Early in my career, I worked as an engineering technician in the R&D department of a company that designed and manufactured active signal processing components in RF and microwave technologies. An engineer at the company had been assigned a project to come up with a design for an RF two-power power splitter with an insertion loss from 100 MHz to 1 GHz of +/- 0.25 db across the entire frequency span.
The physical design of a splitter is fairly simple: It consists of two miniature toroids, wire, one capacitor, and a gold-plated flat pack. Well, one morning, the design engineer had me wrap the toroids with No. 36 AWG wire, forming an auto-transformer on one-half of the toroid, and a simple balanced coil with a midpoint tap forming two equally wound halves to create a splitter with two output ports, each having an impedance of 50 ohms. The tap of the coil would be attached to the tap of the auto-transformer, and the capacitor connected at the junction of the taps to match the input port to the two output ports for minimal insertion loss.
After I had wound the two toroids, I mounted them and the capacitor into a metal flat pack (to be hermetically sealed). We took the device over to the test bench to run the device through its test parameters, measuring the standing wave ratio, input to output ports insertion loss.
We readily noticed that the insertion loss was measuring about +/-0.5-1.5 db. We were looking for an insertion loss of +/- 0.25 db across the entire frequency range. For a couple of days, the design engineer went back and forth between calculations and testing. He finally gave up, claiming that the insertion loss requirement was not doable. As the design engineer, understandably frustrated and angry, got up from the test bench, I asked him if I could try something. At that point, he really didn't care, so he let me take a shot at the design.
Now, this is where the analyzer comes in. I had noticed that the insertion loss from input port to either output port was cutting off short like a low-pass filter. This told me that toroids were not wound with the right wire gauge, and that the auto-transformer toroid needed more twists per inch and a change in wire gauge to achieve a different characteristic impedance.
I went to my workbench and began winding some new toroids with No. 37 AWG THN wire with about 10 twists per inch. The number of turns for the auto-transformer and the output coil remained the same, and I kept the same value of capacitance that would balance (match) the input impedance to the two outputs of 50 ohms each. This meant that an impedance of 100 ohms was needed at the junction of the two taps and capacitor in order to have a balanced split.
When I was finished winding the toroids and mounting them, and the capacitor into another gold-plated flat pick, I took the device over to the test bench and ran the test parameters. Insertion loss was +/- 0.25 db flat across the entire frequency span.
At first, I couldn't believe I hit it on the first try, so I made certain my frequency generator was set correctly, and that the RF analyzer was properly zeroed out to account for any loss introduced by the test setup and test fixture. Once I was certain my setup and measurements were correct and repeatable, I called the design engineer back over to the test bench and showed him how this design met the customer's specifications.
He then took over the testing himself and came up with the same measurements. We made sure the design was stable by building a few more of these two-port power splitters using the design I came up with. The design was repeatable. He then had me take these packages, hermetically seal them, and run them through ambient air, elevated heat, and cold according to DOD specifications. The design was solid, and the insertion loss and standing wave ratio remained within specifications.
There you have it -- a design that was abandoned and brought back to life in a spectacular way.
Why is this situation memorable? This story is quick and simple, but it gave me confidence by demonstrating my ability to interpret analyzers. And it was my first job in an R&D department, and I was looking to make an impression.
Submit your product repair or redesign story as part of our Frankenstein's Fix competition on EE Life, and you could win a Tektronix MSO2024B digital oscilloscope. The deadline is Oct. 26, 2013. Submission details and full contest rules here.