Power supply EXPLOSION

ROTFPMSL! Or fill it with a non-electrically conductive liquid.


This was something i thought about a LONG time ago..
But 1) I dont like the idea of a big block of Oil/(such) sitting in the top
of my case ready to POP and leak
and 2) I dont think it's going to move the heat well enough to cool it.. I
have some MUCH better ideas that I would love to try and build (to
market&sell to casemfgrs) but they will break the "intel specifications" so
I figgure what the hell and havent put more into it yet ..

On Thu, 22 Jul 2004 10:42:30 GMT, "rstlne"
wrote:




Water cooler has no pump?
If it has a pump, "inaudible" would be a better description than
completely silent. Fans can also be inaudible, and while
inaudible, can keep OTHER components in the system cooler than
the water cooling since it only focuses on specific components.

MAC cube is a different design, not an alteration of a design
meant to use fans.

Traditional home PC's were not made with "fans" in mind..

Wrong, the fan in the power supply was sufficient for the amount
of heat generated. When it became necessary to increase airflow
to combat higher heat parts, that's exactly what they did.

Find an asus/tyan/msi/abit/gigabyte/Leadtek/Soltek/DfiLanParty board
specification that REQUIRES case fans in their specifications..

A board isn't a case. Some of them do make cases, with fan
mounts, often fan already in the mount, and barebones systems
with fans in them. Where do they specify that you even need a
case at all?

On Thu, 22 Jul 2004 11:58:36 GMT, kony wrote:

On Thu, 22 Jul 2004 09:34:38 +0100, "Peter Hucker"
wrote:

On Thu, 22 Jul 2004 04:18:40 GMT, kony wrote:

On Thu, 22 Jul 2004 03:35:45 +0100, "Peter Hucker"
wrote:


I sell you a 9V battery for $150.
I warrant it to power your system for 3 years.
Of course the product is unfit for advertised use per it's specs,
but so are many generic psu.
Of course it won't, so when you try it the battery fails
(hopefully nothing else does) and you return battery to me. A
month or two later (maybe longer) you get new 9V battery in the
mail. This goes on over and over and after 3 years I've still
retained a profit but you still don't have a viable power
solution. Granted this is an absurd extreme but goes to show
that a warranty is not an indicator of expected lifespan. MTBF
"could" be in a perfect world but this one isn't so (perfect).

Doesn't bother me. I have the use of a silent power supply for 3 years. Anyway we have the sale of goods act, and I'd get my money back after repeated failure.

That's just it, you DON'T have the use of it if/when it fails.
You have the use till then, but a system failure and downtime
till manufacturer receives it, gets around to sending you another
replacement. If you can't wait for the replacement then you have
to buy another power supply, so it's now 2X the cost.

So I use a noisy cheap one while I wait.

Keep in mind that a decent power supply will last close to a
decade, 3 year warranty should only be a factor for a low-end
unit, insufficient capacity for the system, defect or failure
fairly isolated from the design of PSU. 3 years is very short
lifespan for a name-brand PSU that typically costs much less per
same true wattage.

I probably won't have it in 10 years. People (well not me anyway) don't stick to the same computer parts!

... and it is your decision to make! If you want disposable
parts that's your choice but it's not very kind to the
environment, particularly when today's 3GHz systems have enough
processing power to remain viable for much longer than those from
10 years ago. Even if you personally don't want the system,
someone else might.

I don't really like wasting things, but they are selling it, so I am assuming that Thermaltake et al are not charletons. Perhaps I shall forward this conversation to them and see what they think of you.

Go ahead, I'd like more details of how they did it. Keep in mind
that I did not speculate on their specific implementation beyond
it's limited amperage rating, rather your implementation which
was a hack job.

While you're out looking for support, drop in to some web forums
and ask if they think it's a good idea to run a fanless system
that isn't optimized specifically for this, including parts
selection.

