The signal itself is generated from the counter on the 9171. The output of the counter can be adjusted in 1 tick increments (from 2 timebase ticks up to 2^32 timebase ticks), so you can fine-tune the signal in 12.5 ns increments when using the 80 MHz timebase on the 9171. This output is passed through the digital front end of your choice (9401, 9402, etc.) and available on the I/O connector.
The module used for the front-end (e.g. 9401 or 9402) will affect the general shape of the pulse (e.g. rise and fall times) coming out of the module. However, the precision of the pulse generation should be determined entirely by the counter on the 9171.
I'm not aware of an actual jitter spec for this setup--the timebase on the 9171 should be accurate to within 50 ppm but it doesn't seem to have a published jitter specificaiton (perhaps somebody at NI with access to the oscillator datasheet could look this up if it's something you really need). The "pulse width distortion" on the modules should be less than 10 ns and 25 ns for the 9401 and 9402 respectively, but this doesn't necessarily give a good representation of consistency from period-to-period.
Adding the timebase accuracy and pulse width distortion specs should give you an upper-bound on pulse accuracy. Jitter would be well-within this value but I can't give a value for it based on what is published in the datasheets. The actual rise and fall times of your signal would depend on the load you are driving so I imagine it might be difficult to get an actual number for this from anybody else--if precise characterization of jitter is important for your application I'd suggest trying to measure it under the conditions you intend to be operating under.
Hope that helps!
Best Regards,
John Passiak
