středa 26. března 2014

Remote symmetric matchbox - part 4

I finally completed the matchbox and did some preliminary testing with my network analyzer.

The remote part was inserted into a waterproof box and all external connections were waterproofed by using pigtails that go through water-tight cable sleeves.

The external connections are:
  • 2x connection for symmetric antenna (black connectors on red wire)
  • asymetric RF input (pigtail with N female)
  • power (red and black wire with connectors)
  • control (UTP cable with RJ45 connectors, waterproofed by candle wax)
Almost completed matchbox in waterproof box.
The only piece missing is the waterproofed RJ45 pigtail.
There are two things I would like point out before I continue:

One component that was not described before is the input balun. To suppress the common mode current effectively the balun has to be on the low impedance side and that is the RF input. Look at the middle bottom of the coil board and you will see two coil wires entering a black cylinder with white rims. That is a set of five T50-43 toroid cores with 3 bifilar turns. The design impedance was about 500 ohms at 3.5Mhz.

Another change I had to do was to solder a second inverter (74ACT04) on top of all the inverters I had on the relay driver board to boost the current capability. Driving two relays (coil switching) in parallel needed more juice (34mA) than a single output was able to provide.

Now back to the testing…

After I assembled the whole box I connected it to my miniVNA and used 1000 ohm power (5W) resistor as a dummy load. I then checked that I am able to tune it to look like 50 ohm load.

I was able to do it, but I quickly realized that I have made a mistake in the capacitor bank's relay wiring and the value I am sending over the control line has to be inverted (oops..). The coil switching is not affected by this as I fixed the polarity during PCB layout phase.

So currently the control signal contains the following bytes:
  1. byte ignored
  2. coil inductance; 0uH = 0x00, max uH = 0xff
  3. capacitor capacitance; 0pF = 0xff, max pF = 0x00
I again used Bus Pirate for sending the SPI as I do not have the control box ready yet. All bytes were sent using 30kHz SPI with most significant bit first.

The command that matched 50 ohm output of the VNA to the power resistor was
SPI>[0x00 0x1e 0xe9]
and the resulting values were close enough to the values predicted by SimSmith (about 100pF and 4.9uH). Check the two following images to compare:
Predicted impedance tuning in SimSmith
Measured impedance during step-by-step tuning
You might notice some small back-jumps in the tuning sequence. Those happen at places where multiple bits change (for example 0x0111 to 0x1000) and mean that the MSb coil does not have the full inductance of the three lower bit coils combined. Also the trajectory does not exactly follow the proper Smith chart lines which means there are some parasitic properties associated with the tuning elements.

In case you wonder what software I use to talk to the Bus Pirate's serial interface then the answer is Cool Term when I am on Mac and minicom when I am using linux (Fedora in my case).

I hope I will get to connecting this matchbox to the antenna soon so I can evaluate the real world usability.

Žádné komentáře: