(OT) Heating Hot Water with your CPU (Processor)


C

casey.o

There's a few videos on Youtube where guys are overclocking Pentium 4
systems and using nitrogen to cool them. That does seem a bit
excessive, not to mention crazy. But one of them claims they are
running a 4ghz processor at almost 8ghz.

I'm not intending to do any of that, nor even overclock, because I know
that can do a lot of damage. I did have an old Pentium 1, that someone
overclocked from 133mhz to over 200mhz, and that machine kicked butt.

However, running my latest computer, which is a Pentium 4, 3ghz, I can
not believe the amount of heat pouring out of that fan/heatsink. And
that heatsink is still hot to touch. I can already see my electric bill
going up from running that thing. But while sitting in the house on a
recent HOT day (without an AC), I started to think about using that
heat. In the winter, it will help heat the house, so it's getting used.
But in hot weather, it's not only going to waste, but actually making it
more uncomfortable for me.

I have an electric water heater, and it's not too far from my computer.
I got to thinking about designing a heat sink with copper tubing going
thru it, which is then fed to a pre-heater water tank, which is then fed
to othe intake of the regular water heater. I actually did this same
thing with some solar panels once and it cut my gas bill in half.
(Another house, with a gas water heater).

I'm just wondering if anyone has ever seen or heard of anyone doing
anything like this. Besides using that wasted heat to heat my water, it
should make the processor run more efficient, and thus increase computer
power. The more I read about processor cooling, the more it's apparent
that a CPU has more power when it's cold. And these P4 processors run
MUCH hotter than the P3 and lower. Heck, that fan on my P3 1ghz
machine is tiny in comparison, adn that heat sink is barely warm.

BTW, there is a vido on YT where the guy removed the fan from a P4 dual
core CPU and in seconds the CPU just explodes. That's crazy....
 
Ad

Advertisements

P

Paul

BTW, there is a vido on YT where the guy removed the fan from a P4 dual
core CPU and in seconds the CPU just explodes. That's crazy....

That's probably the video that was faked.

Modern processors (P4 included) have THERMTRIP.
If the processor gets too hot, it turns off
the power supply. Just like that. THERMTRIP is
set about 20C above the throttle temperature.
First, the processor attempts to throttle,
and if the temperature is still climbing, it
can turn off the power. THERMTRIP is not "programmable"
and the processor doesn't get a say about leaving
the power on. THERMTRIP is a low level hardware function.

THERMTRIP triggers, before the material used to
package the processor, gets damaged. That's why the
thing has protection. The silicon might be able
to take another 35C or so, without being permanently
damaged. But some of the packaging materials are
degraded at a lower temperature, and so THERMTRIP is
set to help protect them.

Paul
 
C

casey.o

That's probably the video that was faked.

Modern processors (P4 included) have THERMTRIP.
If the processor gets too hot, it turns off
the power supply. Just like that. THERMTRIP is
set about 20C above the throttle temperature.
First, the processor attempts to throttle,
and if the temperature is still climbing, it
can turn off the power. THERMTRIP is not "programmable"
and the processor doesn't get a say about leaving
the power on. THERMTRIP is a low level hardware function.

THERMTRIP triggers, before the material used to
package the processor, gets damaged. That's why the
thing has protection. The silicon might be able
to take another 35C or so, without being permanently
damaged. But some of the packaging materials are
degraded at a lower temperature, and so THERMTRIP is
set to help protect them.

Paul

I know someone who has a Dell P4 dual core (piece of shit). The CPU fan
died, and that computer would only run a few monutes and shut off. I
was asked to fix it. I quickly found the fan was dead. I took a window
fan and some cardboard and made a funnel pointed the tip of the funnel
at the heat sink and blew the fan into it. The computer continued to
run for hours. Dell wanted almost $100 for a fan, and it's a special
scoop design, so not any fan can be used. But I found a used one on
Ebay for around $30. I dont understand that computer, it's slower than
my P3 1ghz (much slower). I thought that CPU was damaged from the heat,
but I ended up getting two similar conmputers given to me, and they are
all like that. That's why I will never buy a Dell. They dont run
right, and the parts are outrageously costly, and custom made so you
cant use common parts.

