Varactor diode speed question

[jgraygoza]

Hello everyone.
I have a very small and simple question.
What is the best way to fast switch a varactor diode?
I ask this because I built a couple of tunable filters using varactor diodes and they work pretty well,
except the fact that they are very slow in switching.
I have built other filters using pin diodes and they are pretty fast.
But I am looking into the possibility that I could make one using varactors to meet size limitations.


Any suggestions would be greatly appreciated.

Thank you very much!

 

[biff44]

varactor diodes sometimes confuse those that do not know the underlying physics. They are PN junction diodes used in reverse bias mode. As such, an area at the PN junction gets swept out of any charge carriers. At a certain reverse bias voltage, the width of this depleted layer is fixed. The P material to one side, and the N material to the other side look like ohmic contacts. The structure, as far as microwaves see it, is simply a parallel plate capacitor, where the plates are that depletion width. Lets say you are at 2 volts reverse bias, and at that voltage, the diode looks like a 1 pF capacitor.

BUT, lets say you want to change the reverse biase from 2 volts to say 5 volts. The depletion width has to get wider. To make it wider, you need to remove electrons from the N side and holes from the P side until the 5 volt reverse bias sets up an equilibrium condition. At the new bias voltage lets say the new RF capacitance looks like it is 0.6 pF. But as far as the bias circuitry is concerned....all that charge had to move out of the diode...looking more like maybe a 10000 pF capacitor (just a guess) was changed!

So when you want to make your bias network very low impedance to allow a lot of current to flow whenever the bias voltage changes. IF there are any series resistances in the bias network, you would have to parallel them with large capacitors to provide a spike in charge/current to accomplish the change. If you really need to move the bias point fast, you might have to overshoot the voltage temporarily to provide more of an impulse in charge.

 

[jgraygoza]

Thank you for the brief refresher on the physics of the varactor diode.
So, in other words, if I wish to switch voltage fast, I have to use some sort of RC charge/discharge for the voltage that gets placed
on the diode so that the displacement of charge is done much quicker rather than changing voltage at a steady rate, correct?

 

[chuckey]

No you need transistors that can shift a lot of current quickly, if you use a series R you automatically introduce a low pass filter which slows the charge removal rate, hence biff44's suggestion for "speed up" capacitors across any series R. Use the smallest value of RF choke to keep RF out of your biasing networks.
Frank

 

[RCinFLA]

The greatest consideration is the circuit topology to drive the varactor at the cold RF side and having a low resistance at the RF side, like the coil it is tuning to ground.

If you have the varactor in a low sideband noise oscillator you want chokes feeding the bias to eliminate adding biasing resistor KT noise to the oscillator.

Basically you don't want RC time constants in the tuning path. Use chokes or direct low side, low RF impedance drive to varactor.

 

[vfone]

Just do not use big value bias resistors and the problem will be solved.

 

[biff44]

You need a high speed op amp with some current output capability (at least 100 mA). You also need to protect the varactor from inadvertent burn-out if forward biased. So there are two basic circuits:


The single ended rail-to-rail op amp can not blow out the diode with a forward bias, so you can hook it up directly. The dual supply op amp needs some current limiting device, such as the resistor, but you need to parallel that resistor with a capacitor to allow a spike of bias current to go in/out of the diode at each transition.

I do not show the RF bias circuitry, but obviously you want it to be relatively low shunt capacitance, series inductance, and no series resistance.