Pneumatic Sequential Shifter
 

I'm running a 5 speed dog box in my car and the manual shifting is difficult for sprints and launches. It's a circuit based box so it has 1st in the standard 5th location above reverse so that 2nd through to 5th are in the left most 'H'. This becomes an issue for fast launching as the shift from 1st to 2nd is slow. It's also a major risk on the circuit due to the way miatas like to go from 2nd to 5th on a traditional box. If I do that same miss-shift with my dog box it's game over for the engine as I'll be popping it into 1st without a clutch. To mitigate this I have added a much stronger centering spring on the shifter mechanism to make it hard to miss-shift into the far right gates. It's ok but I'm still worried about it.

So I thought up a design project. Why not make a mechanical shifter that converts a sequential shift motion to a H pattern shift. I have been going around in circles with lots of different designs and I ended up on something I think would work. It's fairly elegant with only a couple of moving parts, but it will be hard to get right and tweak.

https://cimg8.ibsrv.net/gimg/www.mia...b8b252dbc5.jpg

BUT then I watched a video of a Mini with an Evo engine and a Pneumatic sequential shift and I figured... why go for a mechanical solution when I could shoot straight for gold. May as well go all high tech and build a pneumatic solution. An hour of googling later and I've ordered the following on ebay:

4 x 16mm bore 25mm stroke dual action air cylinders
4 x 5 port 4 way pneumatic valves
20 elbow fittings and some 8mm tube

All up only $70USD

https://cimg2.ibsrv.net/gimg/www.mia...000e8b2e7c.png


https://cimg1.ibsrv.net/gimg/www.mia...842c14a179.png


The concept is fairly simple. I'll control the pneumatics with an Arduino that is basically switching the 4 valves in preset sequences up and down. 2 pneumatic cylinders will be paired sequentially in two sets (Up - Down / Left - Right) with each cylinder moving the shifter the equivalent of 1/2 a shift. In this way I can make the selector move down from 1st to N by retracting one of the two up/down cylinders, move across to the left two gates by opening both left/right cylinders then back up to 2nd by opening the up/down cylinder. All of this should happen in the blink of an eye... in theory. I'll also need an air compressor and tank but for now I'm going to bench test this concept and my workshop compressor.

I've never worked with pneumatics or Arduino so it's going to be a bit of a steep earning curve to get it all to work but nothing to tricky hopefully. I'll need a good amount of custom fab to build all the required linkages but I figure I can place it on the passenger floor and use cables for shifting to keep things easier for now. Initially I'll push and pull on the shift lever as its easier but will eventually move the mechanism to act directly on the shifter input bar. I don't want to drive the shift forks themselves as mechanical interlocking is very important if there is an issue with the electronics, but I could make my own interlock mechanism I suppose. First stage is a simple proof of concept.

Have any of you played with pneumatics controlled by ardiuno before? It would be useful to have someone to bounce ideas off.

I'm going to go out on a limb and say "no" holy crap this is an awesome project though.

Would you mind posting a link to that 4 port solenoid? That could be useful for a side project I'm working on.

One thought,

What failsafes would be incorporated?

With a mechanical shifter you would still get feedback from the transmission and you'd never ram in a gear when it wasn't ready for it, but with pneumatic you'd have to be very careful and only 'push the button' when the box was ready to avoid game over?

With a dog box you don't get that luxury. If you are changing gears without the clutch it will go into any gear especially if shifting hard and fast which you have to do to make to dogs engage cleanly. If shifting with the clutch you get the opportunity to feel the engine load and reengage it but then you are losing shift time.

The arduino controller could read the CAN output from the ecu and decode the rpm bits. It would also need the vehicle speed. Then you can do more interesting things like hold a downshift command until the target revs are acceptable or even auto change under braking (like preselect a low gear for an upcoming cirner and the controller holds the gears until the vehicle speed drops).

I think I might need to drive the clutch with an actuator in that case as I currently need to tap the clutch to unload the dogs to get it out of gear due to the high backcut on the dogs.

air cylinder Bore 16mm Stroke 25mm M5x0.8 Double Acting Mini Pneumatic Air Cylinder | eBay
Solenoid valve DC 24V Pneumatic 5-Way 2-Position Electric Air Solenoid Valve Brass 4V210-08 | eBay

I have no input other than this reminds me of this:
https://www.nengun.com/ikeya-formula/sequential-shifter

But that one requires you to clutch in every time you want to shift.

