Good heat but poor electrical conductor

Discussion in 'The Projects Forum' started by gdylp2004, Feb 25, 2012.

  1. gdylp2004

    Thread Starter Member

    Dec 2, 2011
    101
    0
    Hi folks,

    I've posted this question in another forum but getting no response at all, and hence I'm posting in here hoping for a better.

    I am currently doing a project and need to dissipiate heat away from a specific electronic component to the frame of the system, in my case, the frame would be the electronic board and the stands for the board.

    More importantly, I think the 4 cylindical stands is a good place to get the heat off to the surrounding since it is 1) a metal 2) a good surface area with adequate space apart from each stand.

    My first thought would be using a flexible tube or something which is good in bending and carry the heat away from the heat source to the stands. But this material has to be an electrical insulator BECAUSE the frame could not get potentially energised to 100V, yes, the heat source unfortunately has a potential difference (w.r.t. GND) of +100V.

    I've considered epoxy or some thermal gel, but the problem is these doesn't have the physical property I desired because I can't just simply apply the gel across the heat source to the stands as it'll be to messy since it looks pretty much like a liquid.

    So does anyone has a clue which material is good for me to use?

    Probably, something which has:

    1) Good heat carrying property
    2) Extremely poor electrical conductor (or good elect insulator)
    3) well-definable shape (definitely not liquid, or gel-like substance)

    Please see attached for the placement of the heat medium.

    Thank you in advance!
     
  2. crutschow

    Expert

    Mar 14, 2008
    12,977
    3,220
    Beryllium oxide (BeO) (beryllia) is one of the best thermal conductors that is an electrical insulator (only diamond is better but is not often used for that purpose for obvious reasons ;)) and is commonly used where such a material is needed.

    Its main problem is that it can not be cut or machined expect in a controlled environment since its dust is hazardous. So it must be purchased in the shape you need.

    Another common conductive insulator is aluminum oxide (alumina). It has a poorer heat conduction than beryilla (about 10 times worse), but is cheaper and not hazardous.
     
    panic mode likes this.
  3. luvv

    Member

    May 26, 2011
    186
    31
  4. jimkeith

    Active Member

    Oct 26, 2011
    539
    99
    Check out aluminum nitride it is almost as good as BeO:
    http://www.anceram.com/pdf/aln_artikel_bull.pdf

    Make the bulk of your dissipating structure out of aluminum and provide the electrical insulation it via insulating washers and insulating fasteners. Surfaces must be flat because both beryllium oxide and aluminum nitride are quite brittle. Also, coat interface surfaces with conductive thermal grease as well. If you make the mating surface as large surface area as possible, you may be able to get away with aluminum oxide insulator(s)--much cheaper and more commonly available.
     
  5. gdylp2004

    Thread Starter Member

    Dec 2, 2011
    101
    0
    I am just glad that I didn't post at the wrong forum. Many replies here! =D

    Thank you! I'll try to do as many of the suggestion, and will post the result here after I plot the temp vs. time for all the diferent methods.

    Thanks again! But please, if you still have other suggestion, dont stop!
     
  6. jimkeith

    Active Member

    Oct 26, 2011
    539
    99
    Note that only about 200 linear ft/min air flow will approx halve your temp rise.
    This is easily done with a tube-axial fan.
     
  7. MrChips

    Moderator

    Oct 2, 2009
    12,415
    3,353
    Never be afraid to look outside the box.

    1) 100V is not such a high voltage that should prevent you from using metallic heat sinks with suitable insulation.

    2) I have had to cool various electronic instruments using tap water circulation.

    3) I have also used a heat pump, a thermo-electric cooler (TEC) to move the heat energy away from the source.
     
  8. gdylp2004

    Thread Starter Member

    Dec 2, 2011
    101
    0
    Hmm,

    If I am using a metallic conductor, say copper or brass with ring terminals as ends (see below). Consider the heat source to be a TO-220 MOSFET, would a thermal-conducting, electric-insulating washer with an adequate amount of heat compound between the joints solve the problem? Has anyone tried that?

    In this way, my choice of the heat medium (as shown in blue in the image from my first post) is no longer restricted by non-metallic good heat conductor only?

