After further investigation I'm starting to believe the rogue 120MHz(ish) signal I am seeing is not caused by spurious oscillation but is more likely external interference.
The interfering signal is superimposed on any real input signal. This leads to false triggering of the schmitt trigger as the real signal causes the trigger threshold to be crossed. The effect is worse for small input signals (around 100mV). In this situation the wideband amplifier gain is increased by the AGC loop so the signal / noise ratio is worse.
It's difficult for me to accurately measure the frequency of the interfering signal as the timebase on my 100MHz scope only goes up to 0.05uS/division. I measure about 6 cycles per division which equates to roughly 120MHz (+/- 10% or so). I really need a spectrum analyser!
The frequency of the signal seems to remain solid - even if when I touch various parts of the input amp with a finger.
I'm not sure if the unwanted signal is mains born or radiated.
It's still there if I move the test setup to another part of the house.
I see no evidence of the unwanted signal if the input is left disconnected.
As soon as I connect a BNC lead to a signal generator the counter indicates random frequencies. This happens even if the generator is powered down or the output signal level is set very low.
I've tried a couple of different signal generators.
A BNC lead just connected to a 50 Ohm termination does not give the unwanted signal.
A BNC to croc clip lead with the two clips joined to form a small loop antenna does produce the unwanted signal.
Judging from the counter display, attaching a scope probe to various points in the wideband amp appears to makes things worse as does attaching the scope probe earth.
Unfortunately I don't have a screened room or a Faraday cage although putting the counter inside a grounded metal bread bin seems to help!
The UK frequency allocation data suggests the signal may originate from FM broadcast or Aeronautical Radionavigation:-
https://www.ofcom.org.uk/__data/asse...k-fat-2017.pdf
For now I have reduced the sensitivity of the counter by adjusting the 'Set DC' and 'Set AGC' trimmers.
Maybe I should cut a hole in the front panel next to the input connector and fit a LP filter switch as per the TF2431A model. This is a bit drastic and will only help when using the counter to measure LF signals.
I've attached a reverse engineered schematic of the input section.
It may not be completely accurate.