8133A Pulse Generator CH1 falling edge glitch

[jhenderson0107]

No good deed goes unpunished. 

Recently, I obtained a sampling scope.  This has revealed a problem with my pulse generator which was invisible on my 3054T oscilloscope.  Specifically, both + and - differential outputs of channel 1 of the 8133A exhibit a glitch on just one edge of each pulse, as depicted in the attachment.  The problem remains even if the outputs are inverted via the COMP button.  The problem is not mitigated by changing signal amplitude, offset level or phase.  The differential outputs on channel 2 on the 8133A behave normally. 

Has anyone encountered this phenomenon and implemented a fix?  I have downloaded the service manual, but hope to benefit from others' experience before diving in.