What size water pump?

Archived from groups: alt.comp.hardware.overclocking (More info?)

I have an Intel P4 2.8 GHz and an ATI 9600 All-in-wonder GPU that I want to
cool.
I plan to use a water block on each processor. My cooler is a small 5 cf
refrigerator.
With a 3 gallon reservoir. There is a 1/20hp compressor.

I am looking at Petsmarts web page. They have a pump rated at 80 gal. per
hour for
$15.99 and another pump rated at 140 gal. per hour for $22.99. The added
beauty of
my design with a submersable pump in the water reservoir I can simply add
ice if there is
to much load on the compressor.

I don't know the total btu's put out by these chips when over clocked.

Thanks in advance!

Robert Miller
8 answers Last reply
More about what size water pump
  1. Archived from groups: alt.comp.hardware.overclocking (More info?)

    Your 5 cubic foot refrigerator may not do the job. Basically a refrigerator
    that small has just enough cooling ability to handle the heat leakage
    through the insulation (as can be seen in how long it takes to make a tray
    of ice), opening the door once every few hours, and eventually cooling room
    temperature items down over a few hours span. When you add up the heat load
    produced by a CPU and GPU, along with the heat leakage into the water tubes
    and refrigerator the compressor/evaporator may not have the capacity.

    A BTU is a measure of ENERGY. A Watt is a measure of POWER. Thus, 1 BTU
    equals 1055.056 Watt seconds (a BTU is the amount of energy needed to raise
    the temperature of one pound of water by one degree F.)

    ALL the power consumed by CPU, chipset, and GPU turns into heat. If, for
    example, a CPU draws 100 Watts, that's 100 Watts X 60 minutes/hr X 60
    seconds/minute = 360000 Watt/seconds heat energy produced per hour.
    Converting to BTU'/s, that is 360000/1055 = ~340 BTU/hr. Add another 75
    Watts, perhaps, for the GPU and chipset, depending on the graphics card.
    Also consider that a submergable pump is cooled by the water. Figure on a
    total of 200 Watts turned into heat (680 BTU/hr.)

    Now, for your pump question; 680 BTU will raise the temperature of 680
    pounds of water 1 degree F. A gallon of water weighs 8.382 pounds, so 680
    pounds = ~ 81 gallons of water per hour will be raised in temperature by 1
    degree F. Circulation of 80 gallons per hour is more than enough for
    cooling as a temperature rise of 2 or 4 degrees F is likely less important
    than the design and construction of the heat exchanger on the CPU.

    The questions you should be asking yourself:

    *Does the refrigerator have the cooling capacity? (Put a gallon jug of
    water at 120 F in the empty refrigerator and check the temperature after one
    hour. If the temperature of the water is down to 40 F, then then 680 BTUs
    of cooling was done.)

    *Is the pump suitable for use at 40 F rather than room temperature?

    *Is the pump capable of working against the friction of the water path?

    *Is the water circulation path sealed? If not, then the difference in
    height of the various components will add to the pump work load, reducing
    the performance.

    *Is the pump reliable?

    *Will the heat exchanger you place inside the refrigerated compartment
    handle the heat load?

    *Do you really want to have ONE refrigerator tied up cooling your computer
    system while your shlep ice from another refrigerator to another?

    On the other hand, if you are just going for a record, rather than practical
    use, try a mixture of cracked dry ice and isopropyl alcohol as a cooling
    bath, replacing the dry ice as necessary.
    --
    Phil Weldon, pweldonatmindjumpdotcom
    For communication,
    replace "at" with the 'at sign'
    replace "mindjump" with "mindspring."
    replace "dot" with "."

    "Robert Miller" <stargazz@hotmail.com> wrote in message
    news:d075$421e45ec$a227fd59$18914@ALLTEL.NET...
    > I have an Intel P4 2.8 GHz and an ATI 9600 All-in-wonder GPU that I want
    to
    > cool.
    > I plan to use a water block on each processor. My cooler is a small 5 cf
    > refrigerator.
    > With a 3 gallon reservoir. There is a 1/20hp compressor.
    >
    > I am looking at Petsmarts web page. They have a pump rated at 80 gal. per
    > hour for
    > $15.99 and another pump rated at 140 gal. per hour for $22.99. The added
    > beauty of
    > my design with a submersable pump in the water reservoir I can simply add
    > ice if there is
    > to much load on the compressor.
    >
    > I don't know the total btu's put out by these chips when over clocked.
    >
    > Thanks in advance!
    >
    > Robert Miller
    >
    >
  2. Archived from groups: alt.comp.hardware.overclocking (More info?)

