6dB Attenuator

Sometimes it is useful to be able to reduce the output of a radio transmitter below the lowest power setting on the radio. This was the case when I wanted to drive a 23cms transverter from a 2m multimode radio. Technically the 2.5W output of the FT290 was within the input range of the transverter, but I figure that all the unused power has to go somewhere, and I didn't really want it heating up the transverter.

So the requirement was for a reduction of power by about a factor of 4, which is 6dB in decibels. But it would also need to handle 2.5 Watts, safely.  The input and output impedance needed to match the 50 ohm system impedance.
There is a tutorial for designing pi attenuators here: https://www.electronics-tutorials.ws/attenuators/pi-pad-attenuator.html
Which even gives the values for a 50 ohm 6dB version in the table. 150.5 ohms and 37.4 ohms.
I checked out the design in LTSpice https://www.analog.com/en/design-center/design-tools-and-calculators/ltspice-simulator.html

37.4 ohms is not readily available, however, I worked out that four 150 ohm resistors in parallel gives 37.5 ohms. Paralleling resistors also allows them to handle higher powers. Wire wound resistors will not work at VHF, because they behave as inductors. Surface mount (SMT) film resistors would be good. I found that the 2012 size SMT resistors can handle 750mW each so in theory four should cope with 3 Watts, although some derating is needed if they are close to other resistors that are generating heat. It is also possible to combine four 150 ohm resistors, using two series resistors in a parallel pair combination, to give 150 ohms. Using LTSpice, I was able to measure the current in each resistor and calculate the power dissipated. In fact R3 and R6 share most of the power, equally between them so with good heat-sinking, the required 2.5W rating will be comfortably exceeded.

Then I did something I haven't done for years. I etched a printed circuit board using ferric chloride. I cleaned up a piece of board and cut it to fit in the dicast aluminium box. Then I masked the areas of copper that I wanted to keep using PVC electrical tape, and schmoggled it about in ferric chloride for about 10 minutes, until the exposed copper had gone. Gave it a good wash and ended up with what you see at left.

On a double-sided PCB, it is possible to give the tracks a characteristic impedance, like coax cable. If this is made to be the same as the system impedance then it will help to ensure a good VSWR at the input. Using a table in the back of the VHF/UHF Manual (Edited by G.R. Jessop, G6JP, 4th edition, published by the RSGB in 1983), I made the tracks approx 3mm wide - although it is not possible to do this where the resistors are in parallel, I wrapped copper tape around the edges, so that the ground was well connected to the copper underneath.

When I came to mount the board in the box I cut three aluminium plates from 2mm thick sheet, so that the board was sitting on a 6mm thick block of aluminium. This brought the board up to a level where the centre pin of the BNC connector could be soldered to the PCB, using a very short wire. Again this helps keep the VSWR low, because lengths of wire act as inductors. I used a smear of heat sink compound between each of the plates, because I wanted the heat from PCB to be conducted to the outside of the box.

Before I finally assembled everything, I tested the board at d.c., measuring the resistance at input and output. Remember that you need to terminate the attenuator with 50 ohms if you want it to measure 50 at the input. When the output is open circuit it measures about 83 ohms at the input.

I used a bench power-supply to feed 11.2V (equivalent to 2.5W RMS) into the attenuator, and left it for about 20 minutes. Without the aluminium block, the board does get quite warm, almost too hot to touch, so I guess it is running about 60 degrees C. The aluminium helps to draw some of the heat away, so it runs a bit cooler when it is in the box.
With the lid screwed down, I fed some RF into it, and measured the VSWR. At 145MHz, the VSWR is very low. There is about 0.1W of reflected power with about 3W forward power. Pleasingly, it has a similar low VSWR at 433.000 MHz too - which is always a good sign.

73
Hugh M0WYE