1
\$\begingroup\$

I designed/had built a PCB to handle higher current draw. The P Channel MOSFET (Q1) I selected is used as a ON/OFF switch for the rest of the circuitry. R2 gets grounded through a switch when flipped "ON", hence I just have it grounded in the picture. It is getting very hot (230F) at 20A current draw after about 5 minutes of being on.

enter image description here

  • Q1 = SQM120P06-07L_GE3

  • D1 = SMAZ10-13-F

  • R1 = 1/10W 0603

  • R2 = 1/4W 0603

  • The MOSFET has a Absolute Max Vgs of +/-20V

  • When measuring the voltage across the Gate and Source, I get -10V which the datasheet says should provide the minimum RDS(on).

  • The MOSFET is properly heatsinked on 3 layers of 1oz copper through vias.

My question is, why is this getting so hot? Did I incorrectly select R1, R2, or D1?

\$\endgroup\$
4
  • 1
    \$\begingroup\$ The datasheet specs R=0.0067 Ohm at V_gs = -10. That's U_drop = 0.134 Volts, and at 20A, that's 2.68 Watts. I'd say that's quite a significant amount of heat. \$\endgroup\$
    – akwky
    Feb 10 at 16:02
  • \$\begingroup\$ 6.7mOhms at 20A gives over 2.5W, that is a lot of power to get out of a FET. I am not surprised that it got hot. As mentioned, a heat sink is required. \$\endgroup\$
    – Puffafish
    Feb 10 at 16:03
  • 1
    \$\begingroup\$ 20 A is absolutely not "low-ish" current. You should probably not be using a PCB heatsink for this. \$\endgroup\$
    – Hearth
    Feb 10 at 16:50
  • \$\begingroup\$ R2 will also fry (324 mW dissipation). \$\endgroup\$
    – Andy aka
    Feb 10 at 19:00

2 Answers 2

3
\$\begingroup\$

Heatsinking is required. Doesn't matter how low the source-drain resistance seems. With a junction-to-ambient thermal resistance of 40K/W that's still 107K rise even if you only dissipate 2.7W...at room temperature The electrical resistance increases as you the MOSFET gets hotter. Between 1.5-2x when running at it's max rated temperature.

PCB heatsinks just aren't that effective. You would have to show us your thermal layout. But after working with a real heatsink that something like a TO-247 package bolts to, doing it on a PCB leaves a lot to be desired. I only used real heatsinks before and always felt that a PCB couldn't really stand up the same way. But I heard people doing it so gave it a try and it was about as effective as I thought it was. And I was using 4-layers of 4oz-6oz planes that were more than 2 square inches in area. I'll stick to discrete heatsinks now unless size or weight is a concern.

20A isn't very low either.

\$\endgroup\$
9
  • \$\begingroup\$ Should I just select a more "efficient" MOSFET with a lower junction-to-ambient thermal resistance? I wont go anywhere near the 120A capacity that this MOSFET can provide. Most likley just 40A. \$\endgroup\$
    – Nick_Nack
    Feb 10 at 16:05
  • \$\begingroup\$ @Nick_Nack Yours is already low especially for a difficult to source PMOS. Not like a sub1mOhm NMOS but reasonable. I do not think you will get much benefit. Can you connect a heatsink to the opposite side of the PCB or something? Or a fan? \$\endgroup\$
    – DKNguyen
    Feb 10 at 16:07
  • \$\begingroup\$ Using a fan is not possible as this is located inside a watertight box. Heat sinking the bottom of the board is also not a possibility. I will either have to move to a through-hole package and put a heatsink on that or use something like FDB9503L-F085 that has 2.6mohm RDS(on) at -10V Vgs. \$\endgroup\$
    – Nick_Nack
    Feb 10 at 16:44
  • \$\begingroup\$ A heatsink won't help in an enclosed box either. The heat needs to get out. If it's in a water tight box then you might have to do something like putting a heatspreader against the MOSFET or PCB and butting up the heatspreader against the wall of the metal box, possibly even with a heatsink on the other side of the wall. If it's plastic box and not submerged then maybe you can cut away a window and replace it with a sealed aluminum or copper plate epoxied or siliconed in. \$\endgroup\$
    – DKNguyen
    Feb 10 at 16:48
  • \$\begingroup\$ Why is it water tight? Is it submerged or are you expecting spray from all directions? Down-facing vents won't work? (you would still need a fan). Also, if it's water tight you may also get condensation issues if there is temperature cycling. \$\endgroup\$
    – DKNguyen
    Feb 10 at 16:54
2
\$\begingroup\$

As stated in previous answer, at room temperature (25 ºC), RDS(on) will be 6.7 mOhm at VGS=-10V. When the transistor switches on, power will be 2.7 W, and temperature will rise, but also RDS (positive temp. coef.). I estimated thermal resistance of PCB based on your remark: after 5 minutes, 230F = 110ºC, making one minute iteration of RDS and power losses until 15 minutes. This is only for comparision, it results about 28 ºC/W. After 15 minutes, the junction temperature will rise to near limit, RDS = 12.5 mOhm, and power losses about 5 W.

enter image description here

With a heat sink below 10 ºC/W (and through-hole FET), Tj will be about 56ºC, RDS=8 mOhm, and power loss 3.2 W. Consider to use 2 parallel MOSFET (current 2 x 10 A), with PCB arrangement and same gate circuit. With 1 squared PCB, according datasheet RthJA=40 ºC/W, and temperature will rise to 57 ºC, total losses will be 2 x 0.8 = 1.6 W. This may be better supported by a closed box.

enter image description here

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.