The ITU database has a clue: on 20.6°E, a USA satellite (nick)named USOBO-5A is registered.
Transponders (bandwidth):
20252.5 (4 MHz)
20452.5 (4 & 5 MHz)
20462.5 (5 MHz)
20752.5 (200 MHz)
The trouble that bothers me now, is the SBIRS-Geo 2 TLE, that apparently has the correct belt crossing time, but wrong longitude. I can trace back at least for half a year, the TLE always had 19.2...19.4°E. But always found this rather strange, another satellite so close to the 19.2°E Astra constellation.
The SBIRS-Geo 2 discussion don't really belong in this thread, but need to finalize it!
After PM/email discussion with Llew and others, it is now confirmed Llew received above 2 frequencies from SBIRS-Geo 2, that is located at 20.6°E (nominal).
The wrong longitude values (19.3°E) for SBIRS Geo 2 that I mentioned above are caused by a DLL called DJTSabLib.dll that GorbTrack, WXtrack, and probably also some other tracking softwares use.
The bug is apparent only on some inclined satellites.
I can't figure out the further conditions leading to wrong longitude output!
Unfortunately, the author of the DLL has told me it is very unlikely he will fix it.
It has hardly been maintained for many years. It is open source though.