Next step appears to be looking at synthesising a sinewave using a square wave. This has the advantage that the square wave is much more efficient on the transistors, but after filtering, the load sees a sinewave.
Using a technique called Pulse Width Modulation (PWM), we can actually generate a sinewave simply by using pulses (the square wave). By changing the width of the pulse, we can make it look like the
average voltage is higher or lower.
The easiest way of synthesising the correct width PWM pulses is with a sawtooth wave and the sinewave we want. When the sawtooth wave is at the start of a tooth, we start our output pulse. As the sawtooth ramps up in voltage, we compare it to the value of the sinewave. Once the sawtooth gets to the level of the sinewave, we stop the output. Now, this requires the sawtooth wave to be higher in frequency than the sinewave. This probably doesn't make much sense, but here's a picture so you can see!
The top wave, is a 180 degree section of a sinewave. Underneath is the sawtooth wave, at approximately 80 times the frequency of the sinewave. Lastly is the approximation of the PWM wave that would represent the sinewave.
http://www.hawks.dircon.co.uk/aoa/pwm.gif
The output is a bunch of pulses, where the width approximates the sinewave output. Filtering this by passing it through an inductor, and you have a sinewave.
Unfortunately, this is more complex, and requires more parts...
It needs an inverter to generate +230V, and then four transistors connected in a bridge to control the PWM pulses.
By seperating the inverter from the power output, we can control the voltage easily - we turn down the inverter, but continue the same PWM frequency. That way the voltage goes down, but the frequency remains the same.
The inverse is also true. Just by changing how quickly we send the PWM pulses, we can change the frequency with changing the voltage at all.
I'm thinking of building a digital controller for this on a testbed initially, to see how it works at 12V. I it works well enough, I'd like try it on the 230V generated by an inverter. The inverter would take the power from the 12V line, so no real mains will be available. This will also isolate the pump from the mains, so that a massive pump failure won't cause live mains to appear in the water cooling system!
Yes, I know it's complex, but I'm hoping it'll work. I'm thinking that the PWM waveform will be stored in a ROM as a table, and that a counter can just step through the ROM. If I change the speed of the counter, the frequency output changes.
AidanII