I kind of wondered about that video. It looks kind of fake. The CPU
flies right in the air with a bang, and ends up in about 5 or 6 pieces.
There is also a hole under the CPU right thru the motherboard. If the
CPU flys upward, there should not be a hole in the MB. I almost think
someone fired a gun from under the table, but probably just a powder
charge, but with a good amount of powder. I just cant see how the whole
CPU would suddenly explode and the whole thing fly up in the air, and
end up in pieces. But a powder gunshot would do all of that. It even
sounds like a gunshot. Plus it explodes about 5 seconds after he
removed the cooler. It just wont get that hot that fast.

I like the one where the guy runs 240V directly to a motherboard. Then
shows the video in slow motion. The caps blow first, and it turns into
a fireball. Pretty spectacular, adn I do believe that is real.

Just search YT for "exploding computer".

Some people will do anything to get a "hit" video on there. They are
paid $5 for every 1000 views. So, something that "goes viral" can make
some big bucks. Like that song Ghandam Style by Psy. He got around
$60,000 for that video (so far). One of my favorites that has gone
viral, is a commercial from the Shetland Islands, where they have a
Shetland pony do a moon dance. That is really cute. (search for
danceponydance) (exact spelling). There are two versions, regular and
christmas. Same video, changed scenery. The commercial is for a soem
sort of software company, but the commercial is great.
 
P

Paul in Houston TX

There's a few videos on Youtube where guys are overclocking Pentium 4
systems and using nitrogen to cool them. That does seem a bit
excessive, not to mention crazy. But one of them claims they are
running a 4ghz processor at almost 8ghz.

I'm not intending to do any of that, nor even overclock, because I know
that can do a lot of damage. I did have an old Pentium 1, that someone
overclocked from 133mhz to over 200mhz, and that machine kicked butt.

However, running my latest computer, which is a Pentium 4, 3ghz, I can
not believe the amount of heat pouring out of that fan/heatsink. And
that heatsink is still hot to touch. I can already see my electric bill
going up from running that thing. But while sitting in the house on a
recent HOT day (without an AC), I started to think about using that
heat. In the winter, it will help heat the house, so it's getting used.
But in hot weather, it's not only going to waste, but actually making it
more uncomfortable for me.

I have an electric water heater, and it's not too far from my computer.
I got to thinking about designing a heat sink with copper tubing going
thru it, which is then fed to a pre-heater water tank, which is then fed
to othe intake of the regular water heater. I actually did this same
thing with some solar panels once and it cut my gas bill in half.
(Another house, with a gas water heater).

I'm just wondering if anyone has ever seen or heard of anyone doing
anything like this. Besides using that wasted heat to heat my water, it
should make the processor run more efficient, and thus increase computer
power. The more I read about processor cooling, the more it's apparent
that a CPU has more power when it's cold. And these P4 processors run
MUCH hotter than the P3 and lower. Heck, that fan on my P3 1ghz
machine is tiny in comparison, adn that heat sink is barely warm.

BTW, there is a vido on YT where the guy removed the fan from a P4 dual
core CPU and in seconds the CPU just explodes. That's crazy....

Water cooled cpu. Put the radiator outside in the summer.
My main comp is o/c'd about 33% on cpu, ram, vid, bridges, etc.
It generates a LOT of heat.
Water cooled cpu. Put the radiator outside in the summer.
Bring it in for the winter to use as a space heater.
Eight fans including an A/C powered 10" mainboard fan.
Lights dim when the comp turns on.
 
C

casey.o

Water cooled cpu. Put the radiator outside in the summer.
My main comp is o/c'd about 33% on cpu, ram, vid, bridges, etc.
It generates a LOT of heat.
Water cooled cpu. Put the radiator outside in the summer.
Bring it in for the winter to use as a space heater.
Eight fans including an A/C powered 10" mainboard fan.
Lights dim when the comp turns on.

If there was a radiator, that would be an option. I'm sure I could make
something to work as a radiator, but if I'm going that far, I may as
well just use it to heat my hot water. The thing is that it will only
extract the heat from the CPU, not the other chips. But the CPU is the
majority of the heat produced.

