Connect with us

Heat Rise in a rack mounted chassis....

Discussion in 'Electronic Basics' started by [email protected], Sep 16, 2005.

Scroll to continue with content
  1. Guest

    I'm looking for some advise on heat rise. If I have a chassis (10 x 17
    x 19) and I mount two components that dissipate 385W and 256W
    respectively, and then have a power supply that supplies those
    components power to operate, then I'm assuming (if the power supply is
    80% efficient) that my total heat load is somewhere in the 1400W range
    (641W + 770W (supply input pwr)). First off, is that a correct
    assumption of the heat loading or am I making this way too simple?

    Next, based on this heat load, the heat rise is a function of the total
    surface area of the chassis (calculated to be about 9.5 sqft). Based
    on that, the Watts/square ft = 147W/sqft (1400/9.5 = 147). Curves show
    that the heat rise will be off the charts since the charts only go up
    to 20W/sqft for something close to 80F temp difference (heat rise as a
    function of the ambient air temp)

    Does anyone see errors in my calculations or is this the expected heat
    rise for this scenario?
     
  2. John  Larkin

    John Larkin Guest


    Your box is roughly half the size of a toaster oven, and if you set it
    to "broil" the power is in the 1/2 ballpark, too. They get pretty hot
    on the outside, as I recall.

    John
     
  3. Rich Grise

    Rich Grise Guest

    Well, 10 x 17 x 19 is about twice the size of any toaster oven I've ever
    seen - maybe you're thinking of an ordinary oven?

    In either case, unless the OP is actually _making_ an oven (in which
    case he'd be better off to insulate it) he'd better poke some holes
    in that cabinet and add a bank of fans. With 1400 Watts inside a sealed,
    10 x 17 x 19 aluminum rack-panel cabinet, he could easily cook dinner
    on top of it.

    Cheers!
    Rich
     
  4. John  Larkin

    John Larkin Guest

    Yeah, I meant twice.
    My point.

    Something like 100 CFM air flow would carry off 1400 watts without too
    much exaust air temp rise. But the devil's in the details.


    John
     
  5. John - KD5YI

    John - KD5YI Guest


    Assuming your components of 385W and 256W are resistors (or equivalent),
    your total heat load caused by the components is 641W. You state that the
    device which supplies them power is 80% efficient. Therefore, the power
    supply is itself adding 160W to the heat load bringing the total heat load
    to 801W. (Total watts = load/efficiency)


    I agree with your surface area calculation, but I use a slightly different
    rule-of-thumb involving square inches. You have 801W/1366 sqin which gives
    about .59w/sq-in. I use the crude rule of about 100 times the w/sqin as the
    temperature rise (because it is easy to remember. I think it may be higher
    than that). So I would guess at about 60C or so rise in temperature of the
    surface of the box.

    Re-do your estimate based on the lower heat load (85W/sqft) and see if our
    results seem to be reasonable.

    John
     
  6. Dan Akers

    Dan Akers Guest

    "I'm looking for some advise on heat rise. If I have a chassis (10 x 17
    x 19) and I mount two components that dissipate 385W and 256W
    respectively, and then have a power supply that supplies those
    components power to operate, then I'm assuming (if the power supply is
    80% efficient) that my total heat load is somewhere in the 1400W range
    (641W + 770W (supply input pwr)). First off, is that a correct
    assumption of the heat loading or am I making this way too simple?
    Next, based on this heat load, the heat rise is a function of the total
    surface area of the chassis (calculated to be about 9.5 sqft). Based on
    that, the Watts/square ft = 147W/sqft (1400/9.5 = 147). Curves show that
    the heat rise will be off the charts since the charts only go up to
    20W/sqft for something close to 80F temp difference (heat rise as a
    function of the ambient air temp)
    Does anyone see errors in my calculations or is this the expected heat
    rise for this scenario?"
    ____________________________________
    Re;
    I'm wondering what sort of components these are; are they discrete,
    passive, components, like resistors, or are they circuit boards? Is the
    chassis vented in any way, or are you relying on the heat to conduct
    through the sides of the chassis? What is the max. temperature rise
    allowed? In what sort of environment will it operate?
    In any event, I don't see how you get 1400W. The total for the
    components is 641W and the 20% of the power supply is 128W giving a
    total of 769W. Never-the-less, in my opinion, the cabinet is going to
    need some forced ventilation to obtain a viable temperature rise if
    semi-conductors and such are exposed inside this chassis. If you assume
    1000W dissipation inside the chassis, you're going to need to move about
    320cfm to obtain a 10F temp. rise; for a 20F rise you need 160cfm; and
    so on.
    I do think that your assumption for natural convection around the
    cabinet based solely on it's total surface area is overly simplistic in
    that the heat flux will certainly not be evenly distributed over the
    surface area of the chassis if the components are simpling naturally
    convecting and radiating within. Most of the heat will convect, and to
    a lesser extent, radiate off of the upper portion of the chassis, with
    lesser amounts from the sides and ends and even less from the bottom.

    -Dan Akers
     
  7. Jasen Betts

    Jasen Betts Guest

    Just looking at the figure 1400W tells me you've got a firestarter there.
    1400W - that's a fan heater... tring to cool something that powerful
    without using a big fan and/or a wheelbarrow load of heatsinking is
    asking for trouble.

    So what will it be? a radio transmitter or something?

    Bye.
    Jasen
     
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 (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day

-