Paul & Sabrina's cheap DIY 144v motor controller - Page 170 - Fuel Economy, Hypermiling, EcoModding News and Forum

steiner · 06-12-2009, 06:43 PM

Is the thermal expansion of the mosfet legs something to be concerned with?

I made a few assumptions. If the legs were simply bent into a 90 degree angle and soldered, I assumed the length of the legs between the mosfet and the board was .200 inches. I measured a small mosfet I had laying around and the width and thickness dimensions of the legs were .035" x .017". I then assumed the legs would be heated from 25C to 100C. This meant the legs would grow .0002484 inches in length. As some of you have said.....this doesn't sound like much. However, the question becomes....how much vertical force does this put on the body of the mosfet that is attached to the heat spreader with an estimated 15 lb force. The answer is each leg applies 11.8 lbs of force due to thermal expansion. Three legs equals 35.4 lbs of force. I don't know what the coefficient of friction is between the mosfet body and the heat spreader. I am assuming you will be using thermal grease between them. Certainly the coefficient of friction will be less than one. Probably closer to 1/2.

If we assume a coefficient of friction of .5 and a 15 lb clamping force, this will require a 7.5 lb load to move the mosfet body. If we assume two of the three legs are experiencing most of the thermal expansion (third leg is not conducting many amps), then each leg applies a 3.75 lb load. Doesn't sound like much. However, this works out to a stress level of 6300 psi in each leg. That is quite a bit for copper.

So here is the dilema. If you apply a great deal of clamping force between the mosfet and the heat spreader....we will get good heat transfer. However, this will allow the stress levels to get much higher in the legs due to thermal expansion as it will be harder for the mosfet body to move relative to the heat spreader. The bend in the legs will reduce the stress level greatly as now they will buckle slightly due to the thermal expansion. The only down side I can see to this is fatigue (bending the legs back and forth every heat cycle).

My two cents.....put the bend in the legs but try to do it as gently as possible. Meaning....don't put sharp bends in them. A nice smooth radius is always better (lower stress riser). I know this is easier said than done.

If you could just put a smooth radius in the legs with them starting out straight down from the mosfet body and gently curving them outward leaving the last portion straight such that it will be parallel with the board. This would give you the clearance you need between adjacent mosfets.

Rick

squiggles · 06-12-2009, 08:30 PM

I think the flex of the circuit board will allow for 99.5 % of the expansion of the legs.

Rick is right though, a small radius, just enough so that the legs are not straight would be adequate.

MPaulHolmes · 06-12-2009, 09:52 PM

Sounds good to me! That has been nagging at me for all these many months.

blowncurtice · 06-12-2009, 11:33 PM

Paul
Attached is a photo of Curtis 1231c showing Mosfets and diodes attached through hole they do not try to bend legs except for the center leg of the mosfet which is bent back on itself and screwed through the body of the mosfet on to the heat sink(spreader) I have plenty of photos of any view of the Curtis that you may like.
John

MPaulHolmes · 06-13-2009, 12:43 AM

That's a pretty good idea to use the heat spreader as the way to carry M-. I heard that Curtis sues people that try to do too similar a thing as their layout. I wonder if it's legal to attach the middle leg like that?

Thank you for the picture!

Cyruscosmo · 06-13-2009, 02:25 AM

First I have to finish the transmission so I have something to control… my new lathe will be here any day now. ;-)

I have one of those developer kits ordered for the processor you have used for this controller so I figured I’d start by making the traction motor drive work then tinker with the remote mount part. I was just wondering if my thinking was correct as far as mounting the Fet drive components in with the Fets and connecting the control box with the driver box via a CAT5.

Also I work in Ballard and live in Woodinville my main job description is jack of all trades where I work. I repair all the tools as well as build new ones for specific jobs. I also do all the welding and repairs on the trucks and machines that need done. So if you need anything welded or built in steele, stainless steele or aluminum let me know.

Cyrus

wjdennis · 06-13-2009, 10:19 AM

I don't want to take up room on Paul's thread, so would one of you electronic whizzes contact me directly about some DC/DC questions I have on my EV?

wjdennis@qwest.net

Thanks.

Bill

MPaulHolmes · 06-13-2009, 12:10 PM

Quote:

Originally Posted by Cyruscosmo

I have one of those developer kits ordered for the processor you have used for this controller so I figured I’d start by making the traction motor drive work then tinker with the remote mount part. I was just wondering if my thinking was correct as far as mounting the Fet drive components in with the Fets and connecting the control box with the driver box via a CAT5.

It may be easier to have a small driver board that basically has the mosfet driver(s), and solder points for the gate resistors, and driver output wire, and power section ground wires (perhaps 2 of the power section ground wires). The power section grounding wires that connect the driver board to the power section should have lots of surface area, like a wide braid of copper. Very low inductance. Then, connect the control board to that driver board via CAT5, and the driver board is permanently connected to the power board (with the mosfets/diodes/caps). I think that would be a very good way to go! You should try it!

vince · 06-13-2009, 06:43 PM

This is my first post and i just wanted to say that i am thoroughly impressed,unfortunately,i dont understand a word of it,that is other than the goal.
I feel like i am a pioneer supporter at least.I am sure youhave already gathered that i hope to jump in at the end by buying one already made if its within my budget? So if i may speak for the silent majority,great job,vince.

Cyruscosmo · 06-14-2009, 07:18 PM

A question about the throttle pot does it have to be a 0 to 5k pot or could I use a 1k to 5k range TPS from an EFI system?

The reason I ask is that I have a couple old 350 chevy truck TB’s laying around that already have a place to hook a throttle cable with idle adjuster (which is used to set the TPS to 1.0K) range limiter, return spring and TPS all in one. It would make hooking the unit up to a throttle cable so much easier.

Cyrus