Channel 2 of a 7A18 showed a strange square-wave response on all input settings below 500mV/div. This made me suspect a capacitive or resistive loading error at the back end of the attenuator, as with only small values of attenuation the 'scope input is seeing a higher source impedance.
To cut a long story short, it turned out to be C412, a 1p8 tubular ceramic capacitor in series with 560 ohms and shunted across the input. It measured 44pF! This is the second time I've found one of these caps going dramatically high in value.
John
John Sykes