You might check out Formula SAE/Formula Student builds, there are a lot of cars out there that use pneumatic/electronic shifters with Arduino. My team did an electronic one, but I was not a part of that project

FSAE shifters are much simpler. Bike motors only need a single rotation to shift, so a lever arm and single pneumatic actuator. We didn't need any controls for ours, just a single solenoid, and single actuator. The driver would just hold the button until the shift completed. Most standalone ECU's have the capability of controlling a bike shifter as well. It basically just actuated the solenoid until the sensor indicates a completed shift, while retarding spark.

Madjak, are you planning on keeping the mechanical portion of your shifter, and just use air cylinders to actuate the single lever arm forward and back? Or are you using the cylinders to position the shifter throughout the H pattern? You should check out the Mastershift system just for some inspiration. They use push pull cables with electric actuators remote located in the trunk. Their controls are interesting that you tap the up-shift before you need it, and it puts the up-shift on queue until it detects the clutch. So basically hit the upshift at say 6000 rpm, then just stab the clutch at 7000 and the shift happens.

Cool project. Very ambitious but very cool. Who made your gearbox? I've never seen it before where 1st is where 5th normally is. I've seen other transmissions with "dogleg" gear shifts where 1st and reverse are near each other, but never where 1st and 5th are swapped while selecting 2-4 remain the same.

In FIRST Robotics Competition teams do a lot with Bimba pneumatic cylinders, CTR Pneumatic Control Module, National Instruments roboRIO and Labview, and all the other stuff that it takes to make a functioning pneumatic actuator system like compressor, accumulator, solenoids, fittings, tubing, etc. I don't know that it's automotive or motorsports grade stuff, but it's a start. Take a look at Andy Mark and Chief Delphi and US First Robotics Competition Forums and National Instruments FRC Forums. I'm sure there are many similar resources available in Australia as FRC is a global phenomenon. Like Megasquirt, be prepared to do a lot of reading and research before you start to get the hang of it. Good luck!

First step I was thinking of just actuation the shifter so I can see and feel if and how it engages. The linkages are the easiest to design and I can easily modify the stock shifter to attach the gear cables to. Movement in this case is simply left right and up and down so no need to make anything complex.

The second step will be to remove the entire shifter and turret casing from the gearbox (5 speed it just unbolts) and then make a lever based assembly that mounts to the same holes that moves the main shifter rod forwards and backwards and rotates it to select the various gates. This should make it a bit easier for the system to shift as there is less friction and slop.

You can get some awesome 3 position air cylinders that would be ideal as they simplify the design and you can customise the lengths of each stage when ordering. I don't know what bore or stroke I need yet so I'm going to prototype with the $5 ones one ebay then once I know the specs order the $200 multi position cylinders.

When I measured the 6 speed gates they are all slightly different spaceing so being able to order cylinders in 1mm increments becomes more important. I think the 5 speed box as more equal spacings which helps me with my el cheapo cylinders. With the ebay cylinders I can scale the leverage of the entire axis but not each stage independently. Pneumatics are on / off at a designed length, there is no fast way to stop them at 85% travel.

With the automation, I still want to drive the car and not make it fully automatic. I could program a target gear per corner based on GPS and let the gears change based on optimal power / revs but I think that takes too much away from driving.

For now it's just going to cycle through gears up and down. Spark cut, clutch control can come later.

I made a thread about this about a year ago or so. It's doable. My design I was going to use pneumatic actuators with linear potentiometers to measure position so when I tell it to shift, it can measure position of the shift in realtime so it never misses a shift for any reason. Some microcontroller to handle logic, arduino would be fine for this. Clutch done by an actuator as well. After thinking about it I swapped in an auto instead, then another, then a Ford auto. Funny enough the Ford auto has a sequential shifter, lol, and shifts in probably 20ms. It's stupid fast. If you got any arduino or control system questions let me know.

Thanks Pat... I've been following your thread closely. I think the area I'll have difficulties is in the arduino and electrical side of things. The mechanical linkages and fab won't be an issue. The Pneumatics themselves are actually fairly straight forward so other than choosing the appropriate valves and sizing the air cylinders it's just a matter of connecting them up correctly.

I was thinking of setting up some sort of contact sensor at each shifter location to signal a successful change, but I suppose the linear potentiometer is far more flexible. They aren't cheap though!

If you got electrical questions too, let me know. I have a fair amount of experience with car electronics and control systems. I'd love to see you build this and get it working! I have no regrets on putting a strong auto in my car, but I do wish I'd built the pneumatic system just to see how well it would work. So getting to see you build it and test it out would be very cool to see.

