-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Fri, Nov 16, 2007 at 12:59:34PM -0500, Spehro Pefhany wrote:
> >          V+
> >          |
> >         LED
> >          |
> > 	 C
> > 3.3V---B<
> >          E
> > 	 |
> >          R
> >          |
> >         GND
> >
> > 'Cause I built one, and it seems to do all sorts of weird and wonderfull
> > things only somewhat related to current limiting... Not sure if I'm
> > understanding it correctly.
> 
> It forms a current sink, but the voltage at the collector has to be
> more than the base (approximately, it can be a bit less in fact, more
> like 2.6V as you observed since Vce saturates at some tens of mV) for
> it to work properly. If the condition you are mentioning can exist you
> might want to add a base resistor of Re * 10 or so to keep the current
> out of the PIC within reason.

Ahh, good point, forgot about that. Could even be causing some of the
weirder stuff I was seeing as the amount of current that would be
flowing would end up being a big fraction of the total power
consumption.

I think the voltage at the base vs. the collector thing might have been
doing it. I noticed that after reaching a bit of a "plateau" in the
limiting, current started actually going *down* as I applied more
voltage to the leds. Then again, at one point it also looked like I was
getting 10mA of current to flow through a 1n4001 in the reverse
direction.

> The main difference between using Vbe and the above type of circuit is

Could you draw a schematic of what using Vbe looks like then? I don't
think I understand the difference between what I built, and what you
were describing.

> that the unit-to-unit variation and temperature change in Vbe doesn't
> impact the above type of circuit nearly as much, since the 0.6-0.7V is
> swamped by the 3.3V voltage, so it's 5x better in performance. The
> downside is that you waste a bit of power and drop a bit of voltage in
> the resistor. If you want the best of both worlds you have to put more
> circuitry in there, such as an op-amp to use a very low value resistor
> (eg. 100mV drop), but then you might have to worry about response times
> with the PWM. The transistor won't have any problems with the PWM
> frequency and the circuit is very simple. Short-circuit proof too,
> like any constant-current circuit.
>
> Say, didn't someone around here mention something about a beer sometime,
> Peter?

We did you know. I'm pretty busy this week, but drop me a line offlist
about the week after.

- -- 
http://petertodd.org
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)

iD8DBQFHP0ub3bMhDbI9xWQRAmkkAJ4wA5/rDBGjUmgpwYkpZkXPY1HnTgCfU7Ez
qRo8RSUqm1wt6OE16JSHDEQ=
=KFbd
-----END PGP SIGNATURE-----
-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist