Power supply EXPLOSION

On Sat, 24 Jul 2004 14:17:47 +0100, "Peter Hucker"
wrote:


I found one hot inductor/coil/transformer/whatever it is
(just two wires to it) that was warm, I put a passive
heatsinnk on that and it kept it at less than doby temperature.

That's a choke, and the problem with heatsinking inductors is that
their cores don't conduct heat well or have little surface area for a
heatsink. And toroid chokes have wire wrapped all around them,
meaning you don't want a heatsink mounted against the wiring since it
could be shorted.

The outside of it was square in shape, and flat. A breeze of air wouldn't have cooled it any better than the heatsink (the entire exposed surface of it was below body temperature with the heatsink on it.


I can't be certain I've seen all Sparkle/Fortron power supply
designs for PCs, but none come to mind that have a square, flat
inductor or choke with a surface anywhere near easy to affix a
'sink.


75% to me sounds rediculously low, but then I don't know much
about the design of switched modes. I thought switched modes gave
of a lot less heat than the old ones?

75% is above average for an ATX supply, and linear supplies haven't
been used for PCs since maybe 1980. I suspect switching supplies have
been preferred mostly for their lighter weight, not increased
efficiency, because by operating the transformer at a much higher
frequency, it can be made much smaller for the same amount of power,
and the filter capacitors can also be smaller.

I assumed better efficiency, as when you get one of hose plug in power supplies (for charging a mobile phone etc), the ones that are heavy (presumably just a transformer and a rectifier) get hot, while the very lightweight ones do not.

Yes, that's a good example, but it is relative that the phone
uses so little power to charge... perhaps a half-dozen watts
instead of a few hundred.



Kony is right about Panaflos, and large fans can be quieter because
they don't have to spin as quickly for the same airflow and will
produce less bearing noise. Also sound reflections inside the
computer case can make fans sound louder, so consider lining the
interior with padding, always taking care to not use anything
flammable. Teflon-coated fiberglass cloth for draperies may be the
best choice, and it doesn't shed fibers nearly as badly as ordinary
fiberglass cloth does.

Is padding not going to insulate the case so I would then need more cooling anyway?

There is a relatively insignificant amount of heat conduction
though walls of a metal case, if case feels warm it's a sign that
system is running too hot to begin with. Insulation of metal
walls (provided it doesn't significantly lower I/O airflow) could
even allow slightly higher RPM fan per same noise level, so
ultimately it is a positive factor in case cooling. Ultimately
the most noise escaping from a decent case is from fans mounted
to external panels, if on a desk the system might not allow
fan(s) mounted in front at all, as even the rush of air could be
heard if fan could not... even though a front fan helps move air,
it is still a constriction of the front air passage compared to
similar size hole without body of fan blocking it. Also avoid
direct path of sound to external environment, overlap bezel with
staggered holes so sound must bounce a couple times before making
it outside of case, and if you lined that area with
sound-absorbant material (if possible) so much the better.

On Sat, 24 Jul 2004 15:20:17 GMT, kony wrote:

On Sat, 24 Jul 2004 14:17:47 +0100, "Peter Hucker"
wrote:


I found one hot inductor/coil/transformer/whatever it is
(just two wires to it) that was warm, I put a passive
heatsinnk on that and it kept it at less than doby temperature.

That's a choke, and the problem with heatsinking inductors is that
their cores don't conduct heat well or have little surface area for a
heatsink. And toroid chokes have wire wrapped all around them,
meaning you don't want a heatsink mounted against the wiring since it
could be shorted.

The outside of it was square in shape, and flat. A breeze of air wouldn't have cooled it any better than the heatsink (the entire exposed surface of it was below body temperature with the heatsink on it.

I can't be certain I've seen all Sparkle/Fortron power supply
designs for PCs, but none come to mind that have a square, flat
inductor or choke with a surface anywhere near easy to affix a
'sink.

It was basically the same as a standard transformer with the iron core.... oh I'll just take a picture!

http://80.229.155.158/temp/coil.jpg

75% to me sounds rediculously low, but then I don't know much
about the design of switched modes. I thought switched modes gave
of a lot less heat than the old ones?

75% is above average for an ATX supply, and linear supplies haven't
been used for PCs since maybe 1980. I suspect switching supplies have
been preferred mostly for their lighter weight, not increased
efficiency, because by operating the transformer at a much higher
frequency, it can be made much smaller for the same amount of power,
and the filter capacitors can also be smaller.

I assumed better efficiency, as when you get one of hose plug in power supplies (for charging a mobile phone etc), the ones that are heavy (presumably just a transformer and a rectifier) get hot, while the very lightweight ones do not.

