Custom Personal Vaporizer, Need to glue a PCB before soldering.

Discussion in 'The Projects Forum' started by UsagiCurry, Nov 17, 2015.

  1. UsagiCurry

    Thread Starter New Member

    Nov 17, 2015
    9
    1
    I am building a custom personal vaporizer and the MOSFET(s) I am planning on using are for flat mounting to a pcb. I want to directly solder the drain tab to a thick piece of copper and the source and gate to a pcb that is glued to the piece of copper. Because the copper is going to be a plate it will need to be heated either in an oven or on a hot plate to be soldered do the adhesive that holds the board to the plate must be able to withstand 600 degrees or more for at least the amount of time it takes for the solder paste to set. I want it to be soldered to the copper as a heat sink. If you have any ideas i would like to hear them.
     
  2. ErnieM

    AAC Fanatic!

    Apr 24, 2011
    7,395
    1,607
    Solder it before you glue it.
     
  3. ScottWang

    Moderator

    Aug 23, 2012
    4,856
    767
    Could you draw a block diagram.
     
  4. UsagiCurry

    Thread Starter New Member

    Nov 17, 2015
    9
    1
    Mosfet mounting.jpg
    This is what i'm trying to do.
     
  5. ScottWang

    Moderator

    Aug 23, 2012
    4,856
    767
    Is this a top view or side view?
    I still not quite understand the meaning of block diagram, does the bottom shows that it is a 8 pins mosfet?
    How is the meaning of labeled mosfet in the white area?
     
  6. UsagiCurry

    Thread Starter New Member

    Nov 17, 2015
    9
    1
    the top portion is a side view. The label for the mosfet is in the white area because it is represented in black.the bottom part is a bottom view and is to illustrate that it is a 8 pin mosfet. but the bottom view shows the top of the mosfet as the drain conductor is the top of the device.
     
  7. kubeek

    AAC Fanatic!

    Sep 20, 2005
    4,670
    804
    You could also try using a different style of mosfet, like a DPAK, which will have much better heat transfer into that copper pad than a soic.
     
  8. mcgyvr

    AAC Fanatic!

    Oct 15, 2009
    4,771
    971
    Personally I wouldn't ever try to build/design something like that.. As kubeek suggested a different package or something else would be my direction..
    I'd be concerned with differing thermal expansion rates and many more factors..

    But
    The copper layers on circuit boards are attached to the fiberglass substrate via epoxy based adhesives..

    Note: If done wrong.. You have made an excellent "thermal insulator" which you don't want at all.
     
  9. UsagiCurry

    Thread Starter New Member

    Nov 17, 2015
    9
    1
    The 4 drain pins of the mosfet are actually a large metal pad on the back side that i was planning on soldering 100% for full heat transfer. I chose this mosfet because of the low threshold(<1v) and full output at 2.5v. For a 120 amp mosfet those are good specs. Im only using 6 to 8 volts across it so it is hard to find mosfets that will reach minimum restriction for this application. http://www.st.com/web/en/resource/technical/document/datasheet/CD00231699.pdf is the spec sheet. I will likly be using a .07 to .09 ohm load at 6.8 volts.
     
  10. kubeek

    AAC Fanatic!

    Sep 20, 2005
    4,670
    804
    I don´t think you are reading the datasheet correctly, the 120A is valid only if you have infinitely large heatsink that can hold the that metal pad at 25°C. For your case the Tpcb=25°C, which is only 28A.

    I think I would try glueing the pcb first and then soldering the pads, you should not need too much heat to attach the four pads.
     
  11. UsagiCurry

    Thread Starter New Member

    Nov 17, 2015
    9
    1
    Insanely large heat sink is precisely why I was looking for a high temp adhesive. Fortunately I have found a solution, https://www.menards.com/main/p-1444439435291-c-6881.htm?tid=5524387748285262162. This stuff is good for 2000 degrees. It is Sodium Silicate based. The entire top of the vaporizer is going to be a copper heat sink. It will be about 1/4" thick 1 1/4" wide and 2 1/2" long. If I'm correct for 120 amp operation at 6.8 volts my load would be .057 ohms, the mosfet is at max .003 ohms so the power dissipation would be around 46 watts. Which would be way too much for direct mounting to a 2oz copper PCB with a 1" drain pad. I may put 2 of these in parallel to reduce the amount of heat generation, but I also plan on using these at about 75 to 90 amps not 120. I could use a thermal cutoff device for safety but the ones i would have used in the past appear to be no longer available. I also don't plan on this thing being on for more then 10 sec at a time. I may also use a thermally conductive potting compound after assembly to increase the heat disapation.
     
    Last edited: Dec 2, 2015
  12. UsagiCurry

    Thread Starter New Member

    Nov 17, 2015
    9
    1
    Next question. With this mosfet, since this is a battery powered device, reverse polarity is probably an issue. would using a diode across my switch be effective.
    fetswitch.jpg
     
  13. ScottWang

    Moderator

    Aug 23, 2012
    4,856
    767
    For your application,
    Vds = 90A * 0.003Ω = 0.27V
    Pd = Vds * Id = 0.27V * 90A = 24.3W.
    You may concern the power dissipation about 3 times of rated power, W= 24.3w *3 = 73W.
    If you using Id = 28A then you will need some more mosfet, num = 90A/28A = 3.2, you may need 4 pcs of N mosfet.
    And you need four flywheel diode for the emf of motor to protecting n mosfets.

    The circuit maybe more complicated for the Vg input and Id.
     
  14. UsagiCurry

    Thread Starter New Member

    Nov 17, 2015
    9
    1
    The load is a heating coil not a motor. I used the image as a reference, specifically, to ask if putting a diode across the switch in the illustration to saturate the gate in a reverse bias situation would prevent current from passing through the load. As for the thermal situation, I have realized, that due to the package of this mosfet and the size of the pad for the drain, that the design of the heatsink is going to have to be more robust for continuous operation. If I am correct, which I am not always correct, 2 of these in parallel @90 amps would have a voltage drop of .18v causing a thermal load of 16.2 watts over a surface area of 56 mm^2, that my heatsink would have to be 4.21*C/W to maintain a contiuous Tc of less then 100*C at 45 amps effective across each mosfet. well under the 75A @ 100*C Tc rating. The 28A rating is for if you mount it as intended on a 2 oz copper PCB, but then i think the rating would actually be lower then 28amps. I am looking at different mosfets but I don't think thermal load will be an issue because the duty cycle of this device is going to be 10 seconds max with at least 120 seconds rest.

    Most people that build these devises use a mosfet with no heastink encased in cheap epoxy and have few problems.Which, in my opinion, is a worst case scenario. They also have no reverse bias protection so if someone puts the batteries in backwards the device operates continuously and, due to the lack of thermal management, the mosfet fails. Ideally I would like to build a fully mechanical switch,but due to space constraints, making a 120 amp safe switch would be difficult.
     
Loading...