    Thanks rms & Phil! I had hoped it really would be this easy, but only a
    week or two
    into learning there was a serious option to water cool my PC. I'll do much
    more
    reading and thinking this thing through.

    Not that it will matter, anybody have the conversion of hp to watts? I'd
    guess off hand
    that 200 Watts is more than 1/20 hp. I think I will go with something
    different.

    Thanks again!

    Robert


    "Phil Weldon" <notdisclosed@example.com> wrote in message
    news:EHxTd.6530$Ba3.988@newsread2.news.atl.earthlink.net...
    > Your 5 cubic foot refrigerator may not do the job. Basically a
    refrigerator
    > that small has just enough cooling ability to handle the heat leakage
    > through the insulation (as can be seen in how long it takes to make a tray
    > of ice), opening the door once every few hours, and eventually cooling
    room
    > temperature items down over a few hours span. When you add up the heat
    load
    > produced by a CPU and GPU, along with the heat leakage into the water
    tubes
    > and refrigerator the compressor/evaporator may not have the capacity.
    >
    > A BTU is a measure of ENERGY. A Watt is a measure of POWER. Thus, 1 BTU
    > equals 1055.056 Watt seconds (a BTU is the amount of energy needed to
    raise
    > the temperature of one pound of water by one degree F.)
    >
    > ALL the power consumed by CPU, chipset, and GPU turns into heat. If, for
    > example, a CPU draws 100 Watts, that's 100 Watts X 60 minutes/hr X 60
    > seconds/minute = 360000 Watt/seconds heat energy produced per hour.
    > Converting to BTU'/s, that is 360000/1055 = ~340 BTU/hr. Add another 75
    > Watts, perhaps, for the GPU and chipset, depending on the graphics card.
    > Also consider that a submergable pump is cooled by the water. Figure on a
    > total of 200 Watts turned into heat (680 BTU/hr.)
    >
    > Now, for your pump question; 680 BTU will raise the temperature of 680
    > pounds of water 1 degree F. A gallon of water weighs 8.382 pounds, so 680
    > pounds = ~ 81 gallons of water per hour will be raised in temperature by 1
    > degree F. Circulation of 80 gallons per hour is more than enough for
    > cooling as a temperature rise of 2 or 4 degrees F is likely less important
    > than the design and construction of the heat exchanger on the CPU.
    >
    > The questions you should be asking yourself:
    >
    > *Does the refrigerator have the cooling capacity? (Put a gallon jug of
    > water at 120 F in the empty refrigerator and check the temperature after
    one
    > hour. If the temperature of the water is down to 40 F, then then 680 BTUs
    > of cooling was done.)
    >
    > *Is the pump suitable for use at 40 F rather than room temperature?
    >
    > *Is the pump capable of working against the friction of the water path?
    >
    > *Is the water circulation path sealed? If not, then the difference in
    > height of the various components will add to the pump work load, reducing
    > the performance.
    >
    > *Is the pump reliable?
    >
    > *Will the heat exchanger you place inside the refrigerated compartment
    > handle the heat load?
    >
    > *Do you really want to have ONE refrigerator tied up cooling your computer
    > system while your shlep ice from another refrigerator to another?
    >
    > On the other hand, if you are just going for a record, rather than
    practical
    > use, try a mixture of cracked dry ice and isopropyl alcohol as a cooling
    > bath, replacing the dry ice as necessary.
    > --
    > Phil Weldon, pweldonatmindjumpdotcom
    > For communication,
    > replace "at" with the 'at sign'
    > replace "mindjump" with "mindspring."
    > replace "dot" with "."
    >
    > "Robert Miller" <stargazz@hotmail.com> wrote in message
    > news:d075$421e45ec$a227fd59$18914@ALLTEL.NET...
    > > I have an Intel P4 2.8 GHz and an ATI 9600 All-in-wonder GPU that I want
    > to
    > > cool.
    > > I plan to use a water block on each processor. My cooler is a small 5
    cf
    > > refrigerator.
    > > With a 3 gallon reservoir. There is a 1/20hp compressor.
    > >
    > > I am looking at Petsmarts web page. They have a pump rated at 80 gal.
    per
    > > hour for
    > > $15.99 and another pump rated at 140 gal. per hour for $22.99. The
    added
    > > beauty of
    > > my design with a submersable pump in the water reservoir I can simply
    add
    > > ice if there is
    > > to much load on the compressor.
    > >
    > > I don't know the total btu's put out by these chips when over clocked.
    > >
    > > Thanks in advance!
    > >
    > > Robert Miller
    > >
    > >
    >
    >
  3. Archived from groups: alt.comp.hardware.overclocking (More info?)