While you're out, drop into the document section of your CPU
manufacturer's website and see what they have to say about
chassis cooling. Ask your motherboard manufacturer too. You
might even luck out and find they claim it's ok, since the part
may last till the end of a warranty period, even though the
lifespan may be reduced by 50% or more.

You seem to have no grasp of the situation... your PSU is dead
because of your effort and yet you STILL can't learn from the
mistake. There ARE cases designed to passively cool, but it
takes a bit more than just traditional water-cooling, which is
meant to suppliment case cooling, not replace it.

Hey I won't get anywhere without experimenting.



--
*****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

What's the German word for Vaseline?
Vienerslide.

On Thu, 22 Jul 2004 11:58:36 GMT, kony wrote:

On Thu, 22 Jul 2004 09:34:38 +0100, "Peter Hucker"
wrote:

On Thu, 22 Jul 2004 04:18:40 GMT, kony wrote:

On Thu, 22 Jul 2004 03:35:45 +0100, "Peter Hucker"
wrote:


I sell you a 9V battery for $150.
I warrant it to power your system for 3 years.
Of course the product is unfit for advertised use per it's specs,
but so are many generic psu.
Of course it won't, so when you try it the battery fails
(hopefully nothing else does) and you return battery to me. A
month or two later (maybe longer) you get new 9V battery in the
mail. This goes on over and over and after 3 years I've still
retained a profit but you still don't have a viable power
solution. Granted this is an absurd extreme but goes to show
that a warranty is not an indicator of expected lifespan. MTBF
"could" be in a perfect world but this one isn't so (perfect).

Doesn't bother me. I have the use of a silent power supply for 3 years. Anyway we have the sale of goods act, and I'd get my money back after repeated failure.

That's just it, you DON'T have the use of it if/when it fails.
You have the use till then, but a system failure and downtime
till manufacturer receives it, gets around to sending you another
replacement. If you can't wait for the replacement then you have
to buy another power supply, so it's now 2X the cost.

So I use a noisy cheap one while I wait.

Keep in mind that a decent power supply will last close to a
decade, 3 year warranty should only be a factor for a low-end
unit, insufficient capacity for the system, defect or failure
fairly isolated from the design of PSU. 3 years is very short
lifespan for a name-brand PSU that typically costs much less per
same true wattage.

I probably won't have it in 10 years. People (well not me anyway) don't stick to the same computer parts!

... and it is your decision to make! If you want disposable
parts that's your choice but it's not very kind to the
environment, particularly when today's 3GHz systems have enough
processing power to remain viable for much longer than those from
10 years ago. Even if you personally don't want the system,
someone else might.

I don't really like wasting things, but they are selling it, so I am assuming that Thermaltake et al are not charletons. Perhaps I shall forward this conversation to them and see what they think of you.

Go ahead, I'd like more details of how they did it. Keep in mind
that I did not speculate on their specific implementation beyond
it's limited amperage rating, rather your implementation which
was a hack job.

You stated that a power supply without a fan would not last 3 years.


--
*****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

Pizza IS the four food groups!

"kony" wrote in message
...
On Thu, 22 Jul 2004 10:01:01 GMT, "rstlne"
wrote:

Rubbish. I can only hear my water pump if I place my ear on the side
of
the radiator. The pump is underwater, inside the radiator, all sound is
absorbed.

... and your system failed, didn't it?
So what if other parts haven't failed YET?
There is a clear relationship between heat and lifespan, and your
system CANNOT have most of not all components not under a water
block, running as cool as when an inaudible fan in employed.
I have systems with inaudible fans still running fine, not a
single part has failed.


His Cooling didnt fail, His PSU fail'd..

Did he not suggest it failed because it overheated?
What would you call that if not a failure to cool it?

The failure he show'd was not down to undercooling (You can see the parts
that went, they wouldnt have been the first parts to go)

This is the most common failure that there is in HomePC's..
in fact I would say that when everything is done correctly in electronics
that many times a SMPS is probably the most common failure of anything..

