[Test-Equipment] Kikusui Freq. Counter problem

[Robert Nickels]

I'm scratching my head over a weird problem with a Kikusui FCO1130 1GHz 
reciprocal type frequency counter.    This type of counter works by 
counting both the unknown input signal and the 10 MHz timebase, and 
calculating either frequency or period with its microcontroller.    The 
counter is basically working and can even - sometimes - display the 
correct frequency, but the result varies with the  setting of the Gate 
Time control.   This is just a pot connected to a multivibrator that 
generates a pulse that tells the MCU when to make a measurement, the 
longer the setting the greater the number of pulses that get counted and 
thus the better the accuracy (at the cost of a longer interval between 
display updates).     The Gate Time control is variable from 100 ms to 
10s, and should only impact the resolution, but in this counter it 
changes the value of frequency or period that is displayed (which should 
not happen).

The 10MHz TCXO output is clean and stable, there was a bit of ripple on 
the power supply so I replaced the filter cap but it had no effect.    
The problem does not change as a function of input signal strength or 
frequency, and happens on both the direct and prescaled VHF-UHF range 
input.    There is a "check" switch that switches the input to the 10MHz 
timebase which should always display 10 MHz of course, but changing the 
Gate Time control can vary it hundreds of kHz.     I'm puzzled because 
this part of the counter function should be primarily implemented in 
firmware!

Kikusui provides an operators manual with a bit of theory and 
calibration info but no schematic.   So I thought I'd see if anyone 
might have tangled with this beast before, or similar reciprocal counters.

Thanks and 73,
Bob W9RAN