It's a PWM. The amount of time on vs off is dictated by the control circuit. The feedback loop for how much difference there is between on and off is provided by the thermistor and that dictates what the FET switches. That would be the big transistor on the back of the switch, the side you can't see. I looked at this for a couple of minutes earlier and it's a more complex version of what I was expecting.
Essentially, over a period of time. say 6 seconds - at level 1, 5 seconds would be off and 1 on, when the temperature that the circuit is set to achieve is reached. For level 6, it might be 6 seconds on, for level 3, 3 on, 3 off. This "temperature to be reached" is probably of the order of 40C if the newer Octavia VCDS threads are anything to go by (and they probably are). So the control board in the switch is turning the pad on or off relative to the thermistor resistance and the dialed-in temperature. This is achieved by comparing the currents pushed back via the thermistor to the "should be" and then the pad is either turned off or on to suit. The setting of the heat level really is a 5k pot, attached with a gear to the dial.
I was expecting just a transistor but a FET makes more sense. It's a BUK7513, which can switch up to 75A. Sounds familiar - the FETs in the CCCU are quite significantly oversized.
Just getting the pad to react is a good start. But the thermistor value should work with the controls in the OEM switch, and knowing what the approximate currents are that are used for control would be a good start. The implication of the work from Thomas (A2TDI) is that the current goes up as the temperature does and that feedback current is important to stop things getting too hot. The calcs from earlier imply <10mA allow current to flow - I'd be looking to emulate this with any replacement mat.
I don't have proof of this, it's conjecture based on logic and some electronics understanding.
@Joga: I suspect it's a really slow frequency, of the order of 0,1Hz - what's the point in going faster? Still, there's an ST Microelectronics chip on there, too, a 90M something that I can't identify, so it's not just an analogue feedback loop - this is the predecessor to the current fully digital systems.
- Bret