I read the Ubiquiti Nanostation M5 instructions which say to hold the reset button down for 10 seconds to return the unit to factory defaults. Suddenly the light flashed in my head! The difference between the CW and FT8 RFI is that CW runs in very short dits and dahs while FT8 is key-down for 14 seconds. The tower, which is shunt-fed on 160 meters, is less than 1/4 wavelength from the Ubiquiti radio and its 175’ of shielded outside Ethernet cable. Even at 400 Watts, enough RF was being induced into the cable and rectified by the Zener to cause a factory reset. This was not a common-mode problem, but rather a differential-mode problem, which cannot be cured with ferrite chokes on the outside of the cable.
Now, How Do We Solve This?
There have been very few Ethernet RFI filters available for HF radio interference until recently when DX Engineering came out with their ISO-PLUS, a small box that has both common-mode and differential mode filtering. While the specifications are excellent, the unit does not pass PoE voltage. This is good news and bad news. The good news is that the reset voltage on the data pairs get blocked. The bad news is that the Ubiquiti radio needs PoE voltage on the non-data pairs to run. The solution: extract the PoE voltage from the non-data pairs before the ISO-PLUS and re-inject it after, wrapping the wire between the two injectors around a ferrite loop to kill any residual RF. This is relatively simple using commonly available PoE injector boxes which I purchased for $7.49 each on Amazon. All parts were mounted in a small watertight plastic box and secured to the 10’ mast where the Ubiquiti radio is mounted.
Does It Work?
And how! Since installing the filter I have made over a hundred transmissions at 1,500 Watts on 160 FT8 without a single problem and in the first week after installation, I added over a dozen new countries to my confirmed list on the band and even started a couple of JA pileups! This puts me as of this writing within three QSL’s of 9-band DXCC. Although 6 meters has not been open, I have done full-power tests with no problems.
This was truly an obscure RFI problem with virtually nothing written about it on the internet, so I have posted this on my site and hopefully others having similar issues in the future will find it. In solving this problem, I learned a lot about PoE and general network infrastructure. Look for me on 160-meter FT8!