If your lights dim when you power up that computer, you must have a big
electric bill. (and a hefty power supply).
 
P

Paul

If there was a radiator, that would be an option. I'm sure I could make
something to work as a radiator, but if I'm going that far, I may as
well just use it to heat my hot water. The thing is that it will only
extract the heat from the CPU, not the other chips. But the CPU is the
majority of the heat produced.

If your lights dim when you power up that computer, you must have a big
electric bill. (and a hefty power supply).

That's called "inrush".

When you flip the switch on the PSU to the ON position,
the main capacitor is empty. On one side, is the big bad
power company, 120V at a zillion amps. There is a bridge
rectifier, to convert that to DC. And that connects
to the capacitors. The capacitors are filled up in one or
two AC cycle times. The current can be a pulse of around
40 amps.

In this picture, C5 and C6 are the main capacitors. They
hold smoothed DC high voltage, to run the primary side
of the power supply.

http://www.pavouk.org/hw/en_atxps.html

The device NTCR1, is a positive temperature coefficient
resistor. When it is cold, it limits the peak current flow.
When the power supply has been on for a second or two,
NTCR1 is a lot warmer, and the warmer it gets, the lower
its resistance gets. The power it dissipates, drops
to a lower level. That component is present, purely
to "tame" the inrush transient.

So when you see the lights flicker, it's because
the C5 and C6 just got charged up. NTCR1 controlled
the current to manageable levels (40A maybe). The
breaker in your panel doesn't trip, because the
time constant of the transient is quite short. So
no breaker trips.

If the supply were to sustain such a behavior for a
couple of seconds, then there would be thermal
damage or breaker trips. But because the current pulse
is one or two cycles of 60Hz juice, the time is quite short.

The only end-user take away from this little description,
is you should not toggle the switch on the power supply
too rapidly. After you turn off an ATX PSU at the back,
wait about 30 seconds for NTCR1 to cool off. It should be
cool, so it's resistance is higher, and the slug of
current flowing will have a lower amplitude. One poster
to these groups, managed to blow a supply when flipping
the switch "oh, about forty times" in rapid succession.
So, um, don't do that.

The power supply in that diagram, also has an actual fuse
in it. But that fuse just never seems to blow. I can't remember
the last time someone traced to that fuse, and found it
blown. Maybe NTCR1 turns into a fireball, before the
fuse F1 pops ? :)

Paul
 
Ad

Advertisements

G

Gene Wirchenko

On Wed, 04 Jun 2014 15:58:03 -0400, (e-mail address removed) wrote:

[snip]
I kind of wondered about that video. It looks kind of fake. The CPU
flies right in the air with a bang, and ends up in about 5 or 6 pieces.
There is also a hole under the CPU right thru the motherboard. If the
CPU flys upward, there should not be a hole in the MB. I almost think

If the CPU flies up, then force is being applied down. Enough
force could do that sort of damage to the motherboard. It does not
seem likely though.

[snip]

Sincerely,

Gene Wirchenko
 
B

Bob F

If there was a radiator, that would be an option. I'm sure I could
make something to work as a radiator, but if I'm going that far, I
may as well just use it to heat my hot water. The thing is that it
will only extract the heat from the CPU, not the other chips. But
the CPU is the majority of the heat produced.

The problem is that the water temperature is the LOWEST temp your CPU can be
operated at with your idea. As the water gets warmer, so does the CPU.
 
J

J. P. Gilliver (John)

In message <[email protected]>,

There used to be one on TomsHardware (that some people discredited but
seemed plausible to me), where the fan was removed from a running Intel
and AMD processor to see what happened; in one case it throttled itself
back to very low speed but otherwise did no damage, in the other rapidly
shot up to hundreds of degrees (probably F), and did actually damage the
motherboard. But that was well before the days of multicore CPUs, I
think before one of the manufacturers had incorporated a sensor and
throttling (I suspect that video might have been instrumental in making
them do so). It wasn't the explosive one you describe.

That's good to know.
[]
all like that. That's why I will never buy a Dell. They dont run
right, and the parts are outrageously costly, and custom made so you
cant use common parts.