"When everything is done correctly" is a pretty vague statement,
but given that generality, a name-brand PSU of ample wattage and
proper surge protection device would be used. In such setups the
average lifespan of the SMPS is higher than the 3 years reported
(IIRC).

I have had PSU's that last a week, Had some dead straight from the box, Had
some that never fail'd to date..
SMPS's are verry tempermental things.. If your using a huge supply with a
small load (ample wattage) then I would expect it to die much faster than
one that's at a 60/70% load through it's lifetime..
I have seen too many SMPS's that were based in and out of PC's and I know
that failure can happen at any time just for the hell of it..

I see your point, and I see his point also..
He SHOULD be able to run a fanless system to be honest..

Fanless is possible, if designed to be so.
Taking a system meant for fan-based cooling and trying to run
without the fans is asking for trouble unless ALL parts that
create excess heat are accordingly cooled to levels conducive to
desired lifespan. Components should outlast the user's desire to
use them... which wasn't the case here.

Transformers, Small VoltageComparitors, Resistors, Caps should not cause
excessive heat.
If those componets cant run without a fan then I suggest that the supply was
crap quality to begin with.

It's certainly true that someone could work backwards, start out
with a system optimized for fanned cooling and remake it into a
passively (fanless) cooled system, but not by just putting
waterblocks on the 4 hottest components/chips.

Passive Cooling is possible..
I have seen a AXP2500+ system that is 100% fanless AND NOT using any type of
water/other cooling methods.. All based on the simple fact that heat rises
and nothing else..


I think I am done replying to you..
You feel it's impossible..
I know otherwise..
time to just leave it at that..

On Thu, 22 Jul 2004 12:54:55 GMT, kony wrote:

On Thu, 22 Jul 2004 09:37:02 +0100, "Peter Hucker"
wrote:

On Thu, 22 Jul 2004 04:24:34 GMT, kony wrote:

On Thu, 22 Jul 2004 03:42:38 +0100, "Peter Hucker"
wrote:


Water cooler has no pump?
If it has a pump, "inaudible" would be a better description than
completely silent.

As I said in the other reply, it is SILENT unless my ear is on the radiator and I am crouched uncomfortably under the desk!!!

Fans can be too.

I've yet to find one. Which particular model did you use, and using what model of fan controller?

Panaflo, "L" speed in the size of your choice. For optimal fan
control try a series of diodes, but just about any fan controller
with suitable range can get close to same result, providing it
has full (or nearly so) range, adjustment down to at least 4-5V.
Lowest cost implementation would be seen by having an adjustment
method, I have a pair of 100Ohm POTs for resistor sizing and a
12-position switch with diode-pairs in series for determining #
of diodes per specific fans. Avoid expensive larger fans like
120x38 mm or at least inquire whether they are voltage-dependant,
as some have internal regulation to achieve fixed RPM at a given
larger voltage range. Sunon and Comair are type makes with this
feature in some models, I don't recall in most others as i
usually deal in bulk with Nidec, NMB, Panaflo. Panaflo are the
quietest of the three due to the hydrowave (sleeve) bearing.

Panaflo eh? I'll go for them then. ANY panaflo? Or do any panaflo contain the RPM regulation?

Of course not, since you jumped to conclusions instead of
pursuing it. People do it all the time, there are reviews and
recommendations all over the web, try a Google search.

I have. It's a maze. There are articles everywhere, and what many call silent are not.

Success may not be seen on first try, and some people DO appear
to be tone-deaf after a certain point, I have heard several
people claim "quiet" for parts I'd consider too loud for a "PC".
Ideally the fans should not be on front wall of case, be as large
and thick as possible given size constraints, but with multiple
fans to allow suitable flow rate at lowest RPM. A Panaflo can
run at around 5V, though part of the noise level is determined by
how it's mounted, the case wall thickness, overall case design,
nearby obstructions and orientation relative to the ear. It's
much easier to make a "quiet" system than silent, and even one
only "quiet" can be inaudible if placed under a desk, without
much consideration of specific fan models provided a speed
controller is used. Ear-level on a desk with higher-heat parts
is the most challenging, can be necessary to line case with sound
absorbant material and have rear of PSU, where most noise
escapes, pointed at a low-density materal.

