PWM Dimmer Project.... I hope?

Discussion in 'The Projects Forum' started by awhaley, Dec 31, 2008.

  1. awhaley

    Thread Starter New Member

    Dec 31, 2008
    4
    0
    Hello everyone! This is my first post here... I've been doing some research on a dimmer to be used for some crazy synchronized christmas light blinking. This year, I was using a simple - on/off controller, the guts of which consist of:

    4 MBI 5027 LED drivers, being controlled using a PC parallel port (one data line, one clock line, and strobe used for Latch Enable) These chips accept data serially, latch it, and can drive up to 80ma per channel with the data, holding their state till the next data is sent.

    The LED driver chips are actually driving an opto-triac based optoisolator, which in turn is hooked directly to an 8A Triac which switches the AC load.

    The system is very simple with a low parts count, which is great for folks who are new to the christmas light control hobby... The best way we've found so far to allow dimming control (operation at perceived levels other than off and on) is to add a PIC and a circuit to detect the zero crosses in the AC waveform and refresh the chip's latches ridiculously fast at the correct time in the waveform... which works, but has the disadvantages (for new folks entering the hobby) of requiring a programmed part (the PIC), adding a little complexity with the zero cross detection, and requiring very high speed data transfer between the PIC and the LED driver chips, which requires very short cables and is still susceptible to errors....



    So what I'm thinking, and need some advice on:

    I'd like to use the MBI5031 chip instead... which is a similar chip, except that it's designed to take a 12-bit level instead of a 1 bit state, and it generates a PWM signal based on that level.

    Now the next step is a little over my head... The triacs we've been using, of course, latch on once they're turned on and remain on until the AC line voltage crosses zero again (well, almost until then....) and to use a PWM signal I need to replace this part with a part that shuts off when the gate is taken low again. Am I looking for a pair of MOSFETS back to back or an IGBT perhaps?

    I can't find any good basic discussion with useful bare-bones example circuits for dealing with IGBTs. I'm looking for the bare minimum parts count/complexity level... and I'd love a 'for non electrical engineers' discussion of what they do and how to use them. I've found several good discussions of how they're made, but I need to know how to build a working circuit with them, if they are in fact the part I'm looking for?

    Anyway. If someone has some suggestions for where to get started with a part that can cycle on and off with the PWM signal and switch AC current, I'm all ears. I guess I should at least ask about that assumption... would my life get much easier (parts count much lower) if I put a 20A Bridge Rectifier on the AC line coming into the dimmer contraption so I was trying to switch 110V Full Wave Rectified DC? I can't think of any reason the christmas lights would mind being fed DC instead, can you?
     
  2. awhaley

    Thread Starter New Member

    Dec 31, 2008
    4
    0
    Well, I saw in someone else's post that IGBTs cannot be used with AC loads, which is contrary to what I'd heard somewhere, but I'll accept it. :)

    I ALSO found an example elsewhere on the web of someone doing something similar by putting a 20A rectifier in front of the dimmer input, then using MOSFETs on the outputs, which may be what I want to consider. His system is much more complex than what I'm hoping to build, with a PIC controlling PWM chips, which are then going through an opto, then to a 'mosfet driver ic'.

    I found one circuit where mosfets were being driven directly from a PIC for low voltage DC... So my big question about the project is now what's the bare minimum parts count for a mosfet driver circuit switching 110V Full Wave Rectified DC? Can the gate be tied directly to the pwm chip? Is the opto enough? Will I burn up the MOSFET if I don't use a mosfet driver, because it's spending too much time in the linear part of it's cycle? Is there a low parts count discreet circuit for this, or is the IC the simplest solution?

    The LED driver chip I'm planning to use to generate the PWM signal can sink up to 17V on the outputs, if that helps me in anyway (I believe I'm understanding that mosfets prefer higher than logic level drive voltages?)

    Thanks for the help!

    Art
     
  3. Bernard

    AAC Fanatic!

    Aug 7, 2008
    4,172
    397
    Not to much to do with your proj.,but try: Bernard's Pictures & Albums-Bernard's Projects-Flag. Uses full wave rectified 24 V AC for 20 rows of 13 incadescent 2.3V bulbs each, each two collums are controlled by a MOSFET. Three collums march across the flag,outer two @ 2/3 brightness, the center @ 1/3. The flash wiped out the waving. Rather than PWM it uses PPM[pulses present mod.] Use due care with 120 rectification, remember the old AC\DC hot chasis radios. MOSFET's should work. Are you using LED's?, how many strings per driver?
     
  4. awhaley

    Thread Starter New Member

    Dec 31, 2008
    4
    0
    Thanks for the reply Bernard! The pictures and albums you mention... are they a function on this forum? I looked around a bit and didn't see them...

    The questions you asked are a bit hard for me to answer, because the circuit, when completed, is going to be available to a wide community of users around the world to do several thing at several voltages. There's a pretty large group of dedicated amateurs involved. At the moment, we're all using triacs and dimming the lights by phase angle control, but this requires a phase sensing circuit, and all the solutions for this so far require a programmed microcontroller. These facts make it difficult for new folks to get into the hobby with a simple solution.

    For my own use, I'd like it to switch 120V rectified DC, up to around 5 amps. In the UK it may be 230V, in Aus/NZ they may use 24V or 36V DC... different mosfets may be used for each voltage. I appreciate the reminder about the dangers of working with higher voltage DC.


    Edit:
    I'm attaching a schematic of what I'm considering for a test circuit, with one channel of the output detailed in. Obviously, there would be 15 more of them.

    I've never worked with MOSFETS before, and my questions about it are...

    1. The PWM chips sink current... do I have the MOSFET connected the correct way to switch current when the output goes low? The Pullup I drew in (R3) won't do what I want it to, will it? It needs to be connected to +120V Rectified to pull the MOSFET gate up to Source, except when the input goes active, won't it? That sounds a little dangerous... putting 120V connected to the same place as the LED Driver chip's output pin. Am I missing something on all of that?

    2. The MOSFET is rated for 'logic level' control... with up to 20V source to gate... But it looks like this system provides 120V source to gate? Am I doing this wrong? The circuit I've modeled this off of used a MOSFET rated at 10V source to gate, and was rated for up to 30V on the switched side... so...

    Am I missing something? Is an opto required? And I'm still wondering if I need a MOSFET driver chip or circuit, to clean up the signal going to the chip to keep it from staying in the transition stage too long and getting overly hot?

    Thanks again for the help!

    Art
     
    Last edited: Jan 3, 2009
  5. Bernard

    AAC Fanatic!

    Aug 7, 2008
    4,172
    397
    First,when you use full wave rectification , neg. output is no longer earth ground, so if going this route isolation is needed. Also a low V power supply [ 12 V DC ] derived from rectified AC such as a switching PS. The isolator may provide the required inversion ,then would need non inverting MOSFET driver. If the high ripple mixed with PDM is objectionable ,add some filtering.' would suggest a high V MOSFET, 400 V ? With 12V supply a wide choice is available. Click on Bernard to view pers. profile.
     
Loading...