bussing MOSFETs together to handle AMPS

[turbobill]

I haven't done this before but need to; a tech told me to bus them together but as for how ... I'm at a loss.

The problem: I need to switch 100A DC to get pulses at about 3 milliseconds. I have the MOSFETs already that handle 73A each but to handle 100A I need at least #4 wire. Which means the leads will melt off the MOSFET, so I need to buss a bunch of MOSFETs in parallel so the amperage will be distributed across several leads.

OK, I understand all that but I can't envision a circuit board that could handle it. I'm thinking of using solid #4 ground wire or copper buss material( maybe flat stock ) with a lug but how do you solder the MOSFET leads to the buss without burning them up?

All suggestions are much appreciated.

Thanks,
turbobill

 

[Glyph]

Computer motherboards (particularly the recent ones) handle 80-120 amps using multiphase buck converters. Essentially using four to eight (and sometimes more) pairs of mosfets to switch all that current. The motherboards seem fine... for the most part.

Ya might get ideas from there. From what i hear they use very wide traces and multiple layers (a dozen last I heard) to handle all that current. Also their mosfets are surface mount or they have very big heat sinks, sometimes with their own fans, to deal with the heat produced.

If soldering is the issue, is reflow soldering an option? otherwise maybe try a high-wattage soldering iron, get the solder the melt and attach the mosfet leg quickly before the heat has time to creep up the leg and damage the mosfet.

 

[Souper man]

4 GAUGE WIRE?!?

Thats Huge!!!

 

[Sceadwian]

You don't generally solder things like that. Why not bend the leads and use some kind of clamp to get good physical contact with the mosfet leads to the bus bar? Generally speaking you want all of the mosfets physically touching each other so they share heat.

 

[Glyph]

Sceadwian has a good point.

In addition i think some discrete devices like high power IGBTs come in a "module" package that have screw-holes so you can bolt down whatever heavy duty wiring you're going to use. I'm not sure if mosfets come in such a package but it might be worth checking out.

An example of one type of module can be seen here: https://www.fairchildsemi.com/products/discrete/packaging/7pmha.html

I have one of these, it is as big as my hand with the screw holes about a half-inch in size.

 

[Sceadwian]

If you really want you can solder it after you get a good physical bond but it's not easy, you need a soldering iron capable of very high wattages to heat up the solder spot fast enough so that the rest of the bus bar dosen't suck off the heat but not to such a higher temperature that you destroy the FET.

 

[turbobill]

IGBT looks good

Yeah, that IGBT looks like "the ticket". However, I've got to use the MOSFETs right now because of budget constraints.

I think I'll try a combination of the suggestions you guys have made. See how this sounds. I'll use flat stock for the buses, heat them up a little with a torch then use high wattage soldering iron to finish them ASAP, line them up with heat sinks and add a fan for good measure.

I hope this works. If not we'll have to bite the bullet and go the IGBT route.

Thanks,
turbobill

 

[jpanhalt]

I have handled the problem of bussing by using copper strips cut from 0.025 copper sheet obtained at a local hobby shop. Thinner would probably work too. I also soldered the drains directly to the copper as shown in the pictures. Everything was mounted to a heatsink. That's important. Althought the mosfets may be rated at 75 A, they do produce heat at that current. Connections to the copper sheet are with brass screws and then to welding cable. The pictures are of a test unit. The actual unit has been working well since 1999. John

 

[turbobill]

This looks great!

Hey, thanks for the post John, this really clears up a lot of questions for me.

How much amperage are you switching with this setup?

Thanks a bunch,
turbobill

 

[turbobill]

Also, I just noticed that it looks like the drain from the top bank is the source for the second bank. Would you mind explaining that to me?

Thanks again,
turbobill

 

[jpanhalt]

Don't know the max current; didn't have any way to measure it. But, it's a lot. The DC motor that is being driven is a Ford longshaft starter that was designed for 6 V. I have the high-speed version with two field coils and no governor coil. It is feed using 2-O welding cable directly from a 12 V marine storage battery. Of course, it is only for intermittant duty in a winch. I have never stalled the winch at full power, but at about 50% PWM, I have stalled the winch for very short periods. Another unit with somewhat smaller cables and fewer mosfets runs a 1.5 KW DC motor for longer periods with no problems.

The configuration is a "half-H." In a full-H configuration, view the vertical legs of the H as your mosfet controllers (i.e., 4 controllers) and the motor as the crossbar. The full-H allows reversing. The half-H is half of that and provides high-side control and a return without the reversing function. Hope that explanation helps. Like many things, it's easier to explain on paper with a drawing than to use words. Maybe someone here has a good picture. I don't. John

Edit: Here's a link to the Linear site that describes my configuration: https://pdf1.alldatasheet.com/datasheet-pdf/view/70323/LINER/LT1158.html

Linear refers to it as a bridge configuration. The motor is grounded. My original post mentioned "low-side" conrol, which I regretted as soon as I hit the post button, as I don't believe it is technically that. Thus, I edited it to say "return."

 

[turbobill]

Thanks John, much appreciated.