Rather than buying an antenna analyzer I have built one myself mostly from spare parts that I had in the mancave. In the end this is what it became:
The measurement originates from a so-called return loss bridge, if there wouldn’t be any signal returning from the antenna into the transmitter, and, if the bridge was symmetric and if there are no internal losses then the output at B would null out. The signal at A is proportional to what we put in to the Wheatstone bride, and the ratio B/A*1000 is what I decided to call the reflection coefficient.
The oscillator is a si5351 left over from the WSPR beacon kit. It outputs a square wave and is able to generate signals up to 150 MHz. Hans Summers has on his website a library for the si3551, it goes via the I2C interface. I decided to use it between 3 and 31 MHz in steps of 100 kHz. Voltages at A and B in the bridge are measured by passing the signal through a Schottky diode in series with a 100nF capacitor. The ADC’s of the atmel328p running the Arduino bootloader do the rest of the work. On the Arduino website you can find how to build the interface to the LCD and all of that, I don’t need to explain that here. The first setup was with a standard Arduino Uno to see if it worked at all, in the second revision I put everything on a experimenters PCB.
In the photo the antenna analyzer is running via the red FTDI board on the left of the LCD display. This is only necessary when you upload the code, in the end you can also remove the FTDI adapter and put it in standalone mode, in which case the interaction goes though the LCD display and both buttons. If you leave the FTDI board in then you can also retrieve the return loss spectra shown below.
For my end-fed antenna I found:
When you zoom in on the first resonance you see that the antenna is at resonance at 7.0 MHz, which is fine, the losses increase at the end part of the 40m band, so maybe the antenna is a bit too long, no worries, without a coaxial switch the autotuner is able to match it over the full range.
When we zoom in at 20m we get to see that the resonance occurs at the end of the 20m band. The 20m band is where this end-fed antenna works best in my opinion.
At 10m the performance of the end-fed antenna is slightly disappointing, actually, it is too long here. Beyond 28.7 MHz I really depends on the autotuner to be able to use it. What I also don’t understand is why the antenna shows a first resonance at 26.9 MHz and a second resonance at 27.9 MHz. Resonances are found by looking for local minima where the the reflection coefficient is less than 390 (thus 39.0%) which is a value based on what I thought was right for this set-up.
Below is the experiment on a breadboard.
I did various experiments with dummy loads and resistors and decided not to try to convert the measured voltages into a SWR values at 50 Ohm. First of all, I did not have any 50 Ohm resistors in the shack, second, I suspect that the voltage measurements are biased and that the bridge is not fully symmetric. Also, the oscillator is outputting a square wave which assumes that the equivalent circuit of the antenna is that of a LCR in series terminating at ground. If we have a sine oscillator then we would not be affected by the returns of the signal at higher harmonics. This could be improved, for instance with the AD9832 DDS which does output a sine wave, that would be the easiest solution. Also, I could compare the reflection coefficient with a calibrated antenna analyzer.
Update 1: handheld case
The handheld version is ready, my solution is, buy polycarbonate and add two plates, this is the easiest temporal solution to carry around experiments.
Update 2: Rotary dipoles analyzed
For fieldwork I have the rotary dipoles (see further on in this blog), and here are the reflection spectra measured today with the analyzer.
Last update: 29-July-2016