Yes, that's a good example, but it is relative that the phone
uses so little power to charge... perhaps a half-dozen watts
instead of a few hundred.

Yes, but I was comparing it to another of a half-dozen watts that DID get hot and was heavy.

Kony is right about Panaflos, and large fans can be quieter because
they don't have to spin as quickly for the same airflow and will
produce less bearing noise. Also sound reflections inside the
computer case can make fans sound louder, so consider lining the
interior with padding, always taking care to not use anything
flammable. Teflon-coated fiberglass cloth for draperies may be the
best choice, and it doesn't shed fibers nearly as badly as ordinary
fiberglass cloth does.

Is padding not going to insulate the case so I would then need more cooling anyway?

There is a relatively insignificant amount of heat conduction
though walls of a metal case, if case feels warm it's a sign that
system is running too hot to begin with. Insulation of metal
walls (provided it doesn't significantly lower I/O airflow) could
even allow slightly higher RPM fan per same noise level, so
ultimately it is a positive factor in case cooling. Ultimately
the most noise escaping from a decent case is from fans mounted
to external panels, if on a desk the system might not allow
fan(s) mounted in front at all, as even the rush of air could be
heard if fan could not... even though a front fan helps move air,
it is still a constriction of the front air passage compared to
similar size hole without body of fan blocking it. Also avoid
direct path of sound to external environment, overlap bezel with
staggered holes so sound must bounce a couple times before making
it outside of case, and if you lined that area with
sound-absorbant material (if possible) so much the better.

Not sure what you mean by this overlap, how can you make sound bounce without making the airflow have to bounce too?

--
*****TWO BABY CONURES***** 15 parrots and increasing http://www.petersparrots.com
93 silly video clips http://www.insanevideoclips.com
1259 digital photos http://www.petersphotos.com
Served from a pentawatercooled dual silent Athlon 2.8 with terrabyte raid

Why does sour cream have an expiration date?

On Sat, 24 Jul 2004 16:45:50 +0100, "Peter Hucker"
wrote:

snip

Also avoid
direct path of sound to external environment, overlap bezel with
staggered holes so sound must bounce a couple times before making
it outside of case, and if you lined that area with
sound-absorbant material (if possible) so much the better.

Not sure what you mean by this overlap, how can you make sound bounce without making the airflow have to bounce too?

That there is not a direct line of sight into case, front bezel
air intake is offset from metal inake, staggered if you will. It
will make airflow travel somewhat longer path but has a greater
effect on noise (especially with sound-dampening material) so
essentially results in greater airflow/noise ratio. It may not
be necessary though, but potentially helpful if system remains
too loud after other measures to combat the noise.

Looking at a side view of front bezel and front metal wall of
case, assuming rough parallelism and as best my poor ascii art
can show;


Metal Plastic
| |
|
|
| |
|
|
| |
|
|
| |
|
|
| |
| |

On Sat, 24 Jul 2004 20:35:16 GMT, kony wrote:

On Sat, 24 Jul 2004 16:45:50 +0100, "Peter Hucker"
wrote:

snip

Also avoid
direct path of sound to external environment, overlap bezel with
staggered holes so sound must bounce a couple times before making
it outside of case, and if you lined that area with
sound-absorbant material (if possible) so much the better.

Not sure what you mean by this overlap, how can you make sound bounce without making the airflow have to bounce too?

That there is not a direct line of sight into case, front bezel
air intake is offset from metal inake, staggered if you will. It
will make airflow travel somewhat longer path but has a greater
effect on noise (especially with sound-dampening material) so
essentially results in greater airflow/noise ratio. It may not
be necessary though, but potentially helpful if system remains
too loud after other measures to combat the noise.

Looking at a side view of front bezel and front metal wall of
case, assuming rough parallelism and as best my poor ascii art
can show;


Metal Plastic
| |
|
|
| |
|
|
| |
|
|
| |
|
|
| |
| |

I see. I always thought that would perhaps cause more if a "rushing air" sound, like with the hose of a vacuum cleaner.




--
*****TWO BABY CONURES***** 15 parrots and increasing http://www.petersparrots.com
93 silly video clips http://www.insanevideoclips.com
1259 digital photos http://www.petersphotos.com
Served from a pentawatercooled dual silent Athlon 2.8 with terrabyte raid

Success always occurs in private, and failure in full view.

On Sat, 24 Jul 2004 21:51:54 +0100, "Peter Hucker"
wrote:



I see. I always thought that would perhaps cause more if a "rushing air" sound, like with the hose of a vacuum cleaner.

