This article describes a low power digital directional wattmeter based on the design published by G8GYW at https://github.com/G8GYW/g8gyw.github.io .
The complete wattmeter was purchased on Aliexpress for around $70, principally for the hardware as it was intended to replace the firmware.
Above is the seller's picture of the wattmeter, it has a different display to that used by G8GYW, different resolution but still an I2C OLED mono display.
Above is a picture of the top side of the PCB. The microcontroller is the 28MLF package at middle right. This type of package is a little more difficult for the DIYer to replace than the common (but scarce these days) 32TQFP.
Above is a pic of the underside of the PCB. Note the transformer construction, more on that later.
Experience says that the quality of SMA connectors and their attachment is a problem in lots of cheap Chinese stuff. In this case, the designer has used a narrow solder mask that prevents a fillet of solder along the sides of the pins. In some pins, the solder has flowed under the pin, and other are secured solely by the solder fillet at the very end. This was repaired by running a little ligquid flux along each pin, and heating the pin with a soldering iron until the solder wicked under the pin for the full length.
The wattmeter is supplied with SMA(M)-BNC(F) adapters of the type shown above. These are commonly found on hand held radios for adapting to a BNC terminated feedline, but for this application they are a poor choice. Experienced uses of SMA connectors know not to allow the centre pin to rotate when assembling the connector as it causes premature wear… and this type of adapter rotates the centre pin.
Not only are they an unsuitable connector from a wear point of view, the ones received were a very tight fit on quality Kings BNC plugs… Chinese Quality.
This issue can be solved by purchasing some adapters with loose nuts.
A sweep with a NanoVNA gives a picture of InsertionVSWR and |S21| (-InsertionLoss).
Above, InsertionVSWR is very good up to perhaps 40MHz, a little poorer to 60MHz. Importantly InsertionVSWR is very good down to 1MHz, most implementations I have tested crash out below 3.5MHz, sometimes below 5MHz.
Above is a plot of |s21| (or -InsertionLoss), it is quite good to 30Mhz, and again a little poorer to 60MHz. Again is it good at the low frequency end, a problem with most of these couplers I have tested.
The one design issue I do have with the magnetics is that the Sontheimer coupler uses two independent transformers, and although the way the binocular core is wound it might appear that they are two independent transformers, flux from the current flowing in one aperture results in a little flux in the opposite side of the core… there is small coupling. This is discussed in more detail at AT-100 ATU Sontheimer coupler using a single binocular ferrite core.
New firmware had been written awaiting arrival of the hardware (which was pretty quick, a few weeks). The firmware had been largely debugged on an Arduino Nano, but calibration was needed.
A very quick and approximate calibration was done by taking a set of measurements of PowerIn and the calculated detector voltage (from the ADC count, measured Vref, and nominal resistors in the voltage divider).
Above is a plot of the data points and a second order polynomial curve fit (with zero intercept). The coefficients were loaded and the wattmeter tested. This is a good enough basis for evaluating different presentations, display averaging etc.
A third order fit was tried, but it was no better on this rough data.
A further experiment with better cal data is a cubic spline curve fit (code written for another project can be used).
Above is an example presentation being evaluated, power on the top row, VSWR in numbers and a bar chart from 1 to 3 on the right half of the bottom line. (The resolution of the power value might seem excessive, it is deliberately so for firmware testing.)
Continued at Digital directional wattmeter – based on G8GYW – part 2 .