The probe compensation signal (referred to as the calibrator in the service manual) is different than most other scopes as Jon notes. It will produce a 5 cycle square wave on any sweep rate set between 100ms and 100ns/div. There is a 200ns shift added to the chop signal at the end of some sweeps to desynchronize the chop signal from the sweep rate. The frequency of the calibrator is controlled by the same divider chain that controls the vertical chop rate. So the 200ns can show up in the calibrator signal as irregular pulse width. Apparently it only shows up when measuring the calibrator signal with a separate instrument. But maybe try measuring the signal with trigger set to Single sequence to shut off the sweeps. see if it disappears? Also, the comp signal has output impedance of 50ohms. Maybe try measuring with output set to 50ohm to match impedance?
Frank Mashockie