TL;DR Just watch this video:

This project began when I completed the CAN-BUS data extraction from my ECU as a part of a separate project. With the engine operational data now onboard an Arduino microcontroller; I thought of all the cool things I could do with it. Spawning from that was the plan that I could replace my shift cables with pneumatic actuators and have an Arduino automate the shifting process. Here is a picture of some early ECU analysis:

The design started with a force analysis for the pneumatic actuators. I built a test rig to measure the force needed to change gears. After I had the force data, I calculated the bore size of the pneumatic actuators I could achieve at a range of pressures. I aimed for target shift force at a median pressure, so I could always increase or decrease system air pressure to adjust shift force for optimal real world operation. I sourced the Speedaire actuators from Grainger with part number 5MMF7. (If you are looking to replicate this project; read on to see why these may be undersized) The solenoids are from STC, part number: 4v110–1/8.

After receiving the components I welded the brackets and built the mechanical system. After a few bench test sessions I arrived at a mechanical configuration for shifting that seemed to work the best.

The next step was to integrate the electronics. I designed and wired a small amplifier sub-circuit to allow the arduino’s digital output signals to energize the higher power pneumatic solenoids. I then wired a joystick to be read by the aduino analog inputs to provide a shift signal. After this wiring was completed, I began to build out the embedded software to control the system. The software keeps track of the current gear selected and the desired shift operation. Incompatible requests are ignored. (shift to a gear higher than 5th, etc.) The software handles all of the pneumatic actuation to engage any gear and shift sequentially through them.

Early Attempts: Lessons Learned

I originally attempted to actuate the shift rods directly by modifying the transmission case. This allowed any gear change to be completed in two transitions instead of three transitions if using the factory gating method. I was concerned about the shift speed from the added lateral movement, but once I observed how fast the pneumatics were, I decided to use the factory gating system. So I went from an original prototype of this:

I built my second prototype to use the stock gating and actuate the stock H-pattern:

I had to modify the length of the actuating arm to get the system working well. The pneumatic cycliders were sized for the proper force needed for the original protype. Once I decided to actuate the stock H-pattern they were slightly undersized.

The above photos show just once axis completed. Once I had the longitudinal axis working, I built out the lateral axis actuation.

This brought me to the final stage of prototype development that I needed before road testing. Now I am waiting for the rest of the car to be finished to try the system out.

The faster shifting also mean thats the system will benefit form a clutch disc with a lower moment of inertia. The lower inertia dish will help save the syncros when attempting to shift rapidly. Below is the 4 puck unsprung unit I chose compared to the old 6 puck sprung hub disc.