It's strange how some people have VERY noisy PCs, and when I ask them how they put up with the noise, they say "what noise?" By bery noisy, I mean 3 full speed noisy case fans and also noisy processor fan(s).

and while
inaudible, can keep OTHER components in the system cooler than
the water cooling since it only focuses on specific components.

MAC cube is a different design, not an alteration of a design
meant to use fans.

So what?

So you wanted to use it as an example, when the whole reason it's
fanless is that they designed it from ground up to be that way.
Note that the had to reengineer the whole thing

Reengineering is different from moifying in what way?

You have already mentioned more changes (like case orientation)
than you did originally, it may be that you have made enough
modifications to help a lot, but the issue is then how many, how
effective they are. We cannot see your case, only what you
describe of it. To a certain extent it's expected that you
would've mentioned the ways it deviates from "normal". If you
did make enough changes and monitored the result then it may not
be so different from their reengineering, but remeober that they
designed it from the ground up, did not have to make any
compromises at all. Simply starting out with a standard case, no
matter how you orient it, is already a significant compromise.
Note how Apple allowed quite a bit of flow staight up though that
cube.

I'll get there :-) The main problem I think at the moment is the RAM. Even with large heatsinks it is a little hot for my liking with just convection (not even overclocking it - in fact it is still the same temperature when UNDERclocking it?!?)



--
*****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

In Today's Market Activity, Helium was up. Feathers were down. Paper was stationary. Fluorescent tubing was dimmed in light trading. Knives were up sharply. Cows steered into a bull market. Pencils lost a few points. Hiking equipment was trailing. Elevators rose, while escalators continued their slow decline. Weights were up in heavy trading. Light switches were off. Mining equipment hit rock bottom. Diapers remain unchanged. Shipping lines stayed at an even keel. The market for raisins
dried up. Coca Cola fizzled. Caterpillar stock inched up a bit. Sun peaked at midday. Balloon prices were inflated. And, Scott Tissue touched a new bottom. Invest wisely!

rstlne wrote:

snip

I think I am done replying to you..
You feel it's impossible..

As do I.

I know otherwise..

And you have the exploded PSU to prove it.

time to just leave it at that..

Good move, you are being shown up at every post.
--
~misfit~

On Thu, 22 Jul 2004 10:09:26 GMT, "rstlne"
wrote:


Doesn't bother me. I have the use of a silent power supply for 3 years.
Anyway we have the sale of goods act, and I'd get my money back after
repeated failure.

That's just it, you DON'T have the use of it if/when it fails.
You have the use till then, but a system failure and downtime
till manufacturer receives it, gets around to sending you another
replacement. If you can't wait for the replacement then you have
to buy another power supply, so it's now 2X the cost.


You seem to be really confusing things here..


Nope, I"m quite clear on what happens.
A warranty isn't of much use if the part fails in a system that
can't tolerate weeks or longer downtime, so a replacement part
must be bought.

I am starting to think your a troll.. but If not then I say consider this..

Ok, think it, but in the meantime my parts aren't baking to an
early demise. I'm not advocating hurricane force fans, just the
proven solutions that achieve temp conducive to expected
component lifespan.


The PSU company wants to make money..
If they have to replace 100% of their sold stock then that means they'll
probably not be in profit..

Not true when they have a huge markup.
Ever priced something labeled as "silent"? Usually that 6 letter
word adds 50% to the price, if not more.


So in other words.. This stuff will work for most people under most
conditions..