    1 Horsepower = 746 Watts

    A compressor/evaporator cooler (phase change), as in a refrigerator, is a
    heat pump and will move several times as much heat energy as it consumes.

    --
    Phil Weldon, pweldonatmindjumpdotcom
    For communication,
    replace "at" with the 'at sign'
    replace "mindjump" with "mindspring."
    replace "dot" with "."
  4. Archived from groups: alt.comp.hardware.overclocking (More info?)

    > A compressor/evaporator cooler (phase change), as in a refrigerator, is a
    > heat pump and will move several times as much heat energy as it consumes.

    refrigerators are not designed to have an active heatsource inside them.
    He better find out what life expectancy will result from this setup, because
    as soon as it breaks down, the whole experiment is over: it won't be worth
    fixing.

    Also, simply putting a radiator inside of a refrigerator compartment is
    incredibly inefficient in terms of heat transfer, i.e., he will not be
    seeing the results that you calculate above, but only some percentage
    thereof. At the very least, fans would need to be installed on the
    radiator, or the radiator would need to be bolted onto the cooling coils
    inside the fridge somehow (good luck on that).

    From a practical standpoint this project is a nightmare, but hay! let us
    know if you actually build it. Someone has to take the plunge!

    rms
  5. Archived from groups: alt.comp.hardware.overclocking (More info?)

    "rms" wrote
    ' refrigerators are not designed to have an active heatsource inside them.
    He better find out what life expectancy will result from this setup, because
    as soon as it breaks down, the whole experiment is over: it won't be worth
    fixing.'

    'Also, simply putting a radiator inside of a refrigerator compartment is
    incredibly inefficient in terms of heat transfer, i.e., he will not be
    seeing the results that you calculate above, but only some percentage
    thereof. At the very least, fans would need to be installed on the
    radiator, or the radiator would need to be bolted onto the cooling coils
    inside the fridge somehow (good luck on that). From a practical standpoint
    this project is a nightmare, but hay! let us know if you actually build it.
    Someone has to take the plunge!'

    We both pointed out potential problems, but ...

    Refrigerators DO have to be designed to cool warm objects, warm air that
    comes through an open door, and heat leakage thrrough the insultation.
    There is no difference in quality between that heat load and the load of an
    'active heatsource.' It is how much heat load that is in question.

    As for 'incredibly inefficient in terms of heat transfer', that is beside
    the point. The total heat transfer is what counts, not the efficiency.
    Efficiency in this case can always be made up by more tranfer surface. What
    counts is the heat load the refrigerator must handle. Not to mention that
    refrigerators have active air circulation and/or cold surfaces (in the case
    of small refrigerators.) It is not a question 'only some percentage
    thereof', but questions of capability, reliability, bulk, noise,
    inconvenience ...

    It's been done before, and described in this newsgoup.
    --
    Phil Weldon, pweldonatmindjumpdotcom
    For communication,
    replace "at" with the 'at sign'
    replace "mindjump" with "mindspring."
    replace "dot" with "."
  6. Archived from groups: alt.comp.hardware.overclocking (More info?)

    "Phil Weldon" <notdisclosed@example.com> wrote in message
    news:EHxTd.6530$Ba3.988@newsread2.news.atl.earthlink.net...

