Partially because the F-22 had a really bad case of DMSMS (Diminishing Manufacturing Sources and Material Shortages). In short, because Intel stopped producing the i960MX CPU, they could not build any more F-22's unless the systems involved were totally redesigned, at great costs, beyond the fact that the component was already obsolete. Practically every attempt to redesign or replace the F-22's avionics was cancelled due to excessive costs.
In a nutshell, the F-22 avionics uses a Common Integrated Processor approach based on the i960MX processor and is written in ADA, whereas the F-35 Integrated Core Processor uses PowerPC based architecture written in C++ like standard PC software. Later production F-22's have a hybrid PowerPC and i960MX architecture and some software code is ported over, which allowed the avionics to combine legacy software with newer software from the F-35 program, but not to the level of being able to shortcut development if they resumed production.
As a result of the issues facing the F-22, there were many lessons learned when the F-35 was being initially developed:
1. A decision was made early on to use C++ instead of ADA as the programing language to broaden the programmer base and to use standard development tools
2. The F-35's avionics approach is to use “virtual machines” (VMs) extensively to avoid having core specific functions
3. The F-35's avionics software uses a middleware so that a change in hardware does not require a complete rewrite and recertification of the avionics suite
4. All major avionics components and weapon stores are connected via fiber-optic interconnects which allows for future upgrades without rewiring the jet
5. The F-35 isolates, in discreet VMs, the individual processing of avionics data then fuses the results. This isolated each VM from any bugs or problems another VM faces. The F-22 ran multiple functions in a single computer instance so if there was a problem, it affected multiple functions (remember the dateline issue)
6. The F-35 was planned for all IOC functions in its current hardware state and only every other Block requires significant hardware changes (likely adding/upgrading ICP cards)
Finally, because the F-22's development took many years, and acquisition processes aren't known for being very fast and responsive, when the decision was made early on in the F-22's to use the i960MX chip, the chip was considered state of the art. When the F-22 went into production, it was already obsolete because they held onto legacy platforms too long. And by the time the F-22 IOC'ed, you could literally find parts needed to keep the F-22's avionics and ground equipment in a museum.
The i960 RISC was not a very common platform to begin with (only the F-22 and the HAL Tejas, which is also a 1980s design that went IOC much later) which further complicated matters. Some of the newer F-22 CIP boards have more powerful PowerPC chips, along with FPGAs that have more compatibility with the F-35 and can share parts of the code base, but in order to replace the F-22's avionics, it will cost a whole lot of money even if they went off the shelf (say, took the avionics computer from the F-35), as they still need to redesign and reprogram everything else on the aircraft to be compatible. For one, the EW receivers on the F-22 are simply not compatible with anything else, and the radar is also a generation behind the F-35's radar even though they share common development heritage. Also, porting the avionics software over from ADA to C++ is also essentially a total re-write of the software package itself, which is a lengthy process in itself, and very expensive.
In my opinion, avionics maintainability and upgrade costs could actually get the F-22 retired to the Davis-Monthan boneyard by 2030. The F-22 is a great example of how NOT to design avionics systems for future updatability.
Отредактировано dell (2018-07-24 13:00:55)