I was running my U3S for a while in WSPR mode here in Arizona on 40 meter and collected nice overseas spots with about 1 Watt of output into a Windom antenna, as far as to Perth Australia, Antarctica, South Africa, Japan, New Zealand and others.
One night I surprisingly just got only one single spot in Seattle and nothing else. What is wrong? I found I forgot to switch on power to the power transmitter stage in the U3S. So, the driver IC delivered a few milliwatt output that reached the antenna and made the spot. A few milliwatt for 1200 km! That awoke my interest in low power propagation in WSPR mode!
I needed an attenuator with a defined dB value, better two of them, one with -10dB, the other with -20dB attenuation. I always wanted to build or buy attenuators for my ham experiments. Some time ago I purchased project boxes made of extruded Alu (Amazon Part B010DHQLEI). The SO239 socked fit right on both ends of the box instead of the end plates. I never finished this project and I lost interest because I could not easily find the proper resistor values in the right wattage.
It happened that I needed a bunch of 1kOhm 1/4 Watt 5% resistors for another project, and I ordered well too much. I had three feet of taped resistors left over. Would these resistors be useful for my attenuators? I got the numbers for the Pi version of the 50 Ohm, -10dB attenuator. R1 = 96.25 Ohm at both ends and R2 = 71.15 Ohm. I choose 10 parallel 1K resistors for R1 and 14 resistors in parallel for R2. Doing the math in reverse for the selected resistors results in 9.85 dB attenuation with an impedance of 51.6 Ohm and a maximum power of close to 5 Watts. That was perfect for my requirements.
What about the -20dB attenuator? 16 resistors for R1 at both ends in parallel and 4 resistors for R2 results in -19.91dB attenuation and 51 Ohm impedance. Perfect again.
I squeezed all these 1 kOhm resistors into the boxes, paying attention to good soldering joins a solid ground connection between the two SO239 sockets. Tests with the NanoVNA showed results well within my targeted +-5% accuracy for attenuation and impedance from the low HF into the UHF band.