Post by thread:PWM synch to ac line

>Hi all, > > I have an interesting challenge. I am using a 16C57 to drive a >nitinol wire actuator with PWM. The drive goes to multiple actuator and >as it turns out I need a whopping current to drive them. I am lifting >something heavy. Each takes 6 amps average. Everything was going along >happy (in denial) until I started to purchase the power supply to provide >the current. > > Now I am wondering if any of you have experience with sourcing >PWM current from an un-regulated supply. I am thinking that maybe I can >simply step down the ac line, full wave rectify it, and synch my PIC PWM >timing to the zero crossing. I use IRF520 MOSFETs to switch the current. >This would insure that my driving waveform would be uniform for each cycle. >I do not believe that having a non-DC source current is any problem. After >all, I am only heating wire. > > Also how about some clever (meaning cheap/simple) ideas for PIC >sense zero crossing from the transformer secondary. > Use a full-wave rectifier to develop a bus having positive half-cycles of AC (i.e. 120 Hz pulses). The transformer and diodes have to be rated for the sum of the current through all the wires that will be heated at the same time. Connect one end of all the wires to this bus, and the other end to the anode of a SCR that controls that wire. The cathodes of the SCRs go to ground, and the gates are driven by the PIC through resistors (as if they were NPN transistor bases). About any SCR will work, for example the S2006L (about $1.20 each in Digi-Key catalog) will switch 6A RMS and turn on with 15 mA of gate current. If you're driving multiple wires from each channel, 70 Amp SCRs are available (about $7.00), but you'll need to add a driver transistor (emitter follower from PIC pin, with collector on 5V bus) to supply the 50 mA gate current required. The SCRs will turn on when positive gate voltage is applied, and turn off during the zero-crossings of the bus voltage. If you want on/off control with 1/120 second resolution, synchronizing is unnecessary. Just leave the gate drive on while the wire is to heat up, and turn it off when not. If proportional control is required, synchronize the PIC with a zero-crossing voltage derived from the bus (by comparing when the bus voltage rises above 1V or so), and generate a delayed pulse for the gate. The more delay of course, the less power will be applied. But the relation between phase delay to power is not linear (think of integrating the sine squared function), easily corrected with a table if required. As far as conditioning the bus voltage for the PIC, it could be simply applied to RTCC through a suitably large resistor to limit the current to <500 uA when the protection diode clamps the pin voltage to 5V. Loading the bus with a resistor to ensure that it does return to 0V at each half-cycle would be a good idea. The turn-on point would be about 3V. If earlier detection is required, a more sophisticated voltage comparator, for example driving the base of an NPN transistor and using the collector with a pull-up to 5V, would switch on while the bus voltage is more than 1V or so. -Mike

> > Now I am wondering if any of you have experience with sourcing > >PWM current from an un-regulated supply. I am thinking that maybe I can > >simply step down the ac line, full wave rectify it, and synch my PIC PWM > >timing to the zero crossing. I use IRF520 MOSFETs to switch the current. > >This would insure that my driving waveform would be uniform for each cycle. > >I do not believe that having a non-DC source current is any problem. After > >all, I am only heating wire. I'm wondering - why not just use a triac and NOT convert to DC? Yes, you need to sense the zero crossing. It's EASY! Just add one resistor. The app. notes show you how. -- -- Paul Greenwood -- (pabloKILLspamaustin.ibm.com) Newton's Fourth Law: Every action has an equal and opposite satisfaction.

>>how about some clever (meaning cheap/simple) ideas for PIC >>sense zero crossing from the transformer secondary. AN521, "Interfacing to AC PowerLines" works like a charm. I have 2 projects where I did just that and they have been working fine for over a year. The only real reason to sync to zero cross is if you are using SCR or Triac as the current switch. Then you let the zero cross turn off the device. If you are using MOSFETs, then you can turn them off anytime. So, set the PWM frequency much higher than 60 Hz and don't worry about sync. This does not take into account switching transient noise if that is a concern. Hope this helps, Gary Skinner, ESI

Thanks all for the suggestions. I am curious about Gary Skinner's suggestion. How much higher than 60Hz is "much higher"? I do a single PWM control loop that drives 8 separate actuators. They all function with the same master cycle. All turn on at the same time and I just toggle the bit for each off when the time for that one runs out. The 20MHz PIC is doing about 230Hz at best the last I checked. Is that fast enough to avoid some sort of beat frequency effect? Thanks, Martin Kirk Arizona State University .....mlkKILLspam.....asu.edu (602) 582-5718 On Sat, 5 Aug 1995, First Last wrote: {Quote hidden}> >>how about some clever (meaning cheap/simple) ideas for PIC > >>sense zero crossing from the transformer secondary. > > AN521, "Interfacing to AC PowerLines" > works like a charm. I have 2 projects where I did just that and they > have been working fine for over a year. > > The only real reason to sync to zero cross is if you are using SCR or > Triac as the current switch. > Then you let the zero cross turn off the device. > If you are using MOSFETs, then you can turn them off anytime. > So, set the PWM frequency much higher than 60 Hz and don't worry about > sync. > This does not take into account switching transient noise if that is a > concern. > > Hope this helps, Gary Skinner, ESI >

Martin writes: ML> Thanks all for the suggestions. I am curious about Gary Skinner's ML>suggestion. How much higher than 60Hz is "much higher"? I do a single PWM ML>control loop that drives 8 separate actuators. They all function with the ML>same master cycle. All turn on at the same time and I just toggle the bit ML>for each off when the time for that one runs out. The 20MHz PIC is doing ML>about 230Hz at best the last I checked. Is that fast enough to avoid some ML>sort of beat frequency effect? I don't really know exactly what much higher means. The PWM is chopping up the sinewave into pulses. If it chopped up each cycle into at least 3 or 4 pieces, then there should not be any "beat frequency" type of problems. I suppose you could do a lot of math and verify some results, but I would probably build a simple device and measure the RMS output as a function of PWM on/off ratio. I expect it will be pretty linear with 230Hz period. Please let us all know how the experiment works out. Thanks, Gary Skinner ESI

-> The only real reason to sync to zero cross is if you are using SCR or -> Triac as the current switch. Sorry to butt in here Gary... The main reason to sync to the zero crossing is to minimize in-rush current and its resulting EMI. Connecting the load to the mains while at peak voltage results in a high dI/dT surge that many loads don't appreciate. As well, EMI emanations will abound up into the MHZ region. An inductive load just complicates this even further. This is one of the reasons one can't (or shoudn't) use a dimmer switch with an inductive load. Regards, Dana Frank Raymond - Foxtrot Systems Ltd. Internet: EraseMEdana.raymondspam_OUTTakeThisOuTcanrem.com. Compuserve: 73362,3052