DIY HP J06-59992A, or making two 10MHz square waves for HP 53131/2 adjustment?

[TERRA Operative]

In an effort to perform the 'Fine calibration' procedure on my HP 53132A counter, I need to supply two 10MHz square waves in various configurations to the two inputs.

Usually (according to the service manual) this is achieved by feeding the output of a HP 8130A pulse generator into a HP J06-59992A Time Interval Calibrator.
The problem is that the 8130A is a boat anchor, and the J06-59992A is rather hard to find and expensive....

Luckily, the requirements for the generated signal isn't too esoteric, basically just two 10MHz square waves, 1V Peak-to-Peak, 50% duty cycle, centered around 0V, with the ability to flip one square wave by 180 degrees on command.

The jitter should be low, and transition time ideally <1ns but <10ns is acceptable, phase difference between the two square waves should be <100ps.

There are four output configurations required, 1/2 and 3/4 may seem redundant, but they allow the counter to cancel out any phase difference etc between the two signals up to a point, making for a more accurate adjustment.

The 4 output configurations needed are:

1. Square wave A to Output A + Square wave B to Output B
2. Square wave A to Output B + Square wave B to Output A
3. Square wave A to Output A + Inverted Square wave B to Output B
4. Square wave A to Output B + Inverted Square wave B to Output A

I was thinking of using a microcontroller (although I have no idea about software stuff), maybe with the provision for a 10MHz reference input if possible for a more accurate output.
Then generate two 10MHz square waves from two pins and have an input pin used to invert one of the square waves with the press of a button.

After that, a DPDT relay on the way to the output BNC connectors so output A and output B can be transposed.

How accurate of a square wave can a microcontroller produce? I have to admit, I have very little software experience...

[aeg]

I would skip the micro and just use a TTL can oscillator and a dual XOR gate. (Dual so you have the same delay in the inverted and non-inverted outputs.)

[TERRA Operative]

I'm thinking about using a couple opamps as inverting and non-inverting schmitt triggers. Would this sort of thing give me reasonable stability (assuming a good input signal)?