What's "this stuff"?
A specific PSU has a specific load rating and an expected
lifespan at a specific temp. "Most" people do not put water
blocks in their power supplies, take out all their case fans,
etc. That's not "most conditions" either. "This stuff" is only
tried by fringe users annoyed by junk components and deluded by
kiddie website articles. An optimally designed system can be
VERY quiet, OEMS prove it year after year, but they're
constrained by EMI emission limits, while an end-user already
demonstrating a willingness to modify, can have even quieter or
cooler (or both) system than seen from an OEM.

You cant get 600w fanless PSU's because they just cant manage it..
You can get 200/300/350 Out the wazoo because they can manage it.. They are
not going to make kit that HAS to be replaced..
It's like making a car that's so poor in quality that the whole car will
need to be replaced.. It's not something the mfgr's do.

Actually, any PSU has a finite lifespan. They all eventually
would need be replaced. The issue is then if the lifespan
exceeded user's needs... not only the buyer, but whoever else
might use the system. For example, right now I'm sure I can find
someone who will want my current most-used systems when I'm ready
to replace them again.

I"m not implying that it's guaranteed that a passively cooled PSU
willl die before it's warranty is up, but the odds are very high
that it's lifespan will not be anywhere near that of a
traditionally designed name-brand PSU of same price range. I am
now drifting into vagueness, the specific PSU needs be evaluated
for fitness powering specific components, it's temp used as one
gauge of whether it is appropriate for the system (as it is
designed).

On Thu, 22 Jul 2004 13:04:54 GMT, "rstlne"
wrote:

snip

Did he not suggest it failed because it overheated?
What would you call that if not a failure to cool it?

The failure he show'd was not down to undercooling (You can see the parts
that went, they wouldnt have been the first parts to go)

They may not have been the first parts to go, merely the most
visually obvious. Failures can and do induce other part failues.



This is the most common failure that there is in HomePC's..
in fact I would say that when everything is done correctly in electronics
that many times a SMPS is probably the most common failure of anything..

"When everything is done correctly" is a pretty vague statement,
but given that generality, a name-brand PSU of ample wattage and
proper surge protection device would be used. In such setups the
average lifespan of the SMPS is higher than the 3 years reported
(IIRC).

I have had PSU's that last a week, Had some dead straight from the box, Had
some that never fail'd to date..

"some" is a bit too vague.

If the percentage was really high enough to make it significant,
worth mentioning here, you had a breakdown somewhere in the
supply (commondity supply, not power supply) chain. Name-brand
(actual manufacturer's name on the label) units do not often fail
within the first years, unless they had an initial defect. We
can see the odds are against that here, it worked fine, was
modded, continued to work for a while, then failed.

Try diffeent power supplies, it is not typical to have this
expectation of failure at all/any moment.


SMPS's are verry tempermental things.. If your using a huge supply with a
small load (ample wattage) then I would expect it to die much faster than
one that's at a 60/70% load through it's lifetime..

Define huge.
Typical PC PS/2 power supply is now 300-550W. Typical modern
system uses at least 90W, more often closer to 170W. Looking at
the extreme example you imply, running a 90W consuming system
from a 550W PSU, it should run for many years providing that 550W
PSU is a decent name-brand, with quality fan.


I have seen too many SMPS's that were based in and out of PC's and I know
that failure can happen at any time just for the hell of it..

Too vague to be even slighlty useful. Isolate the failure mode
then we have something to work with. Decent PSU do not just
randomly fail at a rate that would be of concern to typical
single or few unit PC user. Next time you find a failure (and
the next, and so on) track down the fault. Understand why and
correct that fault.

There's nothing in a PC that's particularly prone to failure
compared to the rest of a computer, many of the same
components/types are used, so if the failure rate is that high
another possibility is that your power supplies are overheating,
which as I've stated all along, will indeed reduct lifespan
(ended with a failure). Since these PSU seem to be dropping like
flies where you are, show me a manufacturer's name-brand unit
used as intended, that failed, and that the part which failed
wasn't due to heat or faulty AC input.


