Microwave Construction Discussion Page

[Ray Cannon]

This topic is for a reference presenting useful construction techniques and practices specific to microwave.  It is not meant to be a Q&A page - for questions and answers please post them as normal messages.  

[Ray Cannon]

To start things off...

One of the interesting issues encountered at the microwave frequencies is that stray signals may form unexpected feedback paths through the air inside that cabinet.  An oscillator is an amplifier with positive feedback.  This feedback can come from any other nearby component via the air.  This is especially true at microwaves because short wires and circuit board traces act as antennas at these frequencies.  The question arises how can we minimize potential feedback paths.

There are multiple techniques involved in commercial practice.  An obvious technique is to put shields around each amplifier with bandpass filters on its inputs and outputs and bypass filters on DC supply lines.  Depending on how many stages there are and area available this may not be practical.  Another guideline in to avoid running parallel wires or traces which can increase the likelihood of feedback.  In commercial microwave amplifier circuits the signal path is laid out in a linear fashion.  Also when interconnecting circuit board use coax for RF connections.  The coax should be high quality with either a single or double braid shield.  Do NOT use audio cables as their shields are not braided and may be nothing more that aluminum foil.  For microwave work I prefer UT-141 or UT-85 for long runs however it is difficult to work with.  RG-178 or RG-316 will work for very short runs (a couple of inches) and is much easier to work with.  Be sure to ground the braid on both sides of the connection.  Another technique is to mount the boards at right angles to each other to minimize the coupling.

Okay ... I have tried my best but there are still unexpected oscillations that are not simply harmonics - what can I do?  The next step is to try to determine where the feedback is.  To do this you move a probing tool around inside the box to try to squelch the feedback.  A surprisingly effective  and inexpensive tool for this purpose is a human finger. Poke around and see if the feedback disappears.  Mike has mentioned he likes to use a number 2 lead pencil.  Remember the feedback can also be a result of a bad ground so use the eraser of the pencil to actually push down on the individual parts.  That is the was I found a bad connection in my 10 GHz project.  A question is how do I know the problem went away?  One can use a power meter and if you see output when keying the microphone while in CW it is a s

There is also material designed for RF absorption.  I have some and we could try it to see if it disrupts the feedback path.  If you find a place a finger blocks it that is probably a good place to try this material.  Digikey sells a version listed as 1168-1389-ND.  I used a small piece to kill an oscillation in my SG-Labs 2304 transverter..

I welcome others to post their experiences and advice dealing with errant oscillations.

[Ray Cannon]

You can also erect grounded shield barriers between different sections of your circuit.

Coax -- RG400 is also another low-loss 50 ohm option.

The absorptive material is sold to cover different bands.  Some vendors have trial packs with different samples of rubber-like sheets that are about 1/16 of an inch thick, but each with a different permeability. 

--Konrad Roeder, WA4OSH