----- Original Message -----
From: "Roman Black" <fastvid@EZY.NET.AU>
To: <PICLIST@MITVMA.MIT.EDU>
Sent: Tuesday, May 22, 2001 9:22 AM
Subject: Re: [EE]: Battery charger design


> Bob Ammerman wrote:
>
> > > 1. it's open loop, with no current feedback the constant
>
> > The loop will be kinda closed by voltage sensing of the battery. Exact
> > currents are not that important.
>
> If exact currents are not important stick with the
> two transistor design. Spehros current-mirror is
> more accurate but maybe the complexity is not
> justified.

Yeah, I don't think the complexity is needed. But I would like to know how
to build this 'current mirror'.

> > > Ok, rscale resistance can change up to 5% depending how
> > > hot it gets. You can pay for an expensive resistor, or
> > > simply calbrate it when hot, or use a huge (10w??)
> > > resistor so that its temp doesn't rise much.
> >
> > Yeah, I figured I'd use a pretty big resistor here: 400ma @ 2V = 800mw.
A
> > five watter shouldn't get too hot. How does a heatsinked TIP30 sound?
>
> Hmm. A 5w resistor for 1w disspate is about the limit
> I would use for a HOT resistor. Run a standard 5w resistor
> at 1w and measure it, i've seen 30'C rise, which makes
> it about hot coffee temp and the value will change at
> least 2% with most cheap resistors.

How hot do they get at 5W then?

Ok, 10W then. I was kinda figuring on using an aluminum case and sinking the
PNPs to it. I could sink the resistors to it also or would the additional
rise on the PNPs make that a bad idea?

> I suggest maybe 1v @ 400mA, with the 5w resistor.
> 1v is plenty for a decent current sense.

Yeah but at 0.1 @ 40ma that isn't giving me much slop at all in the Vbe
number. (That's why I created the spreadsheet, so I could play with these
things easily).

> > > My solution here is to make the voltage across Re
> > > higher, say 2v or 3v. This reduces the error from
> > > Q2 Vbe changing. Again not so critical with two fixed
> > > currents but you may want to change it to PWM etc
> > > later.:o)
> >
> > The problem with this is that my downside compliance gets rather high.
>
> But you only need 10:1 difference between the 2 currents?
> 400mA CC = 2v Re + 0.6vbe =2.6v
> 40mA CC = 0.2v+ 0.6 = 0.8v
> PIC pin will supply 3.5v ok, at 10mA. I don't see
> a problem.

Yeah, that was dumb. I was thinking that I had to keep the collector of Q2
below Vbat, but that isn't true. Dumb!

> > > Cool, will work well, you can use IN4148 as you
> > > don't need schottky. Plenty of volts headroom for
> > > turnoff.
> >
> > Don't I have to bring Vb below Vbe to ensure turnoff?
>
>
> Using the figures above, for the volts at the base
> of Q2;
> PIC pin high = 2.6v base = 400mA CC
> PIC pin float = 0.8v base = 40mA CC
> PIC pin low = <0.6v base = 0mA CC
>
> Just forget the diode, with the PIC pin floating
> the R ladder is holding Q2 almost off anyway, once
> that PIC pin goes low (to about 0.14v) through the
> Rpic resistor Q2 will definitely turn off. :o)
>
> You could make this circuit very accurate, but if
> all you really want is one PIC pin to give 400mA
> or 40mA or off, and you can tolerate a few percent
> current error, it will work great. Better to make
> it simple I think than too perfect.
>
> For efficiency I think you could drop Rsense to maybe
> 0.4v at 400mA, but I would still keep that series
> diode to the battery...

The diode is a very good idea, I think.

Thanks again, Roman.

Bob Ammerman
RAm Systems
(contract development of high performance, high function, low-level
software)

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads