Spectrum Analyser Mixer Dissection - Page 3

OK - with a little methylated alcohol, prise off the black plastics. It sounds gross, but by re-arranging the external hardlines, the mixer could be positioned to run "live" with the lid off, fed from the tracking generator. The plastics do have some kind of effect on the trace, but its all so bad I don't know why they are there!

YUK! By now, the hackles are rising, and I am beginning to mutter that it's probably the work of some recently teenage scribbler now well into RFICs. I so unfairly malign them when I get this way, and I wonder what kind of affliction is it I possess that I care so much about what is, after all, a transient piece of RF hardware that most sensible folk could not work up too much regard for.

Onward...how does that mixer balun work? Wilkinson splitter? Er.. no, well.. sort of - but this one has 51 ohms in parallel with a capacitor on the input, and no 100 ohms across the legs. How does the IF ever get connected? Hmm.. gotta get a look at the underside and find out..

Well, surprise! This is much more like the finished construction I would expect in a piece of quality kit. First notice the hardline entry, very like the one topside, but without the silly piece of bent wire approximation to a Lecher Line. The coax centre goes to a microstrip stub short circuited at the end. While in some respects, it has some of the DNA of a Marchand Balun circa 1944, spread out straight and morphed into microstrip, I know it is not so, because a Marchand Balun has a coupled line that would extend the length of both strips, and has a open circuit end.

This trick of being allowed to solder the hardline onto the top of the microstrip, and having the ongoing transmission line take the centre reminds me of the so called "infinite balun", used to drive the feedpoint of antennas by sneaking up inside the element coaxially, from a grounded part and forcing a standing wave on the outer of the sheath. While I don't yet understand all of it, I did discover this coaxial balanced input technique being used as prior art in these patent references.

Underneath the hardline, and also under the strip that the centre conductor connects to, is groundplane, not a broadside coupled strip, or anything. Here is where I need a little help in understanding what is going on. The hardline outer RG405 is 19.45mm long. The strips are 16.0mm long, the legs of the fork are 13mm before the 40° mitre. All strips are 3mm wide.

But what about the two shorter strips completing the right triangles? (Here I strongly suspected that they were directly over the legs of the "fork" on the opposite side, and indeed it turns out to be so. I am thinking that makes the business bit of the structure in the middle a Suspended Substrate Stripline Coupler. Its connections are by two hardlines with microstrip stubs and a direct approach via a resistor-capacitor combination at the stem of the "fork" on the topside.




Page 1 Special Mixer Mess (description of the problem)

Page 2 - Spectrum Analyser Mixer Dissection - Taking things apart

Page 3 - Spectrum Analyser Mixer Dissection - A closer look

Page 4 - Spectrum Analyser Mixer Dissection - Oh Yes - it is still full of surprises!

Page 5 - Spectrum Analyser Mixer Dissection - How it ended up

How We Finally Got it Right (and all the secrets too!)

The Result - Performance Traces

Some (not so good) Traces


Author : Graham Seale