I would like to use PWM to limit the power to a 1kw resistive load @ 240vac, but have not found a way to handle the switching.
So at a PWM of 50%, the power would be limited to 500w, at 25% limited to 250w, etc.
My first thoughts was to use a thyristor circuit, but unfortunately they are not able to switch and recover at 500hz (the speed of the PWM).
Has anyone any idea how to step up the PWM output and use it to control relatively large loads?
Normal PWM just isn't very compatible with variable AC power control. Variable control of AC usually involves using a thyristor device (triac or back to back SCRs) where you delay the normal turn on time from the normal zero crossing of the input AC voltage source that occurs every 8.333 millisec. Once the thyristor is turned on only the next zero crossing of the source AC voltage will turn it off. Controlling such a device requires an accurate AC zero crossing detector circuit and then precise timing of the delayed turn-on signal to the gate terminal of the thyristor according to the duty cycle value you want for each positive and negative AC cycle.
Lefty
Look into arduino power shield discussion:
http://www.arduino.cc/cgi-bin/yabb2/YaBB.pl?num=1236998971
The aim of the project is to use surplus 'solar' generated power to heat domestic hot water, hence I must match the supply and demand to avoid supplementing the 'free' solar power with 'paid for' power from the grid. I have a Arduino which can determine how much spare power I have at any moment in time, and wanted to use that data to restrict power flow.
The electricity supply meter which monitors power (for the supply company) measure 'real power' which is averaged over a full cycle, so I need to chop into that current waveform to ensure that on average I do not exceed the 'free' power.
Hence you will see that I cannot simply turn it on and off like a thermostat.
KE7GKP:
You must think we are clairvoyant if you expected us to understand that from your original question.
What part of my original post is ambiguous? I thought that my objective was clear..
KE7GKP:
Incomplete. Not ambiguous. You mentioned nothing about "matching supply and demand" or "chop into that current waveform". It appeared to be a conventional thermal control situation from your original statement.
Didn't the fact that I was using PWM, and matching power output to PWM % suggest that I was wanting a proportional power output?
I'm looking for a little guidance & help, not an argument!
I am aware that mosfets are able to handle the frequency and current, but not the voltage, so I'm unsure how to develop this further.
OK, back to the drawing board!
Use triacs/relays and several heating elements in power-of-2 power ratings, 125W, 250W, 500W, 1kW? You will probably confuse the metering systems if you switch large power levels rapidly with PWM or zero-crossing-switching (do they account separately for each half-cycle??).
Using several elements statically switched is guaranteed to be a well-behaved load, if you can find the right heating elements that is!!
Also putting an unfiltered PWM load onto the mains would violate your contract with the power company as you would be putting significant harmonic energy into the grid - (actually more likely to play havoc with your grid-tie inverter, thinking about it).
It might be possible to use PWM if you
(a) bridge-rectify the supply.
(b) have a high power LC low-pass filter to isolate the high frequencies from the supply
(c) push some MOSFETs to their limit - in fact IGBTs are probably better at these voltages.
(d) take care not to send the LC filter into resonance.
A few more relays and heating elements is simpler 
The aim of the project is to use surplus 'solar' generated power
IMHO, solar power isn't fast changing variable, 1-10 second, probably? There is not much difference with thermostat, and why PWM was brought into discussion isn't clear.
The question is how solar power delivered to heater? Is it converted to AC?
If so, than dynamic power distribution could be accomplished by synchronizing DC / AC converter to main grid, and shifting phase of voltage produced by solar source.
http://www.av8n.com/physics/power-grid.htm#sec-phase
If not, than it become not " power sharing " situation, but rather dynamic load, when
all depends how much power solar generated, some "extra" could be delivered to heater, or to battery - that is most common.
MarkT:
Use triacs/relays and several heating elements in power-of-2 power ratings, 125W, 250W, 500W, 1kW? You will probably confuse the metering systems if you switch large power levels rapidly with PWM or zero-crossing-switching (do they account separately for each half-cycle??).Using several elements statically switched is guaranteed to be a well-behaved load, if you can find the right heating elements that is!!
Also putting an unfiltered PWM load onto the mains would violate your contract with the power company as you would be putting significant harmonic energy into the grid - (actually more likely to play havoc with your grid-tie inverter, thinking about it).
It might be possible to use PWM if you
(a) bridge-rectify the supply.
(b) have a high power LC low-pass filter to isolate the high frequencies from the supply
(c) push some MOSFETs to their limit - in fact IGBTs are probably better at these voltages.
(d) take care not to send the LC filter into resonance.A few more relays and heating elements is simpler
Mark, unfortunately I can only physically fit one heater element, but did consider having 3 outputs from the Arduino;
O/P 1 would provide the full load
O/P 2 would feed the load via a Triac - predetermined to fire at half power
O/P 3 would feed the load via a Triac - predetermined to fire at quarter power
The sketch would determine which output to activate.
But your suggestion of using a Mosfet/IGBT via a bridge rectifier seems a much better idea, and one which I had not considered.
Presumably, I would need to isolate the Arduino from the mosfet/IGBT, maybe an opto-isolator?
Do you think that this is approach is viable ("might be possible"), or should I throw the towel in!
Its theoretically OK, but quite a beast - high power/high voltage switching circuits explode if mis-designed, I certainly wouldn't want to risk it (besides there's the expense of the filter) - there must be a better way. Perhaps use a low voltage heating element and a programmable SMPS to drive it?
(surely selling back to the grid is easiest?)
This was prompted by a query in the general area:
Any of these of any use?
http://www.crydom.com/en/Products/Catalog/AdvancedWebPage.aspx?CategoryText1=Control%20Relays
Thanks Dave, I see that they do a 0-5v controller which may do the trick. I'll do some reading!