I see your point, and I see his point also..
He SHOULD be able to run a fanless system to be honest..

Fanless is possible, if designed to be so.
Taking a system meant for fan-based cooling and trying to run
without the fans is asking for trouble unless ALL parts that
create excess heat are accordingly cooled to levels conducive to
desired lifespan. Components should outlast the user's desire to
use them... which wasn't the case here.

Transformers, Small VoltageComparitors, Resistors, Caps should not cause
excessive heat.

Not excessive when rate of cooling allows Temp rise to remain
within check. In a mostly closed metal box it doesn't take a lot
of watts of heat to raise temp, consider an oven on low setting.


If those componets cant run without a fan then I suggest that the supply was
crap quality to begin with.

It came with a fan because it was optimized to use one.
There is no magic involved in SMPS, they're only about 70%
efficient. Do the math, a system needing 200W power delivered
causes PSU to use roughly 286W, meaning 86W of heat. Since you
have no concept of what it takes to remove 86W of heat, Google
for flow rates needed to achieve desired temp rise, and consider
that for each 10C temp rise a capacitor's life is cut in half.
It's a fact, clearly stated on spec sheets from major
manufacturers, let alone the low-end junk components with
questionable specs.



It's certainly true that someone could work backwards, start out
with a system optimized for fanned cooling and remake it into a
passively (fanless) cooled system, but not by just putting
waterblocks on the 4 hottest components/chips.

Passive Cooling is possible..
I have seen a AXP2500+ system that is 100% fanless AND NOT using any type of
water/other cooling methods.. All based on the simple fact that heat rises
and nothing else..

It's not a matter of whether it's possible to get a system stable
without a fan, it's the resulting lifespan if system isn't
specifically optimized toward that end. I agree that it's
possible, but only with very extensive deviation from the
standard form factors.

The real issue here is that there is no good reason to do without
the fan. A fan can be inaudible inches away from the ear when
separated by the thin sheet metal of a system case. Any decent
fan has MTBF of a decade at much higher RPM, may easily last 20
years or more at very low RPM, even longer with today's higher
tech lubricants having higher film strength and molecular
stability compared to dino grease.



I think I am done replying to you..
You feel it's impossible..
I know otherwise..
time to just leave it at that..

Not impossible to get it running, but very hard to attain nearly
same lifespan, and pointless when a fan can do the job quietly,
potentially at only 1/4th as much time or $ spent towards this
end instead of remaking entire chassis, PSU, etc. (and voiding
warranties). I used to sell motherboards... if i knew they'd
been operated in a homemade fanless system your warranty would be
voided. Next time you have a motherboard failure and send it to
the manufacturer for RMA replacement, tape a note to it
describing how you "known" it's ok to run a system without ANY
fans then see what happens!

On Thu, 22 Jul 2004 14:02:45 +0100, "Peter Hucker"
wrote:


Go ahead, I'd like more details of how they did it. Keep in mind
that I did not speculate on their specific implementation beyond
it's limited amperage rating, rather your implementation which
was a hack job.

You stated that a power supply without a fan would not last 3 years.

No, I did not, or at least I didn't intend to...

My argument was that a 3 year warranty isn't a guarantee that it
will work for 3 years, that warranties are a marketing tool the
manufacturer would just as soon do away with if possible, but
given some bean counter with an idea that additional profit seen
from this "feature" will outweigh expense, a warranty is
provided. The truth about warranty replacement is that once a
manufacturer sets up a manufactuering line, they produce a given
number and what they don't sell are shelved for a period to serve
as warranty replacements. There is very little cost associated
with this, all of the R&D, parts purchasing, manufacturing, etc,
is done and they're just sitting around. The higher cost of the
unit previously linked (I am confident it's quite high priced
compared to other name brand PSU of same amperage specs) is way
more than enough to offset warranty replacement costs.