(12-30-2008 01:23 AM)Gary Wrote: Think of the PWM as a light switch, flipping on and off very fast.

Think of the electrolyser as an inefficient capacitor.

Because of the capacitance, the current can stop intermittently without stopping production.

This is how I understand it, in simplest terms.

Think of each single cell more as a leaky capacitor, rather than an inefficient one. The electrolyte between the plates acts in the same manner as a resistor connected between the plates would do.

Visualize a standard electrolytic capacitor with a resistor connected across the positive and negative terminals of the capacitor.

Apply a voltage across the capacitor terminals and the capacitor will charge at an inverse exponential rate, fast at first, then slowing as the capacitor charge approaches maximum.

If we then disconnect the applied voltage, were there no resistor present, the capacitor would remain in a charged state for a long period of time. However, we do have to consider the resistor. This would cause the capacitor to discharge at a rate depending on the resistor value, and charge state of the capacitor. Again, the capacitor charge would discharge rapidly at first, then more slowly as the charge state reduced.

What a PWM is used for is to periodically boost the charge state of your capacitor by, as Gary says, switching the applied voltage on/off, and thereby maintain a reasonably constant voltage level across your capacitor. This is dependent on several factors.

Applied voltage, capacitor resistance, and size, in uF, of your capacitor, and the switching rate and mark space ratio of your applied voltage from the PWM output

What you have is a method of applying voltage for less than 100% of the time, but, with self discharge of the capacitor/resistor, maintaining internal current flow through your system as the capacitor discharges when the applied voltage is discontinued..

Think of this as being a bit like the way in which a switch mode PSU works, but with less components (inductors, feedback loop monitoring, etc), where your resistor represents load across the SMPS.

Not an exact analogy, but close enough, without getting too complex.