    > ALL the power consumed by CPU, chipset, and GPU turns into heat. If, for
    > example, a CPU draws 100 Watts, that's 100 Watts X 60 minutes/hr X 60
    > seconds/minute = 360000 Watt/seconds heat energy produced per hour.
    > Converting to BTU'/s, that is 360000/1055 = ~340 BTU/hr. Add another 75
    > Watts, perhaps, for the GPU and chipset, depending on the graphics card.
    > Also consider that a submergable pump is cooled by the water. Figure on a
    > total of 200 Watts turned into heat (680 BTU/hr.)
    >
    >

    So if I place a $85 5000BTU conditioner outside my computer
    room (in the shade, I'm in central Texas after all) and make a cowling
    to direct the air flow through a couple of heater cores and back to the
    AC; that might be a little over kill? A copper reservoir would also be
    positioned inside the fiberglass sandwich cowling. Insulated 3/4"
    PVC pipes to distribute the distilled water to and from my computer(s).

    Totally silent, (if I can find, or make, a water cooled power supply
    that doesn't scare me.) , with all the noise making parts outside. A
    constant supply of very chilled water.

    Luck;
    Ken
  7. Archived from groups: alt.comp.hardware.overclocking (More info?)

    "Matt" <matt@themattfella.zzzz.com> wrote in message
    news:zGGTd.10585$hV7.10225@news02.roc.ny...
    > Robert Miller wrote:
    > > I have an Intel P4 2.8 GHz and an ATI 9600 All-in-wonder GPU that I want
    to
    > > cool.
    >
    > Probably you will need liquid nitrogen cooling for that rig.

    I like the idea of cracked dry ice and isopropyl alcohol solution.

    Both are cheap, but the risk and effort of that setup inside the CPU case is
    more than
    a little intimidating. I assume the temp would be near 0'F if not colder.

    I will need thick insulation, and likely other steps to prevent condensation
    buildup.

    Thanks!

    Robert
  8. Archived from groups: alt.comp.hardware.overclocking (More info?)

    You put the isopropyl alcohol (99%)/cracked dry ice slush OUTSIDE the system
    case and circulate a propylene glycol/water mix through it, then through a
    water block on the CPU and back.

    You do get a very low temperature - dry ice sublimes at - 78.5 C, 100%
    isopropyl alcohol is a liquid above - 88 C (sorta a sluggish liquid at -78.5
    C.) A pound of dry ice requires 256 BTU of heat input just to go from a
    solid at - 78.5 C to a gas at -78.5 C (that's 100 Watts input for ~ 45
    minutes.) Even over a long time period, if you insulate the piping, have
    good exhange between the inside and outside of the system case, and place a
    few stategic resistors near cold spots you can avoid condensation problems.

    I've tried it, but didn't have that good a CPU to work with at the time
    (Pentium III 1 GHz/100 MHz FSB.) Publix supermarkets in South Florida were
    selling dry ice for $1 US per pound.

    --
    Phil Weldon, pweldonatmindjumpdotcom
    For communication,
    replace "at" with the 'at sign'
    replace "mindjump" with "mindspring."
    replace "dot" with "."


    "Robert Miller" <stargazzr@hotmail.com> wrote in message
    news:182ea$423092c1$a227fddc$16732@ALLTEL.NET...
    > "Matt" <matt@themattfella.zzzz.com> wrote in message
    > news:zGGTd.10585$hV7.10225@news02.roc.ny...
    > > Robert Miller wrote:
    > > > I have an Intel P4 2.8 GHz and an ATI 9600 All-in-wonder GPU that I
    want
    > to
    > > > cool.
    > >
    > > Probably you will need liquid nitrogen cooling for that rig.
    >
    > I like the idea of cracked dry ice and isopropyl alcohol solution.
    >
    > Both are cheap, but the risk and effort of that setup inside the CPU case
    is
    > more than
    > a little intimidating. I assume the temp would be near 0'F if not colder.
    >
    > I will need thick insulation, and likely other steps to prevent
    condensation
    > buildup.
    >
    > Thanks!
    >
    > Robert
    >
    >
Ask a new question

Read More

Overclocking