    [​IMG]
     
  9. MrChips

    Moderator

    Oct 2, 2009
    12,415
    3,353
    Did you know you can get MOSFETs in insulated TO-220 packages?
     
  10. gdylp2004

    Thread Starter Member

    Dec 2, 2011
    101
    0
  11. BSomer

    Member

    Dec 28, 2011
    433
    106
    It is a package like the TO-220 that does NOT have any connection to the exposed metal on the tab, or no exposed metal. Hence the term "insulated".
     
  12. MrChips

    Moderator

    Oct 2, 2009
    12,415
    3,353
  13. gdylp2004

    Thread Starter Member

    Dec 2, 2011
    101
    0
    Pardon me, I should have Google it before asking Whats that?

    Please correct me if Im wrong. Wouldn't such packaging (like the TO-220F) provides poorer heat dissipation mainly because the heat generated within the TO-220 has lower exposure, hence lower heat dissipating capability to the surrounding, and that is the opposite of my intention?

    Moreover, the +100V supply is connected to the system via a ring terminal screwed onto the tab of the TO-220 package. Hence, TO-220 insulated seems out?

    But more importantly because of the former reason rather than the latter.
     
  14. MrChips

    Moderator

    Oct 2, 2009
    12,415
    3,353
    Have you determined how much power will be generated by the transistor?
     
  15. KL7AJ

    AAC Fanatic!

    Nov 4, 2008
    2,039
    287
    Not too many materials qualify for this! Beryllium Oxide is one. Magnesium oxide (used in heat sink "goop") is another. But the best is DIAMOND! Unfortunately, it's a bit pricey for most electronics applications!

    Eric
     
  16. gdylp2004

    Thread Starter Member

    Dec 2, 2011
    101
    0
    Right now, my DC input power is about 103W and the output DC power is about 94W (to the electronic load).

    The Δ is about ~10W but all these losses doesnt appear on the MOSFET only. But it is very safe to assume that the MOSFET is the sole culprit for all the losses. I had in fact used a digital thermocouple to measure the surface temperature of the tab of the TO-220 packaging. The temp slowly climbs from about 23.7°C (room temp) to a max of 130°C and steadystates.

    The tolerated junction-to-case temp as recommended by the OEM is about 175°C, and therefore theortically speaking, I can do without a heatsink at all. But I wish to keep the integrity of the product to ensure it last as long as possible, so any heat sinking methods I would want to try.

    In addition, my test condition is in an air-con lab environment. The real application would be in space, and hence I would expect poorer heat dissipation (since now it is limited to only radiation). The junction-to-case temp would then expect to climb further, perhaps even greater than the recommended limit.

    I would put it in a vacuum chamber for testing once all thermal mgmt is done.
     
  17. MrChips

    Moderator

    Oct 2, 2009
    12,415
    3,353
    Whoa! Wait a minute here. If you are going to put this thing up into space you have a big problem with this.

    Why is the power generated on the MOSFET so high?
     
  18. gdylp2004

    Thread Starter Member

    Dec 2, 2011
    101
    0
    That is because the MOSFET is a high side buck converter switch with an avg current of 1A. Running at 100kHz should amount to quite some switching losses but I've selected the nMOS with the lowest Rds(on) to minimize conducting losses. But even with that, the TO-220 MOSFET rise up to 130°C ++.

    The full operation and design aspects of my 100W buck converter could be found here: http://forum.allaboutcircuits.com/showthread.php?t=62817
     
  19. THE_RB

    AAC Fanatic!

    Feb 11, 2008
    5,435
    1,305
    So you are designing a high performance SMPS to go into space but building it on stripboard and don't understand basic heatsinking for PSUs?

    Somethings a bit fishy! You could switch the 100v with a TO-220 old fashioned bipolar PNP which would only drop 0.5v Vce at 1amp, less than 0.5W dissipated. But you say you are using a specialist "low Rds" FET and it will get to 130°C??

    Can you put a .JPG of your schematic up here for people to discuss? I looked at the other thread but it's 10+ pages long.
     
  20. gdylp2004

    Thread Starter Member

    Dec 2, 2011
    101
    0
    Schematic:
     
Loading...