Hi,
I must have a misunderstanding of SSRs and would appreciate clarification.

I am using a PAA140 opto-isolated SSR.
https://www.ixysic.com/home/pdfs.nsf/0/4D1B681B1709960485256B20006A0CF9/$file/PAA140.pdf

When the photodiode is enabled, everything is fine.
When the photodiode is disabled, I still get a large amount of leakage through the relays when using a sinewave (500Hz up to 20KHz). It increases with frequency. If I don't have the 22K load resistor, the leakage is like the entire signal amplitude.

Putting a square wave on it, you can see the edges leaking through.

What is the mechanism of this leakage? Is there anyway to prevent it? I expected a SSR in the 'off' state to be 'off'.

Thanks

 

What is the mechanism of this leakage?

Likely stray capacitance between the input and output, since it is proportional to frequency.
A small output capacitor to ground should minimize the effects.

 

The diodes in the FET , like all diodes are not perfect, they have a leakage .
at the mains voltage, even a few nA gives a voltage out.

If you want lower leakage, you need to use a relay isolator , possibly in series with the SSR.
Use the SSR to switch the large current, then the relay to stop the leekage.

 

You should be getting less than 1µA of leakage current but that's enough to be seen by sensitive (high impedance) equipment.

 

That device tops out at ~2kHz; trying to run at 20kHz square wave is cause of the ringing.

You didn't show the voltage on the enable line so can't determine if it is being driven or leaking, etc.

 

This effect is caused by the internal capacitance across the FET switch. The datasheet does not specify this capacitance but it could be quite high. Nothing to do with the leakage current (specified at 1uA max) - well, very little to do with it: 1uA x 22k = 22mV.
Using your circuit with the 22k load, it should be possible to work out what the internal capacitance is by measuring at a known frequency.
A capacitor across the load will help.