A mechanical switch could work too. I figured for me that linear pots would be easier to setup/calibrate, etc. Then you can datalog the motion as well to allow for adjusting pressure or dampening as needed. They cost more but I'd do them if they aren't too expensive. I don't remember what they actually cost, haven't used one in years.

I don't have the power band that you have, so I need the rev drop between gears to be as small as possible. Unfortunately that means I have lots of gear changes so saving time between shifts will help a lot.

Here are the specs on the box, although mine is dog engagement for 1st. Check out how close 2nd to 5th are.

The cheapest I can find are around $200 each. I think some cars came out with them so I'll see if I can find some second hand ones.

I did make a 3D printed lever arm for a rotary potentiometer for suspension ride height that I never ran. Maybe I could get a couple of them to work...
https://cimg5.ibsrv.net/gimg/www.mia...8a11bba497.jpg

Oh hey, I thought about this too, though it's not legal for the class I want to build for. Check out if you haven't already. I'd almost be tempted to use 3 pairs of cylinders and work the shift rods directly once you have the control algorithms worked out.

One thing you don't seem to know about is "switch ready" cylinders - there's a magnet embedded in the piston and you have a reed switch on the outside of the cylinder (or non-contact version) to tell you if the piston is at a given location.

I am tempted to directly drive the shift forks but if anything goes wrong it's likely that two gears could be engaged at once. I still want a mechanical interlock and the stock selector gives me that.

in terms of the magnet, does it work like a toggle so it on when the cylinder is at that end? Sounds ideal for checking to see if the change was successful.

Imo your best option is a mechanical solution to turn the h-pattern into sequential. And then use a simple actuator to shift up or down.

You could still leave the little interlock balls to prevent more than one gear being engaged, assuming you're using something still based off the 5-speed. And the cool thing is you can put the reed switches at either end or whatever... all it does is open or close a circuit based on if there's a magnet next to it inside the cylinder. It's much cheaper than linear pots.

I'm picturing something akin to what Subaru were playing with in WRC at one point - sequential gearbox, but mechanically it's H-pattern in case the shifting system goes tits up. Of course, the lever will be sorta flinging around so that might be a bad idea.

http://quaife.co.uk/sequential-gearboxes-honda-s2000-mazda-mx-5-nanb-nissan-silvia-sr20det-toyota-1jz-2jz-supra/

bit expensive though.

I'd combine it with something like our (BMW Motorrad) shift assist pro, which briefly cuts ignition during up on downshifts, auto-blips on downshifts etc. At least a brief ignition cut during upshifts would make for clutchless, faster acceleration

My H to sequential shifter I designed does that but I think I'd need multiple revisions to get it working. The tollerances are fairly tight and there is no space for flexibility meaning it would be remaking parts to change throw.

If the pneumatics don't work then I'll go back to that design. For now I think the paddle shift isn't actually that tough. I've just had a play with an online arduino simulator and I've managed to program the base code to shift up and down using some buttons for gear selelection, resistors firing off leds that are simulating the actuators.

https://circuits.io/circuits/4851337...bed#breadboard

I'm not sure that link will work but basically you should be able to start the sim then hit the two buttons to shift up and down gears. The 4 leds represent the actuators.

There is a lot more code to add but the basics are working which was the first step. Arduino coding is fairly straight forward luckily.

How awesome is Arduino! I picked up a Uno R3 board at lunch time yesterday with bread board, wire kit, leds and some switches and within 1 hour of unwrapping it had my code running doing the sequential shifts. I have a video of it switching the LEDs on and off but it's all a bit boring. I'll wait until I get my pneumatics firing before I upload a video.

My tracking is still saying 2 weeks from china for all the pneumatics. I'll pick up a spare 5 speed to start testing with as I now need to make the mechanical interface to the gear selector.

https://cimg8.ibsrv.net/gimg/www.mia...24d6101a02.jpg

What I'd be tempted to do is make a closed loop ignition cut setup too. Arduino is going to have spare digital inputs right? Have a pulse counter at the trans tailshaft, one reading engine speed, and a map of RPM per gear - by sequencing and a little bit of experience with how fast the pneumatics go you should be able to know when it's in neutral and spark cut the ECU until the engine RPM has dropped for the next gear. I mean hell, do something with more pulses per rev than ignition for the engine and watch the RPM delta to have it figure out when it found neutral.

