|I need to crank out a multichannel dimmer for 8 low-wattage incandescent
120VAC lamps; PC RS232 to PIC to switching devices. I'm seeking opinions on
the absolutely easiest and fastest way to do it.
I need to solve for min compexity rather than expense, this time around.
Detect zero cross and drive some triacs? Can I eliminate the need for optos
+ snubbers if I use packaged SSRs? Can I eliminate the need for zero cross
detection if I rectify line and drive IGBTs instead? If so: can I pipe the
PIC PWM channel straight to the gate? They don't make a PIC with 8 PWM
outputs, do they? How fast do I need to drive it? I guess if triac
switching is flicker free, I could easily kludge 8 channels of PWM onto a
port at pretty much any speed?.
Thanks very much for any input!
Bruce Cannon Style Management Systems
(510) 787-6870 1228 Ceres ST Crockett CA 94525
Remember: electronics is changing your world...for good!
| I did zero-cross detect and then use the capture-compare register on a
16c74a to drive 4 triacs (through optos) in the ShoeBox dimmer at
http://dovesystems.com . This unit also receives DMX at 250 Kbps and
reads 0 to 10V analog inputs. I think I'd run out of processor time if I
tried to extend it beyond four channels.
I did a 24 channel dimmer by having a 16c74a receive DMX, run the
received value through a table to determine how far into the half-cycle
we should turn on the triac. The output of the table was sent to a 12
bit serial DAC that was then demuxed out to capacitors driving
comparators. The other comparator input is driven by a linear ramp in
sync with each phase of the power line (this is a 3 phase unit). I tried
looking at an all digital approach but it just did not seem economical.
It'd be nice to find a single chip with a bunch PIC type PWM outputs
(timers and comparators). We could then reset the timer on zero cross
and serially load a compare register for each PWM output. I couldn't
find such a chip, so I went back to analog comparators. Works great!
But, it's not digital...
I think that triacs are the most efficient way of dimming lights. You
COULD rectify the line, then do a BUCK converter to drop the voltage to
the lamp. It seems that this would take quite a few parts (the FET, high
side driver, catch diode, output filter, etc.).
I've also considered using a pair of FETs to do do phase control of the
AC (instead of using a triac). However, the FET-diode combination has
more voltage drop than a triac, so efficiency is down. One advantage to
the FET approach at line frequency is that you can slow the rise time and
get by without a choke on the output. This, of course, causes a loss of
efficiency. That loss may or may not be as bad as the power lost in the
Finally, I've also considered doing high frequency chopping of the AC
line, resulting in a sort of AC-Variac where the output is sinusoidal. I
built one, but it was complicated and inefficient. The required output
filtering was also substantial (depending upon how much stuff you want to
conduct out the line to the lamp).
On Sun, 21 Feb 1999 15:44:18 -0800 Bruce Cannon <JPS.NET> writes: bcannon
You don't need to buy Internet access to use free Internet e-mail.
Get completely free e-mail from Juno at http://www.juno.com/getjuno.html
or call Juno at (800) 654-JUNO [654-5866]
At 15:44 02/21/99 -0800, Bruce Cannon wrote:
>I need to crank out a multichannel dimmer for 8 low-wattage incandescent
>120VAC lamps; PC RS232 to PIC to switching devices. I'm seeking opinions on
>the absolutely easiest and fastest way to do it.
>I need to solve for min compexity rather than expense, this time around.
>Detect zero cross and drive some triacs?
i guess that's probably the simplest, most robust and most common way.
>Can I eliminate the need for zero cross
>detection if I rectify line and drive IGBTs instead?
>How fast do I need to drive it? I guess if triac
>switching is flicker free, I could easily kludge 8 channels of PWM onto a
>port at pretty much any speed?.
yes, speed is no problem here, with incandescent lamps. anything over 120Hz
as pwm frequency is obviously just fine (that's what triacs do :). but
you'd get interference with the half waves from the line; gotta check on
that. and if you go up with the frequency, i don't know if they would like
it a lot if you switch mains with 1kHz or so to stay away from
interference. maybe filter that?
but i would guess that the "complexity" of rectifying is at least as much
as detecting 0 cross. and triacs are =very= robust.
More... (looser matching)
- Last day of these posts
- In 1999
, 2000 only
- New search...