PIC - PCF8574 - MOC3020 - TRIAC

PWM isn't applicable to AC; for dimming AC lights, you use "phase control:"

(OK, admittedly, at least one of these sources compares phase control to "AC PWM," but I've never used the terms interchangeably.)

Hope This Helps! Rich

You use a zero crossing detector and sync the pwm period to the mains half cycle. It'll work well enough for jazz.

On a sunny day (Fri, 11 Mar 2011 08:55:34 +0100) it happened Anti-Spam wrote in :

Thats is an opto-triac, and once on, it will stay on for the rest of the period. So, no. You need to find the zero crossing and then count until you fire the triac. Very basicaly the PWM will only work if your lights runs from a DC source, and you drive them with a MOSFET or transistor of sorts.

Which PIC has an IC2 port ???

hamilton

Is IC2 better than I2C?

I can add a Zero Crossing Detector, reasonably easily. Scrub using PWM and do my own timing to trigger the Triacs.

Because the MOC3020 (as I understand it) has a zero crossing detector buit in, I assume I will have to use a different device without ZCD?

But will the time it takes for the PIC to issue a IC2 command, and for its content to hit the output port of the PCF8574, not be a problem?

I thought many of them have???

OK, show me at lest one.

hamilton

Sorry, I got the terminology wrong, feel better?

So how about driving a MOSFET inside a bridge, to control AC lights then?

Grant.

On a sunny day (Mon, 14 Mar 2011 23:36:31 +1100) it happened Grant wrote in :

period.

That will work, but you may get aliasing between the PWM and the mains frequency, I had a similar aliasing problem using 3 independent PWM generators for RGB, and fixed it by using a common clock for 3 PICs.

period.

frequency,

the mains.

Quite possibly, I remember a sound to light project I made as a teenager that aliased bass notes with the mains frequency to create a flashing light instead of the expected steady light. It was three bandpass filters feeding three TRIACs for lo, mid and hi frequencies (musicolour, an EA magazine kit from early 1970s).

I've seen an SCR used inside a bridge, but not a MOSFET. At least putting the switch inside a bridge means you're not running high voltage DC around the lighting fixtures and stuff not rated to quench a DC arc.

Grant.

period.

you can do that, I've seen it done on reverse phase angle dimmers, i.e. instead of turning on during cycle and off at zero crossing, it is turned on at zero and off during the cycle. I guess you could do "real" pwm too

also seen back to back nfets used as a switch for mains

with a fet you can get quick turn off so if you measure the current and add that to the control you could make it pretty much short circuit prof

-Lasse

period.

That sounds like a big plus :) An idea for later, at the moment I'm doing low voltage LED lighting, although a recent prototype runs >100V to 50 LEDs in a series string. Powered from 12 or 24V DC.

Grant.

Before i read any others posts; i suppose you could do it that way. You would have to have 40 PWM outputs though (one per channel). Plus some = way to synchronize them with the mains, otherwise the triacs would = effectively be firing randomly.

Bluntly, the triacs are PWM outputs, with mains timed start pulses and = off at the end of every half-cycle.

Forty channels, across phases too? Take care synching to the correct phase ;)

Grant.

OK, did not really think the Dimming problem through. So have decided not to use the PCF8574, and use a PIC in its place. In fact will drop I2C altogether in favour of Serial Comms, and use a PIC with a UART port instead. Will also drop the MOC3020, for a opto/triac isolator, with no Zero Crossing detector. Will do my own zero crossing detection, with a opto isolator, going back to the PIC.

I am thinking on the lines of having 5 slave boards (each having 8 lamp output channels), each slave having its own unique address (set by DIP switches), but all the SERIN inputs of the 5 slaves, wired in parallel.

I read somewhere that a PIC UART can only hold 2 bytes of data, so the Master controller board would send 2 Bytes of data, for each lamp it wants to control. The 1st byte would would contain the address of the board/lamp it wants to switch and the 2nd byte stating how long the fade up/down time should be, for that lamp.

Am I now thinking on the rights lines?

I suggest that you start with mains synchronization. Required just make = a dimmer produce consistent results. Then worry about many channels. Then the fancies.