I think I'm going to order a CAN bus shield so that I can decode RPM and TPS but I'm not sure it will be fast enough for closed loop. I've written a CAN script for the RaceCapturePro already so I know how to translate the data for each sensor already do I figure why not.

I'll know where the pneumatics are as I'm going to run two potentiometers on the shifter rod.... so I should know if the actuation is successful. Ive ordered a 16 segment display and driver ic so I'll use the potentiometers to drive the display rather than the pneumatic code. That way I'll be able to debug the mechanics.

I think the pneumatics are going to be fairly fast so I'm not sure how much rev matching will help. My engine changes revs fast but it will still be slower than the shift speed. I really just need shift cut so I can flat shift but still let the dogs disengage. I can do that by driving the antilag mode (2 step) directly within the Haltech ECU feed via an input from the arduino.

I had a thought today that I'll hook up a clutch switch so that I can only shift from N to 1st or R with the clutch in. That will give me a good safety for not hitting R accidently.

Well, my thinking anyway is that removing torque during the shift can't help but be a good thing. If you had a (fast enough) closed loop cut, even if it doesn't get down to the target RPM before the shift dogs are engaged, it would know that they were engaged because all of a sudden the RPMs would sync up to the new ratio.

I might be trying to come up with this code for an H-pattern box though.

I think it might be worth still actuating the clutch so it unloads the dogs. There is less weight then and the clutch acts like a buffer. I'm not sure if a shift cut by itself will be ok.

Mine is a H pattern box... just in an odd layout.

I have one sort of like that too - old Mandeville-Leeson box, though reverse/first are on the left and first is back to the left. They do weird things inside to get it to be like that. Anyway, this is getting a little offtopic to what you're doing, but my basic algorithm idea is to look at TPS, engine RPM, output shaft RPM, and a strain gauge on the shift lever, calculate current gear from the RPM ratio, spark cut to unload things when at WOT and the lever gets shoved harder than some force value (to distinguish from up and downshifts) then keep spark cut until the engine and output shaft RPM match up one way or another - either because it's in gear or because the shift took too long and it had time to drop enough RPM. Maybe 4 sparks out of 5 cut instead of all of them to keep from having too big of a bang when the fire comes back.

For my application, the engine is a carbureted (required by rules) rotary that will bolted to a small diameter multiplate clutch, and sequential boxes carry a weight penalty, so I'm not crazy in planning it the way I am. ;)

I've managed to get a few more parts of the project working.

First off I'm going to use two potentiometers to measure the movement of the shift selector. I've got one working as a test and it's accurate to around 0.5mm of movement on the lever arm which should be fine for measuring if the selector made the gear change. These potentiometers fee to the analog inputs on the arduino which will drive a 16 segment display. It will also feed back to the main logic loop whether or not the shift move was successful.

I have also mocked up the mechanical linkages on the box itself. They need to be made properly yet but they are ok for testing. I'll cnc them my design is done. There is a little bit of rotation genetated from the linkages in the up down direction but Im hoping its small enough to not be an issue.

lastly my pneumatics arrived. They are tiny little cylinders but pack a punch!

If you dont want to do an H to sequential conversion manually, which I think you should do and do it with a cam and drum design.

You should have 1 cylinder that goes in through the back of the trans case and pushes in and out directly on the shift rod/rail. And then have the second cylinder on a lever to rotate the shift rod/rail. Dont mimic the persons hand moving the shifter, mimic what the shifter does to the shift rod.

That is what I'm doing but with rod ends to lever arms.

this was my first layout but I've switched it around so that the up / down movement is directly onto the end of the shaft.

https://cimg7.ibsrv.net/gimg/www.mia...68a4bb7b57.jpg

This is very impressive!

Do you think you can get finite position control of those actuators with PWM of the solenoid? What pressure are you using to run these? My mind is wandering to other projects.

It's alive!



The shift speed is set low atm. I need to make another metal bracket up for the cylinders rather than using cable ties and then I can try upping the shift speed.

Gear selector mechanism is all mocked up. I've left some extra length on the selector arms for when I set shift length so I can adjust the ratio. Hopefully it all fits.

https://cimg7.ibsrv.net/gimg/www.mia...2b2453a304.jpg

https://cimg9.ibsrv.net/gimg/www.mia...64a0a55dcf.jpg

Cycling through the gears. Each arm moves around 45mm which is what the pneumatics offset.


Now I just need some shift push pull cables and I'll be ready to test.

Looking good. I can't wait to see this on the car, and to hear how it works.

