[time-nuts] Photodiodes for high frequency OPLL

gary lists at lazygranch.com
Sun Mar 31 23:28:28 EDT 2013

Noise adds in a RMS fashion unless it is correlated. So a differential 
amplifier can cancel common mode noise, but that is about it. The diode, 
being a single ended device, seems to me to be impossible to bias in a 
manner where the noise from the bias circuitry will cancel out.

Note it may not be the case you need the full 7GHz if it is only the 
mixer product you are feeding the photodiode. That is, the diode will 
see the difference prouct.

You can get gain and bandwidth in multiple stages. The first stage will 
determine the noise if you do it correctly. It has been my experience 
that anything high gain I try to build oscillates. [And of course 
intentional oscillators just sit there.] You will need to insure there 
is no feedback through the supply rails. In this respect, going fully 
differential will be better.

I can't speak for JFET at high frequencies, but there is a lumped model 
for the BF862. You need to create a subcircuit for it. I had spent quite 
a bit of time designing a high speed photodiode front end, but junked 
the design when a different part of the system couldn't be made in a 
cost effect manner. Anyway, you need to check the limits of the 
manufacturers lumped circuit.

The JFET bootstrap really has two functions. That of a low capacitance 
high impedance input and a low impedance output. If you read the 
literature, there are multiple transistor circuits used to get the 
output impedance lower. There isn't much that can be done regarding 
input impedance.

Trolling the net, RF transformers seem to have an upper limit of 5GHz. 
Just thinking out loud here, you can get a little gain with a 
transformer, get rid of DC, bias the diode, and generate a fully 
differential signal. It is a route worth looking into.


More information about the time-nuts mailing list