I've used enough Dells (especially in the corporate environment) to
disagree. Yes, certainly, they do sometimes use special parts, though by
no means always; their more recent ones, I'd say the only uncommon is
perhaps the ducting - one I had built for me (well for blind friends I
was buying it for) was assembled before my eyes, and used fairly
standard parts. (In fact as they wanted more memory than he had to hand
the assembler had to go buy some memory modules from one of the others
at the show!)
[]
--
J. P. Gilliver. UMRA: 1960/<1985 MB++G()AL-IS-Ch++(p)[email protected]+H+Sh0!:`)DNAf

/Pride and Prejuice/ seems to be remade every ten years - the only real
difference being a new Mr Darcy in a new shirt with fewer buttons. - Marie
Hanerman, in Feedback, Radio Times, 13-19 November 2010.
 
J

J. P. Gilliver (John)

[/QUOTE]

(And shade it. I'm always surprised how many A/C units I see where the
outside radiator is in full sunlight: surely this must use sufficiently
more energy that the cost of a bit of sheet metal to shade it would soon
be recouped.)
[]
That's called "inrush".

Yes, I think PiH meant they just dim when (i. e. at the moment) he
powers it up, not all the time he has it on!
When you flip the switch on the PSU to the ON position,
the main capacitor is empty. On one side, is the big bad
power company, 120V at a zillion amps. There is a bridge
rectifier, to convert that to DC. And that connects
to the capacitors. The capacitors are filled up in one or
two AC cycle times. The current can be a pulse of around
40 amps.

Well, limited only by what the wiring and rectifiers can take, really:
basically it's a short!
[]
The device NTCR1, is a positive temperature coefficient
resistor. When it is cold, it limits the peak current flow.
When the power supply has been on for a second or two,
NTCR1 is a lot warmer, and the warmer it gets, the lower
its resistance gets. The power it dissipates, drops
[]
I think modern units sometimes also have an input _filter_, containing
coils and capacitors - partly as regulations to control the amount of
crud fed back into the mains/line (and maybe control power factor), as
well as limit inrush current.
[]
In the early days of switched-mode supplies, I had a big (for those
days) 13V 20A (IIRR) one (I was a radio ham and much such gear runs on
13.x V, as in car battery); that had little or no (no I think) inrush
limiting/filtering circuitry, and not infrequently blew its mains fuse,
even if I put in the highest available (UK - BS1363 - mains plugs have a
[BS1362] fuse in them, which commonly come in 3, 5, and 13A ratings,
though you can get 2 and I think 10).

Conversely, the sort with filters in can retain a charge: I've had a
belt off the pins of the plug of a piece of equipment, even some tens of
seconds after it's been unplugged from the supply.
--
J. P. Gilliver. UMRA: 1960/<1985 MB++G()AL-IS-Ch++(p)[email protected]+H+Sh0!:`)DNAf

/Pride and Prejuice/ seems to be remade every ten years - the only real
difference being a new Mr Darcy in a new shirt with fewer buttons. - Marie
Hanerman, in Feedback, Radio Times, 13-19 November 2010.
 
P

Paul

J. P. Gilliver (John) said:
In message <[email protected]>,

There used to be one on TomsHardware (that some people discredited but
seemed plausible to me), where the fan was removed from a running Intel
and AMD processor to see what happened; in one case it throttled itself
back to very low speed but otherwise did no damage, in the other rapidly
shot up to hundreds of degrees (probably F), and did actually damage the
motherboard. But that was well before the days of multicore CPUs, I
think before one of the manufacturers had incorporated a sensor and
throttling (I suspect that video might have been instrumental in making
them do so). It wasn't the explosive one you describe.

That's good to know.
[]
all like that. That's why I will never buy a Dell. They dont run
right, and the parts are outrageously costly, and custom made so you
cant use common parts.

I've used enough Dells (especially in the corporate environment) to
disagree. Yes, certainly, they do sometimes use special parts, though by
no means always; their more recent ones, I'd say the only uncommon is
perhaps the ducting - one I had built for me (well for blind friends I
was buying it for) was assembled before my eyes, and used fairly
standard parts. (In fact as they wanted more memory than he had to hand
the assembler had to go buy some memory modules from one of the others
at the show!)
[]
 
Ad

Advertisements

P

Paul

Paul wrote:

Sorry about that, hit the wrong button.

