SMD MOSFET that takes 40 amps

[Uitgeslapen]

thx

[jfitter]

The FDR6672 is surface mount and rated to 65A continuous and 100A pulsed. I use these a lot and have had much success with them.

40A is easily handled by the leg of a smd device - the resistance is very low because the copper is very pure, so the temperature rise is small. Measure the pin geometry and do the calcs to prove it to yourself.

The copper pcb tracks are another thing altogether. You need to get hold of the formulas for designing pcb tracks. Set you maximum ambient operating temperature, maximum track temperature rise, contiuous current, copper thickness, and cooling process and you can calculate the width of track that you need. You MUST do this otherwise you will just be designing a very expensive fuse. Expect to need BIGGGGG tracks.

You must also carefully calculate the cooling and mounting requirements for the FET. The manufacturer says it can carry 65A, but then he goes to a lot of trouble in the data sheet to give you the information you need to predict exactly what operating conditions this capability occurrs at. If you don't do this then I suggest putting the device in another room when you first power it up!!!

I have designed and made 80A speedies - I made couple of dozen or so and sold most of them to modelling mates - it was mainly and intellectual exercise, not a serious business venture and all I wanted was to get back the cost of the pcb panel. They are all smt design except for the fets and reg. They have been tested at 80A continuous and will run all day - make good pocket warmers. Excellent for big cobalt motors. None have ever failed in the two years since I made them.

The trick to success with grunty speedies like this is to do very thorough design work and use the best materials and parts you can afford. This is not going to be a commercial venture unless you live in China, so making it cheap is not a target specification.

[vintage1]

I used to solder tinned copper wire along high power PCB tracks. I've fused a good many of them.

The trick is to use maxiumum avialable board space for any high current tracks - instead of tracks connecting blobs think lines seperating copper areas.

Solder itself makes a good lowering of resistance too, by dint of its THICKNESS.

[Uitgeslapen]

thx

[jfitter]

I think you have other problems more serious than track sizes to worry about. Assuming you get your design down to 1mR total on resistance. At 80A the power dissipated in the fets is 6.4W. To get 1mR you will probably have 3 or 4 very high quality fets. They will not fit on a 3x4cm pcb and leave enough room for the high current tracks and thermal conductive paths.

6.4W is quite a bit to get rid of - it will quite happily melt a 2x3cm pcb and everything on it. You could assume that the power cables will conduct the heat away - many designers do this - but this is very dangerous design practice.

I think the "laws of nature" have got you. It is unlikely that with current technology you will succeed in designing an 80A continuous rated esc to fit on a 2x3cm pcb.

[Uitgeslapen]

nice!

[Comatose]

If he went to the stacked boards approach that Castle uses, he could make a 2cmx3cmx2cm block that would work like a charm.

The other thing you need to be aware of is mosfet current ratings are pure fiction. They're the rating with the case held absolutely at ambient, with the only thermal resistance being between the die and the case. RDSon is the only thing you need to be concerned with, because that will get you an apples to apples comparison. In our application there's no space or weight for elaborate heat-sinking.

If you only use 20V mosfets that will make your life somewhat easier - lower RDS on. If you go to some of the newer, more exotic packagings, then that will leave you with plenty of board space.

Four Irf6609 would get you to where you need to be. That'd be half a milliohm of on resistance and 3.2 watts of heat. They're metal cased, so extracting heat isn't too much of a chore. A small heatsink over them might be prudent if you're running these at 5v gate voltage.

For PCB traces at this power level your options are use normal .5cm wide or so traces and beef those up with 12ga wire layed on top of them and soldered or go with 10oz copper.

The next thing is expect something to catch fire during the development cycle at this sort of power level. I have a 20A product I'm almost done with, and that's caught fire several times. Never the fault of the product, always me shorting something, or testing overcurrent protection, or typoing my way to a popcorn situation. Have a little fire extinguisher handy. At 80A if you have just .1 ohms of resistance in the current path the device will heat up faster than if you'd put it in a microwave.

The other thing to be concerned with is your rectification diode or mosfet solution.

[Uitgeslapen]

done

[Comatose]

The sand is a good idea as long as you don't mind everything getting a little bit sandy.

First, how critical is your 2cm x 3cm size? There are cheaper and easier to work with options if thats not your overriding priority. Anything that has its pads underneath, like QFN, BGA or DirectFET is hard to work with on a prototype scale. An air pencil would work but take some doing. I'd fire up the reflow oven just for the one board, but you probably don't have that option.

If you can deal with it being somewhat bigger, there's a reason that essentially all ESC manufacturers use the SO-8 form factor. They're easy to work with, relatively inexpensive and don't take any heatsinking to approach their rated currents.

SO-8 is easy to parallel, you just run two parallel strips: one side is the drain and the other is the source. Pop the gates in from the other side with a via.

DirectFET (if you really need the small size) your best bet would be to use one of the drain pads and leave the other floating. That would make running them in one parallel strip simple.

[Uitgeslapen]

thx

[jfitter]

....I think the "laws of nature" have got to you to....

...giving more problems is not what I am looking for here, I want to solve them and have joy in desinging and writing C...

regards Daan

The first line is rude - behave yourself.

Regards the second part of the quote. Engineering is about looking for all of the problems and solving for them BEFORE they come as a surprise to you. Pretending the problems aren't there and getting on with the fun of coding is play.

When engineers design and build an 800 million dollar bridge, they don't make a practice one just in case it doesn't work. They design the first one properly. This priciple applies to electrical design equally.

[Ron van Sommeren]

@all
The 'Open Source Motor Controller' discussion group:
www.yahoogroups.com/group/osmc

Vriendelijke groeten ;) Ron van Sommeren
• diy motor building tips & tricks (http://www.rcgroups.com/forums/showthread.php?t=240993)
• diy outrunner discussion group (http://www.yahoogroups.com/group/lrk-torquemax)
• int. E fly-in & diy outrunner meet (http://home.hetnet.nl/~ronvans/), August 28th, Nijmegen, the Netherlands

[Uitgeslapen]

thx

[Uitgeslapen]

Ron,

That site is blocked, the helpdesk here thinks it has adult content. Please put some clothes on the controllers :)

[Ron van Sommeren]

I guess all Yahoo groups are blocked: www.yahoogroups.com/group/lrk-torquemax of it's French pendant www.yahoogroups.com/group/moteur-perso ??