I had a similar idea for my mechatronics project this year but ended up doing something different, I'm excited to see you get it working

It will be interesting that's for sure. I just need to hope the logic I'm building into the ardiuno is good because one miss shift and I'll be down an engine! Hardware wise I'm fairly happy with everything. All the mechanical linkages are fairly easy, I just need to make sure they are made so they don't wear or catch. I'm ordering some shift cables today so I'll be able to test it all on an actual gearbox soon. The software and logic side still needs lots of work.

I think I have upshifts sorted by using a shift cut input on my ECU, or alternatively just switching the ign relay off. The ardunio will receive a shift up input, cut ignition, fire the pneumatics, check for successful gear change via the poteniometers, then re-engage ign and display successful gear change on the led display... hopefully with a nice loud pop down the exhaust to signal a successful shift.

Down shifts are far harder to automate, so for now I'll still need to dab the clutch to unload the dogs, whilst blipping the throttle and hitting the gear down button. How that will all work I'm not so sure but I can really only test it on the track. I'm going to use a CAN shield on the arduino to receive RPM input from the ECU and do a calc on each downshift to make sure I'm not exceeding my rev limiter on the following gear. I think that's the biggest risk to my engine as there is no way to stop an over-rev once I tap that button. I think the pneumatics will be strong enough to rip it out of gear and ram it into another so if I tap that button at the wrong time it could be disastrous. I'm not sure what I'll do but it will probably abort the gear change and wait for another gear down input.

Last night I cranked down the shift delay between the 3 step (eg 2nd to 3rd) shifts. I can get them firing super fast, so fast it sounds like a single move but then there is no drag at all on the cylinders. Once I hook the box up it will most likely slow down a bit.

Don't interrupt power to the coils. If you have just started charging a coil to fire, and then you cut power, it will immediately fire the cylinder with some amount of advance more than it would have if it fully charged the coil to the amount specified in the dwell settings.. Better to tell the ecu to cut all future events, but not any that are currently happening.

I would write the software to measure RPM in 2 different ways, or write it such that it can detect an error if it happens, and if so disable all shifting or at least disable downshifting.

I would also have a clutch actuator so you can just drive it with paddles. I assume you're doing all this work to make the car faster/more consistent, and having the clutch also controlled would make the system work a lot better. A

About 6 or 7 years ago I built a similar system to what you're doing for a FSAE car. My system only needed 1 actuator since it was already a sequential trans. What worked best on our car was, if user says shift, cut spark, wait X time, then turn on actuator for Y time, then enable spark and turn off actuator. It gave the system a delay I could adjust to get the current gear out of a bind and the motor shut off, and a time to shift I could adjust separately. We ended up settling on 50ms total shift times. Although the actuators were faster, 50ms was bomb proof even at lower than normal CO2 pressures. It worked at 40ms, and at 30ms it worked most of the time but sometimes missed.

I wanna see you do the clutch too, because if I ever did what you're doing, I would 100% make the clutch automated as well.

Is it easier to reliably shift your trans from gear to gear when racing vs a syncro box? I assume it is.

With the clutch activation, I'll still need the manual clutch for launching / stopping. If I did it hydraulically I would need to set it up with a parrallel master driven by pneumatics feeding into a tee. But then if I hit the gear down button whilst my foot is on the clutch it will either damage the slave / pressure plate or push my foot back. I could have a clutch switch so that the clutch pneumatics don't fire if my foot is on the clutch.

I could also do it mechanically so that either the pneumatics or the clutch pedal pushes onto the master.

I think first stage is just replacing the manual gear selection. Second stage is optimising it for speed.

Currently shifts are very fast when I'm confident.... but due to the layout of the box with 1st in 5th position I'm a bit tentative going from 3rd to 4th especially whilst cornering. I know if I hit that 3rd gate by mistake its all bad.

Yeah clutch would be pnuematic, like the rest. Have the solenoids setup to connect each side of the cylinder to atmosphere when not firing so the pedal is still easy to use like before. I would just have the pnuematics operate the clutch pedal directly. Have a button where the deadpedal is that you mash with your foot to enable the pnuematic clutch control. Moment your foot comes off the deadpedal, it's back to regular clutch. That would be simple and fairly safe.

Muahahahaha! It works!

It's right on the limit of these little cylinders but I also had the air pressure down at 50PSI. I could crank it up too 100 and the shift speed would increase a lot. Still, it's pretty damn awesome!

