Q Band AFC

This projects aim was to build an AFC (automatic frequency control) for a Q Band GUNN oscillator with varactor input. Being used in a scientific setup, the generated frequency is fed into a resonator with Q values of 1000 and up. And these resonators are temperature dependent. They are routinely cooled down to liquid Helium temperatures. The AFC has to follow the resonator to whatever temperature. This frees the operator from continous tuning. The resonance frequency may also vary with the loading.
The setup :

the oscillator

The oscillator had a mechanical micrometer screw to adjust the frequency between 34 and 36 GHz. The varactor had a range of 100MHz or so between +5 and +15 Volts. The temperature sensitivity of the oscillator was measured to be at 1MHz/Kelvin. A peltier and enclosing the regulated side with the oscillator into some foam, reduced the frequency drift and jitter to 50kHz pp with periods of several minutes. The resulting behaviour is coming from the limited resolution of the peltier controller. Should a higher stability be required, another peltier controller will be used.

The power supply sensitivity of the GUNN and the voltage sensitivity of the varactor was measured to be in the order of 20kHz/mV. We stabilized the GUNN with a linear supply built around a LT1236, a 5ppm/Kelvin reference.
The varactor was controlled with low noise OpAmps implementing several functions plus another LT1236 5ppm/K reference. Since the Gunn supply and the varactor electronics have common GND at the oscillator case, both circuits were independently made floating and their GND conntected there. Fine tuning of the frequency (10MHz or so) was implemented as potentiometer.
The freerunning stabilized GUNN :
freerunning GUNN


One mode allowed tuning the resonator. A triangle wave over the available voltage range of the varactor was also made available to a scope as X in XY mode. The Y signal is taken from the microwave detector measuring the reflected signal.
tuning graph

AFC - automatic frequency control

There are cases when the oscillator is required to follow the resonator, eg when doing CW spectroscopy at liquid Helium temperatures. For some historic reasons sidebands of 70kHz are generated with -30dBc or lower. These are sub-mV at the varactor and yet should be above the noise in the receiver. These 70kHz sidebands are AM modulated in the resonator. A lock-in amplifier (synchronous detection at 70kHz) weighs the left and right sideband and produces a signed signal when they are different. An integrator lets the varactor move the frequency up or down until both sidebands are equall in amplitude.

The resonance at Q=1000 is 30MHz, rather wide. When fully tuned, the reflected signal is minimal. And the signal level is in the mW range. The detected sidebands are at -90dBm. Several amplifier stages provide a gain of 10'000 and more before the switch (AD630) rectifies them. After the integrator a few mV are required.
With a diode detector :
afc stabilized graph
A mixer produced even better results.


In freerunning mode, we achieved a frequency jitter of 50kHz pp after initial drifts.
In AFC closed loop mode, we achieved a stability of 100kHz pp with a preloaded diode detector and perhaps 30kHz pp with a mixer as detector.

sponsored links

projects page

last updated 28.july.03 or perhaps later

Copyright (99,2003) Ing.Büro R.Tschaggelar