[Cialug] electronic/power geek wanted

[Dan Hockey]

Try looking into IGBTs, the ones Nissan uses in their drive motor
controllers can handle about 400a and cost about $200 (x4) a piece.
-dh 

-----Original Message-----
From: cialug-bounces at cialug.org [mailto:cialug-bounces at cialug.org] On Behalf
Of Nathan Stien
Sent: Thursday, March 22, 2007 6:00 PM
To: Central Iowa Linux Users Group
Subject: Re: [Cialug] electronic/power geek wanted

On 3/22/07, Matthew Nuzum <matthew.nuzum at canonical.com> wrote:
> On Thu, 2007-03-22 at 15:59 -0500, James Shoemaker wrote:
> >   I have a electronic/power problem in a personal project involving
> > controlling and reversing 20A of 110V DC and am curious if anyone in the
> > lug has experience in that line.
> >   I can't find relays that can take the voltage and don't know enough
> > electronics to design an H-bridge (constructing one based on a schematic
> > is no problem, but all the schematics I find are for lower voltages)

I'm probably not the only one here who would be curious what you're
doing with that 2.2 kW ;-)

> Are you sure a relay is the way to go? I'd use a FET.

Especially with a simple DC load like this, I would second Mattew's
suggestion.  Definitely go with FETs if you can.  Big FETs are not
typically logic-level FETs, so you might need to use still other
transistors to turn the FETs on.

Alternatively, you could go with a dedicated FET driver chip.  I have
personally used the TC4429 [1] from Microchip in applications, and it
seemed fairly decent.  As I recall, though, we had an issue wherein
you can fry the chip if you apply voltage to the inputs before the
chip itself is powered up.  This isn't a hard problem to solve, but I
suggest you get the DIP parts and use a socket so you can pull out
your dead ones if this happens.

[1]
http://www.microchip.com/stellent/idcplg?IdcService=SS_GET_PAGE&nodeId=1335&
dDocName=en010675

When considering how to drive your FET, you need to look at the gate
capacitance and think about your gate drive voltage.  To turn the FET
on, you essentially have to charge that capacitor.  Charge on a
capacitor is the integral of the current, so if you pump more current
into the gate, you spend less time turning it on.

You need to do it quickly because between "off" and "on" lies the very
unpleasant "linear region" wherein you will burn a whole lot of power.
 A half-on FET can quickly become a smoking hole in your board.
Bigger FETs in general have more gate capacitance, and therefore you
need something heftier to drive their gates.

All of this begs the question of where your control signal will come
from.  In my personal experience, it's always been a microcontroller
with either 5V or 3.3V outputs, hence the need for a FET driver.

On another note:
You may already know this, but aside from Jameco and Arrow, a really
good supplier is Digikey (digikey.com).  They have a lot of stock and
they're fast.  I have historically used them for 99% of my parts for
prototyping work.  I have also used Mouser.com, but digikey is my main
squeeze.

> Of course, that depends on how hot you're running, you'll want to do the
> math to calculate heat sink requirements. I've never tried constructing
> an h-bridge that big, but the nice thing about the FETs is that you can
> connect two (for each corner of the h-bridge) to get twice the current
> throughput. For inspiration, see the "Higher power..." motor controller
> listed near the bottom of
> http://www3.telus.net/chemelec/Projects/PWM/PWM.htm

An important note to take from that design is that you often need a
diode to protect each FET from inductive kick when switching a lot of
current.

One of the keys to reducing your heat output is to get nice quality
FETs that have as low Rds_on as possible.  When the FET is switched
on, you can think of it as a simple resistance, and Power = I^2 * R.
Minimize R and you minimize power dissipation for a given current.

This might be a wee bit un-linuxy to keep discussing on-list, so you
may wish to continue this discussion off-list.  I would suggest that
if you do so, you should include both myself and Matthew Nuzum in the
To: list, because he clearly knows something about this as well.

-- 
Nathan P. Stien
Consulting Engineer / Software Developer
Embedded Systems Electronics and Software
http://linkedin.com/in/nathanstien
Mobile: 309.241.2581
_______________________________________________
Cialug mailing list
Cialug at cialug.org
http://cialug.org/mailman/listinfo/cialug