6
\$\begingroup\$

There are many examples of MOSFETs that, if cooled properly, can handle sustained currents upwards of 200A. Many of there come in surface mount packages.

How is it possible to design a PCB that is capable of handling the 200A drain current?

It seems like you would need to solder theses directly to a copper plate.

Example datasheet: IRLB3813PBF

\$\endgroup\$
9
  • 1
    \$\begingroup\$ Not all PCB's are created equal. This calculator shows that using 6oz copper plating, and allowing a 15 degree C temp rise of the PCB, you can pass 200 A with a trace less than 60mm wide 4pcb.com/trace-width-calculator.html \$\endgroup\$
    – Kyle B
    Jan 29, 2021 at 0:49
  • \$\begingroup\$ @KyleB that's still hilariously huge, especially when the pins are only 2.5mm apart on some of these. Assuming you had whatever ridiculous cooling you needed to run 200A continuous on a TO-220, I bet this would be a good use case for the embedded bus bar boards that appeared on another question recently. \$\endgroup\$
    – K H
    Jan 29, 2021 at 1:30
  • \$\begingroup\$ "There are many examples of MOSFETs that, if cooled properly, can handle sustained currents upwards of 200A" - please provide an example. \$\endgroup\$ Jan 29, 2021 at 1:48
  • 1
    \$\begingroup\$ There's even a 1600A continuous mosfet that's only 6cmx10cm with m6 screw terminals. Huge for a mosfet, but implausibly small for the claimed current. \$\endgroup\$
    – K H
    Jan 29, 2021 at 3:49
  • 1
    \$\begingroup\$ Datasheet for IRLB3813PBF.... It does say "200A continuous" but there's an asterisk.... Read on, later the datasheet states "Calculated continuous current based on maximum allowable junction temperature. Package limitation current is 120A." STill kind of a ludicrous amount, but only about 60% of 200A \$\endgroup\$
    – Kyle B
    Jan 29, 2021 at 18:40

3 Answers 3

6
\$\begingroup\$

The maximum drain current is more theoretical than practical. Let us delve deeper.

The datasheet says 260 amps with the case temperature of 25°C, and RDS(on) of 2 mΩ.

So the transistor must dissipate 260 * 260 * 0.002 = 135 watts while maintaining a case temperature of 25°C. That may not be physically impossible, but it would require a totally unreasonable thermal solution (heat pipe or liquid cooling or something outrageous). So, actually, it is not a realistic continuous current purely for thermal reasons.

Even at the lower rating of 190 amps at Tc = 100°C, the dissipation will be 72 watts. It will be extremely challenging to maintain a case temperature of only 100°C while dissipating 72 watts.

In addition, if you read the fine print, the datasheet actually has a note that says the package maximum drain current is 120 amps, and that the 260 amp figure is just based on a simplistic thermal calculation using RΘjc (thermal resistance from junction to case).

So in the end, the so-called continuous current rating is not even intended to be realistic. It may be useful for comparison with other transistors using the same rating scheme. But other than that you will most likely need to do thermal calculations based on your actual operating conditions. So you should be looking at the maximum junction temperatures and the various thermal resistances and RDS(on) rather than Id continuous. Don't neglect the importance of the PCB footprint area in thermal design of SMT devices. Often they are intended to transfer heat through a large SMT pad on the PCB.

Some MOSFET datasheets now provide current limits based on thermal resistance from junction to ambient, which are more directly realistic/useful.

\$\endgroup\$
6
  • \$\begingroup\$ @Transistor I will do my best to remember. But I have been capitalizing units named after people for a long time. \$\endgroup\$
    – user57037
    Jan 29, 2021 at 17:24
  • 1
    \$\begingroup\$ No problem. You can also use HTML entities &Omega;, &mu;, &deg;, &times;, etc. as well as <sup>...</sup> and <sub>...</sub> in the posts (but they don't work in the comments). \$\endgroup\$
    – Transistor
    Jan 29, 2021 at 17:36
  • \$\begingroup\$ @mkeith, thank you. This answer really helped clarify the situation for me. \$\endgroup\$ Jan 29, 2021 at 18:32
  • \$\begingroup\$ Datasheet for IRLB3813PBF.... It does say "200A continuous" but there's an asterisk.... Read on, later the datasheet states "Calculated continuous current based on maximum allowable junction temperature. Package limitation current is 120A." STill kind of a ludicrous amount, but only about 60% of 200A \$\endgroup\$
    – Kyle B
    Jan 29, 2021 at 18:44
  • \$\begingroup\$ @KyleB I didn't catch that at first. I will see if I can work that into the answer. \$\endgroup\$
    – user57037
    Jan 29, 2021 at 19:21
7
\$\begingroup\$

A better question would be "How do I get the heat out of the package" But the PCB also makes a difference because the heat has to go somewhere or things continually get hot.

Many of them are package limited, but it really boils down to power, the packages can't take the power and if continuous current was used, it would probably melt the package. This fet can handle 400A of current pulses, but the designer must keep the average power and temperature low.

enter image description here
Source: https://www.ti.com/lit/ds/symlink/csd18512q5b.pdf?HQS=dis-dk-null-digikeymode-dsf-pf-null-wwe&ts=1611874884504

In the same datasheet, they give a layout and a thermal resistance number (which is really nice). This one 125C/W this means that 1W of power would raise the package temp 125C (which is close to the limit). Rdson is 3mΩ so sqrt(1W/3mΩ)= 18A

enter image description here

This one 50C/W this means that 2W of power would raise the package temp 50C (which is close to the limit). Rdson is 3mΩ so sqrt(2W/3mΩ)= 25A

enter image description here

If you wanted to get the power higher you would need a heatsink on the package.

\$\endgroup\$
4
  • \$\begingroup\$ Or use IMS .... \$\endgroup\$
    – user16222
    Jan 28, 2021 at 23:49
  • \$\begingroup\$ @JonRB Please clarify what is IMS in this context? "IMS mosfet" brings up nothing useful on google. Very useful answer thanks spike. \$\endgroup\$
    – K H
    Jan 29, 2021 at 1:33
  • 1
    \$\begingroup\$ @KH IMS = Insulated Metal Substrate \$\endgroup\$ Jan 29, 2021 at 3:28
  • \$\begingroup\$ @user8908459 Thanks! \$\endgroup\$
    – K H
    Jan 29, 2021 at 3:40
3
\$\begingroup\$

You can. But it’s not that easy. In fact, I am currently making a 140A PCB with FETS that are SMD!

First, just because a FET says 200A, doesn’t mean it can do it. Unless you use it in a module form not a traditional PCB with module level cooling from the bottom.

But that’s not what I do. I use multiple low Rdson fets in parallel. And then you can have a really high oz board and use the copper for both current carrying and cooling.

\$\endgroup\$
1
  • 2
    \$\begingroup\$ If the FETs are running at a reasonably low frequency, then they could also be put on a separate PCB. That separate PCB can be manufactured to have a lower thermal resistance to substrate. I have experimented with mounting FETs on the aluminum substrate boards you can now get cheaply thanks to all the LED lights that use them. The copper isn't super thick, but the heat dissipation beats most reasonable things you can do with copper. It's at least worth trying - especially if you have some thermal FEM modelling software and can run a decent model to see how locally hot the thinner copper gets. \$\endgroup\$ Sep 1, 2022 at 2:00

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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