There were two videos. There is the one on Tomshardware,
of an early Athlon burning up.

And the one involving an Intel processor, much later,
was one done just to get hits on Youtube. It involved
inserting a pyrotechnic underneath the motherboard,
detonating it, then claiming the "processor blew up".
The materials in a processor, are not suited to
exploding on their own.

The most exotic failure I've seen a picture of, is
a motherboard with a copper plane short inside, in
the CPU socket area. The area all around the
CPU socket got so hot, the fiberglass PCB material
became discolored (dark brown). It didn't catch fire or anything.
That's the most "violent" legit failure I've seen.
The only thing that even comes close, is failures
involving MOSFETs, where the MOSFET gets so hot, the
solder melts. The worst that can happen there, is the
MOSFET falls off the motherboard.

*******

It's my suspicion, the Athlon always had the raw materials
for making an emergency shutdown solution. I think the original
Athlon had a thermal diode, just like the later ones did. It
just wasn't hooked to anything. That's the sort of diode the
engineers would use in the lab, when verifying thermal
solutions (what size of heatsink to use). If you need a Tj,
a thermal diode in the silicon die, is perfect for that.
You can get a Tcase, by putting a thermistor into the heatsink.

The designs they did use in the first generation, were
intended for fan failures. In that case, the temperature
shoots up relatively slowly, because the CPU has to heat up
the heatsink above it. A typical scenario would be, a user
assembles their own computer, forgets to hook up the CPU
fan cable, and the thermal solution would then turn off the
power in time.

What they hadn't planned on, was the fact that the clamp
on the heatsink applied too much retention force. And the
single tab on the socket could snap off, as a result of that
pressure. The heatsink would fall off... and the CPU die could shoot
above 100C in around a second or so. The thermal monitoring
was being done in software, and the software would frequently
crash (because the CPU was overheated), before the software
could turn off the power.

Both AMD and Intel, on their modern THERMTRIP implementations,
use a hardware-only solution. Even if the CPU is crashed, the
THERMTRIP can still turn off the power if the computer
has an overheating CPU. So we no longer have to worry about it.
It responds virtually instantly, because the sensor is embedded
in the thing generating the heat. The otiginal Athlon design,
there was significant "lag" between the sensor and the silicon
die that was overheating. The silicon die could be "cooked", before
the thermistor under the socket was even warm.

Paul
 
P

Paul

J. P. Gilliver (John) said:
I think modern units sometimes also have an input _filter_, containing
coils and capacitors - partly as regulations to control the amount of
crud fed back into the mains/line (and maybe control power factor), as
well as limit inrush current.
[]
In the early days of switched-mode supplies, I had a big (for those
days) 13V 20A (IIRR) one (I was a radio ham and much such gear runs on
13.x V, as in car battery); that had little or no (no I think) inrush
limiting/filtering circuitry, and not infrequently blew its mains fuse,
even if I put in the highest available (UK - BS1363 - mains plugs have a
[BS1362] fuse in them, which commonly come in 3, 5, and 13A ratings,
though you can get 2 and I think 10).

Conversely, the sort with filters in can retain a charge: I've had a
belt off the pins of the plug of a piece of equipment, even some tens of
seconds after it's been unplugged from the supply.

When I designed a power supply with that problem present (inrush
for a long time), I had a three position switch on the front
of the unit. OFF-Standby-ON. In the Standby position, the output
(load) is disconnected, the rectifiers are connected via a current
limiting resistor to the capacitor bank (0.25F). Once the main
capacitor is charged, the user rotates the control to the ON
position. In the ON position, the current limiting resistor
is shorted out (goes from a "soft" to a "hard" connection).
And the load gets connected to the supply.

Never had a fuse blow with that setup. Of course, if someone else
had access to my little home project, they would undoubtedly rotate
the knob too fast, and blow it. Like all my projects of
that sort, the front panel never came with a warning about
how to use the controls properly :)

Paul
 
Ad

Advertisements


Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.

Ask a Question

Top