One thing that should be a guideline is C-Bill cost per ton of item, as a reflection of relative complexity.
For example, IIRC standard Mech armor is 10,000 C-Bills per ton, while a PPC is 200,000 C-Bills for a 7 ton weapon. A PPC is ~28,571 C-Bills per ton.
My preference would be buying Research points initially equal to cost per ton/100, then you buy more research points, and roll a D100 under the second set of research points.
So researching Mech Armor might need 100 Research points initially, then you purchase 5-10 Research points per turn, and roll a D100 at the end of each turn. If you roll under the second total, you succeed on that project. So if you had bought a total of 120 RPs for the Mech armor project (because you are tired of using Commercial Armor), you have a 20% chance at the end of that turn to succeed.
Researching PPCs means you need to buy 285 Research points initially, then buy the remaining 100 to try to succeed. There would be the fun option of rolling 100, where the project simply failed and the PC cannot restart the project without additional data (i.e. ally provides full specs, data copy from a neutral, espionage, stealing a PPC, or just getting shot by one). Different types of additional data give bonus RP for restarting the project.
The other option would be tying Research point cost to the weapon cost per ton. Assuming RP cost is equal to item cost/ton divided by 100, that means each Armor Research Point is 100 MC-Bills, while each PPC Research point is 285 MC-Bills.
For scientists, you might even have it where scientists need a certain skill level before they can even contribute. Schools can train better scientists, but are the trainees worth it?