protection was defacto standard even 30
years ago. It still must be part of any minimally acceptable supply.
And yet with so many 'bean counter' types claiming to be computer
experts, then so many discount power supplies don't even have
essential overvoltage protection. Failures are therefore directly
traceable to a human's lack of basic knowledge; not an electronics
failure.

As Kony noted in another post, these standards are defined by the
industry; not by law. A computer assembler (not to be confused with
one who builds computers) is suppose to have this so trivial and basic
electrical knowledge. Too many do not which is why power supplies
cause computer damage.

Another simple fact made obvious by those specifications. Short
together all power supply outputs AND power supply is not damaged.
Again, many computer assemblers don't even know this. They *assume*
(mostly from experience) that a shorted power supply will be damaged
even though power supplies 30 years ago could be shorted and not be
damaged. Again, it is about first learning basic knowledge; first
learning the underlying theory. Too many computer experts know only
from experience and don't need no stinkin' education. So many
computer components are damaged by a power supply bought only on one
specification - price. It is called damage directly traceable to the
'bean counting' human who cannot be bothered to first learn
technology.

 

[Franc Zabkar]

Correct, it shouldn't be on the market.
But, it is.



It can be sold because people buy it.
It *continues* to be sold because of apathy, who do you know that has
reported their overrated crap power supply to any governing agency or
started a lawsuit over one?

Who do you know who can tell the difference between quality and crap?
The answer is practically nobody. Even those who profess to be experts
have probably never looked underneath the covers and wouldn't
recognise what they saw anyway. I suspect that most of these
self-styled experts are merely critics with dubious qualifications who
do nothing more than regurgitate marketing hype, and who base their
"technical" assessments principally on price and reputation. Real
technicians, OTOH, will have experience in chip-level repair and will
not be fooled by the mindless "you-get-what-you-pay-for" mantra of
advertising droids.

As for PSU specs, I've only ever heard of one PC PSU design that even
attempts to regulate properly. By that I mean that the vast majority
of PSUs "regulate" by taking a weighted average of the +5V and +12V
rails, and then assume that the other supply rails will fall within
spec solely on the basis of their turns ratio. A poor design
philosophy, IMHO ...


- Franc Zabkar