I did a dimmer for a 12/24vdc halogen.  I didn't have any complaints about
emissions, although I will grant you they probably didn't care that much.  I'll
be honest, I didn't use a PIC (hiss from the crowd).  I wanted a voltage
programmable device, 0-10v control, so I used the national LM78S40 switcher
controller.  I modified the basic app note, which is based on harder and harder
to find PNP power bipolars, to operate with MOSFET's.  I just added some level
shift and a source follower to handle the capacitance.  I agree with Mike,
efficiency will suffer dramatically when operating the FET in the linear
region.  I would venture to say that you might be better off either adding a
component in series with the lamp to limit current rate of change (commonly
known as an inductor), or change your concept over to a switching supply,
whereby you use a buck or boost, and an output diode, inductor, capacitor in a
classic switcher.  This way, you are creating 'clean' dc to send down the
wires.  In the case of the PWM controller, that square wave is being jammed
into the wires to the lamp.  The switcher will still be relatively efficient as
well.  If you need programmable dimmness, you might have to get really creative
with the switcher, as most of the switch controller chips are designed to
achieve a fixed target.  That is why I chose the old 78S40, it has a
programmable setpoint.  For more on switchers, study the Linear Tech app notes,
they are great.  For a DC switcher, try either Linear Tech or National simple
switcher.  For an AC line powered supply, try a Power Integrations Power Tops
device.

Chris Eddy
Pioneer Microsystems, Inc.

Mike Harrison wrote:

> On Mon, 13 Oct 1997 18:31:07 +0100, you wrote:
>
> >I'm looking into a project which involves dimming some 12V 50 watt lights,
> >and am planning to use the PWM outputs to switch the current via a FET
> >output stage.
> >
> >Does anyone have any suggestions about how to reduce the RF emissions this
> >is likely to produce? I don't have space or money for large chokes. If
> >anyone has any pointers to relevant web-sites (or can recommend any texts)
> >then I'd be very grateful
> >to hear...
> >
> Control the switch-on/off time so it isn't too fast - the easiest way
> is a resistor in the gate, which together with the gate capacitance
> limits the rise time (but at the expense of efficiency).
> Also, bear in mind that cold lamps have a much lower resistance, so a
> software soft-start would be adviseable to avoid popping the fet.