Depends on how large the opening is. Small opening causes
higher airspeed, more turbulent noise. Larger lets more interior
noise escape, but minimizes turbulent noise and allows lower
speed exhaust fan and/or lower speed cpu fan for same sustained
temp. In implementation it may not be so simple unless you're
willing to really tweak fan speeds instead of settling for those
most easily obtainable, so it could be viewed as more of a
near-final tweak to counteract a situation of high air intake
speed OR insufficient intake which would be louder if opening was
too large. As a case becomes more personally customized the
decisions may be made one at a time, until you're happy enough
with the noise level that you aren't willing to do any more work
to reduce it further.

"Peter Hucker" wrote in message news:opsbm8jxs3aiowgp@blue...

On 24 Jul 2004 01:35:38 -0700, do_not_spam_me wrote:

Did you measure the voltage on the heatsink before you did this,
in case it had 170VDC riding on it?

The heatsink that was bolted the the earthed PSU chassis? No.

I failed to mention that while the heatsink attached to the low
voltage components poses no shock hazard, my concern here was over the
heatsink attached to the high voltage transistors because even if it
measures 0V, as I believe it does in most Fortron-Source supplies,
it's not safe to water cool it or allow it to have any chassis or
external contact. This is because its electrical insulation is very
thin, and good safety practices call for using much better insulation
for such high voltages.

A heatsink bolted to the power supply chassis indicates a
Fortron-Source supply. If so, you ruined one of the best power
supplies available.

I got it for a fiver!

And now it's worth nothing. You should have experimented on one of
your worse supplies because a 300W Fortron-Source will outperform many
350W and even some 400W units.

I found one hot inductor/coil/transformer/whatever it is
(just two wires to it) that was warm, I put a passive
heatsinnk on that and it kept it at less than doby temperature.

http://80.229.155.158/temp/coil.jpg

The outside of it was square in shape, and flat. A breeze of
air wouldn't have cooled it any better than the heatsink (the
entire exposed surface of it was below body temperature with
the heatsink on it.

That picture indicates it's a power factor correction choke, as
evidenced by the "PFC" printed over it, and if air flow wouldn't cool
it significantly, I doubt the manufacturer would have included those 3
air slots on each side. Also the metal box over it probably doesn't
make good contact with its core inside, so a heatsink external to it
would probably be useless. This choke is a high voltage device and
probably located next to high voltage components that aren't
completely insulated, so be very careful in installing a heatsink on
it.

sound reflections inside the computer case can make fans
sound louder, so consider lining the interior with padding,
always taking care to not use anything flammable.
Teflon-coated fiberglass cloth for draperies may be the best
choice, and it doesn't shed fibers nearly as badly as
ordinary fiberglass cloth does.

Is padding not going to insulate the case so I would then need
more cooling anyway?

Only if you have a nearly airtight case. Otherwise air flow matters
far more than heat conduction through the case.

Franc Zabkar wrote in message . ..
On 23 Jul 2004 01:53:04 -0700,
(do_not_spam_me) put finger to keyboard and composed:

www.silentpcreview.com was not very enthusiastic, probably
because the company simply took the same basic 65-75%
efficiency design that's been in PCs since 1980 and tried
to make it fanless by merely enlarging the heatsinks and
adding more vent holes. They should have instead raised
the efficiency to around 85-90% because such supplies are
common for non-PC purposes. The best I've seen was 96%,
but it was expensive.

I wonder if this figure of 96% was for a high voltage PSU that was
optimised for a fixed load. AFAICS, the efficiency of a low voltage
PSU would be limited by the Schottky rectifiers on the load side. For
example, a 3.3V supply would be limited in efficiency to 3.3 / (3.3 +
0.6) = 85%. I suppose higher efficiencies could be achieved with high
current, stud mounted rectifiers with lower forward voltage drops, but
these would be very expensive (?).

The 96% efficient supply put out +-12V, 5V, and some others. It
operated in resonant mode, and to reduce diode drops, almost all the
regular diodes were replaced with transistors (MOSFET or IGBT), even
for the hot side bridge. I twas claimed that the switching was also
controlled to tweak the efficiency more.

"Dorothy Bradbury" wrote in message ...

To create your own PC fanless SMPS involves risk:
o Water cooling the MOSFETs is typically done
o That still leaves thermal output by Transformer,
Capacitors, Coils & Load Resistors
o So the issue then is will convection cool those
components sufficiently

So it's Make or Buy:
o Make - consider the Cost, Development, Piloting & Failure
of your design
o Buy - consider the cost of an Off-The-Shelf design someone
tested to CE approval

For a Bedroom-PC, fanless-SMPS, I'd always go for the Buy option.

