I note that there are no noise specifications in the data sheets however, I
want to switch nanoamps & so it might be a factor. I haven't homed in on a
particular device yet.
What are the noise characteristics of these devices?
Is there a 1/f aspect to it?
Can I safely assume the device is a low value resistor for noise purposes?
This is not something we test or characterise for any of our switches so we
can't guarantee any numbers.
However, I do not think voltage and current noise of the switch is likely to
be a significant consideration in your application. Noise from other sources
in your circuit is likely to be far more significant.
The construction of the ADG733 (and most MOSFET switches) is a PMOS and NMOS
transistor connected in parallel.
For any MOS transistor there are essentially two voltage noise components.
Johnson is essentially thermal noise and is due to the channel resistance.
1/f noise is present in MOSFETs too and typically you can expect the corner
frequency to be quite high at around 10kHz. However, the broadband noise is so
low that this is unlikely to be significant.
You can make a good approximation of high frequency noise (above the corner
frequency) from the Johnson noise equation:
En = sqrt(4kBRT)
Which for the ADG733 assuming an on-resistance of 20Ohms is calculated to be
Current noise in MOSFETs is extremely small and will be in the sub pA/sqrt(Hz)
for the ADG733.
Since noise sources add as route-sum-of-squares, any noise source which is 3 or
4 times larger than other noise sources will dominate and you can effectively
ignore the smaller noise sources.
This is not a quantitive answer to your question but it should convince you
that noise in the switch is unlikely to be a problem unless you are working at
low frequencies with other very low noise components. I suspect that other
errors will dominate such across-talk, off-isolation, noise in other parts of
the circuit. If this does not satisfy, please forward details of your
applications and why you think voltage/current noise in the switch is likely to