If you are in a hurry to get it, go here.
F90X and N90S are exactly the same models inside as well as outside and only model name tells them apart. The same applies to F90 and N90 pair but they differ slightly from F90X/N90S. I will therefore refer to all of them as F90X simply because I own one!
Nikon F90X may be reprogrammed. It has three NEC microcontrollers inside but none of them have Flash memory that is they cannot be upgraded to a new firmware. Microcontrollers' RAM is volatile and its contents is lost when the camera is powered off. However camera uses EEPROM chip to store things that it should not forget - frame number, current shutter speed, operation modes, etc. Each time any control is changed the Nikon writes new settings to EEPROM. By the way EEPROM has limited life expectancy - just about 1,000,000 write cycles. Go figure.
Anyway the same EEPROM keeps shooting data that you may accumulate and download with Photosecretary or alternative PC software via MC-31 cable interface. It also contains adjustment parameters that approximately equivalent to adjustment screws on older mechanical cameras. Say, fine-tuning shutter speed within very fine limits is individual for each camera and is done during final testing at the factory. By the way this is the only setting that is different for each camera.
Since Nikon started to make electronically controlled autofocus cameras their service interface to get to these parameters was DX contacts pad. One of microcontrollers is able to sense when DX pads are being used for serial communication and switches into "service" mode.
F90X generally follows this rule but it has MC-31 that provides very limited communication functionality - it allows write access to only a part of EEPROM that contains shooting data and user settings. Say shutter speed. Write over new value and basically you are controlling camera's settings. That's what Photosecretary in its camera control mode does. MC-31 interface has no provision for writing into arbitrary portion of EEPROM.
However even though it's small, it's still a computer. Just go from a backdoor. Use buffer overflow. Get the full control. However first obtain full user documentation for NEC microchips, learn it, build an electro-mechanical model of the camera, experiment and find a small hole to squeeze into. Good luck! I've been there and I'm not coming back. Maybe later... Some mirror lockup hardware fix :)
We are getting to the point that is useful to an ordinary person (if you call F90X owner an "ordinary":). Getting the camera leave film leader outside when it rewinds the film back into canister. It may be done and even the length of remaining tip can be controlled. I like 1cm (or 1/2") - unobtrusive, will not be trying to mess with shutter blades and difficult to mix up with unexposed film. Nikon will tell you it can't be done via MC-31 (and they will be right). But I don't care.
If you want it too - welcome! The software price is zero dollars and zero cents. But there are some rules you have to stick to. First, you have zero support for it so you are on your own. Second, you are not allowed to use it commercially - if you do it for a friend you should do it free of charge. If you use it in your business - you use it commercially and you are breaking my rule. Currently there is just one shop that has the right to do it for money. This is the place where I buy all my equipment.
Again you are on your own. If camera locks up with Err, try running software again few times and then switch camera off and on again. If nothing helps send me a message I may have time to help. Software was made with a sledgehammer 2.0 and runs in DOS mode on COM1 only. It works for me.
So get it here for F90X and N90S or here for F90 and N90 .
Good luck!
Official disclaimer: I have no relations to Nikon or its affiliates world-wide as well as its employees or associates. I possess no proprietary knowledge or documentation and own only Nikon consumer products acquired via purchasing. Any information and software presented here is the result of my independent product behaviour study and therefore fully legitimate. The algorithms for changing products functionality were developed by me based on common and freely available information. They constitute my intellectual property and are copyrighted by me. Any commercial use without approval is illegal. I am not responsible for results from use by others (including you!) of this software and information.