Personally, if you do decide to modify your PSU:
o Consider running a 12V fan at say 5V
---- to cool transformer, coil, capacitors & sundry components
o Consider using a heatsink-with-internal-water-path
---- these solutions tend to leak least
o Consider turning the PSU upside down
---- so gravity pulls water away from 300V DC & transformers
o Consider using a temperature controlled fan

Usual note re primary MOSFET tabs needing insulatory pads,
since the tabs are often live - some run the heatsinks live also.

You could replace the primary capacitors with 105oC or 125oC
low-ESR units, but there are other thermal risks in a PSU.

Not many home ambient environments are below about 15dB(A).
So running a low-noise fan at 6-7V for transformer/coil/caps is not
a bad idea - leave purity of design to Architects & clients with money.

I wouldn't encourage people to try this because the level of
confidence tends to far exceed the level of skill, something you don't
want when deadly voltages are involved.

On 25 Jul 2004 00:17:43 -0700, do_not_spam_me wrote:

"Peter Hucker" wrote in message news:opsbm8jxs3aiowgp@blue...

On 24 Jul 2004 01:35:38 -0700, do_not_spam_me wrote:

Did you measure the voltage on the heatsink before you did this,
in case it had 170VDC riding on it?

The heatsink that was bolted the the earthed PSU chassis? No.

I failed to mention that while the heatsink attached to the low
voltage components poses no shock hazard, my concern here was over the
heatsink attached to the high voltage transistors because even if it
measures 0V, as I believe it does in most Fortron-Source supplies,
it's not safe to water cool it or allow it to have any chassis or
external contact. This is because its electrical insulation is very
thin, and good safety practices call for using much better insulation
for such high voltages.

You have made a mistake. You are assuming I am safe.

A heatsink bolted to the power supply chassis indicates a
Fortron-Source supply. If so, you ruined one of the best power
supplies available.

I got it for a fiver!

And now it's worth nothing. You should have experimented on one of
your worse supplies because a 300W Fortron-Source will outperform many
350W and even some 400W units.

Worse? That WAS the worse one. I didn't want to break my gold plated dual fan 600 watt thing. The 300W one I broke was absolutely terrible - 10% voltage accuracy?! **** that. This 600 watt one is 2%. And what's Fortron-Source? A make or a design?

I found one hot inductor/coil/transformer/whatever it is
(just two wires to it) that was warm, I put a passive
heatsinnk on that and it kept it at less than doby temperature.

http://80.229.155.158/temp/coil.jpg

The outside of it was square in shape, and flat. A breeze of
air wouldn't have cooled it any better than the heatsink (the
entire exposed surface of it was below body temperature with
the heatsink on it.

That picture indicates it's a power factor correction choke, as
evidenced by the "PFC" printed over it, and if air flow wouldn't cool
it significantly, I doubt the manufacturer would have included those 3
air slots on each side. Also the metal box over it probably doesn't
make good contact with its core inside, so a heatsink external to it
would probably be useless. This choke is a high voltage device and
probably located next to high voltage components that aren't
completely insulated, so be very careful in installing a heatsink on
it.

It seemed pretty insulated to me, and I never even got a tickle off it.

The thing was very hot without a heatsink, and cool with one, so I assumed that the heat was getting to the oiutre casing just fine.




--
*****TWO BABY CONURES***** 15 parrots and increasing http://www.petersparrots.com
93 silly video clips http://www.insanevideoclips.com
1259 digital photos http://www.petersphotos.com
Served from a pentawatercooled dual silent Athlon 2.8 with terrabyte raid

Do not adjust your mind - the fault is with reality.

I wouldn't encourage people to try this because the level of
confidence tends to far exceed the level of skill, something you don't
want when deadly voltages are involved.

I wouldn't water cool a PSU full stop.

However, for the original poster, my key point is:
o Not many home ambient environments are below about 15dB(A)
o Running a low-noise fan at 6-7V for transformer/coil/caps is not a bad idea

Water cooled PSUs are very common in industry (using chillers) since it allows
a smaller package and remote disposal of heat. However, they are designed for
it throughout - like the Cray-1 water-cooled PSUs were similarly.

I also feel the failure was temp/water - doesn't take much of a leak, and
we are talking 300V-DC caps and 110/220V+ AC fused in the amps.
Water will happily conduct such voltages to an output rail & toast $$$$.
--
Dorothy Bradbury.