​​​​​​

Nice work! This is awesome. Can't wait to see it in vehicle.

That is looking awesome! The only thing I will mention is that if you watch your video, those steel brackets that hold the cables are being way too much to be reliable. Those will fatigue fail bending that much with every shift. I would make those from thicker steel, especially the front one. Otherwise looking very solid!

Yeah the cable brackets need to be at least 1/8" steel. I'd even recommend gusseting them even at that thickness.

I like how you kept yourself a little emergency nub to shift with in case it shits the bed or you run out of air.

The brackets are temporary... I'm just using some bend ally to lay it out. The entire assembly will be flopped to the alternate side and I'm going to move it down towards the tail shaft another 50mm for final. I'll machine some nice billet brackets out of aluminum or weld some out of plate steel.

There isn't massive amounts of room in the tunnel so that little nub will be trimmed back.

I think I have a solution to handle auto blip. I'll basically make a servo driven large port idle control valve. It will be a port from after the intake filter that bipasses the throttle and feeds directly into the plenum. I'll make a rotary valve that will open up via the servo motor when I need to blip the revs. I'll have to have a search and see if I can find something that will work. I don't think a standard idle control valve will allow enough air through at 6000rpm to lift the revs.

what about a waste gate or bov or something like that? There is no pressure from the intake to drive it but I will have compressed air onboard.

Superchargers often have a small "bypass valve" that's effectively a 1" to 1.5" diameter throttle valve with a vacuum diaphragm on it to actuate it. You could likely modify one of these to act as a throttle blipper. A small electric push/pull solenoid to replace the diaphram would be pretty easy to construct and control such a device.

EDIT: Like this: http://www.ebay.com/itm/2002-2006-Mi...-/262985782122

why not just use the cruse control cam on the throttle body with a faster actuator than the cruse control one.

I could probably just use peumatics to open the throttle. Close it when I'm at target revs.

There us a haltech idle valve which would hook up via 4 dedicated outputs on the ecu. Its a stepper motor with 19mm ID ports so will be fast and flow a decent amount of air.

Last night I set up the gear indicator. Its a 38mm 16 segment display driven by a controller chip that only requires 2 pins on the arduino. I couldn't use the standard libraries so I'm manually turning on the individual leds based on there addresses which works well enough. I'm using a 16 segment display so that I can display 'N' and 'R'. I've also set it up to show 'P' when the pressure is low and 'E' if there is an error.

https://cimg4.ibsrv.net/gimg/www.mia...1bf57a5be9.jpg

The shield on top is a CAN bus shield with logging and GPS (if I solder to it). I'll see if I can get this receiving ecu data.

I'm another former FSAE guy who did a simple pneumatic shifter and I have to say that I really like the progress you've been making on this. I have had this idea in my head for a while now but I might start to look into it more with your success. :likecat:

I'm surprised how easy it has been so far. The part I was most worried about has been fairly simple so far. Its simply awesome how easy it is to add modules and fuction to arduinos.

Today I spent an hour setting up the CAN-BUS shield. The libraries supplied with the shield didn't have a 1mbit option so I had to add it. I then wired into my existing CAN loom in my car and first go it was reveiving data. So I added some filters for RPM, TPS and MAP from the Haltech and the data streams through super fast. I think I'm seeing around 50 packets per second so thats around 20ms delay from the ecu to the ardunio.

I'll use the rpm data to make sure the revs of the target gear are under a threshold to avoid money shifts. I'll also use the manifold pressure to data log whatever auto blip mechanism I use.

I've just ordered a 150psi pressure sensor for the air supply and also a 4L aluminium tank. I still need to order my final air cylinders which will be custom 3 position tandems so that I don't have complicated pivots and arms driving things.

Final layout on the shifter cables. The entire assembly is now mounted on my dog box and flopped to the other side to clear the ppf.

https://cimg8.ibsrv.net/gimg/www.mia...6055c262f0.jpg


I'll remake the base aluminium plate and the side to side bracket once I get the final pneumatic cylinders. The dog box does shift easier than the stock 5 speed so I'm wondering if I should order 20 or 25 mm bore cylinders. I'm currently testing with 16mm bore so a 25 will be over double the power.

The multi position cylinders cost a lot more but they will also package far better with the cylinders solidly mounted rather than moving like my piggy backed design. The are around 80€ so I want to get them right. They even have magnetic sensors that pick up the cylinder position which saves me from having to set up the potentiometers.

Just here to see how this pans out, so much fun tinkering DIY.

I've designed a simple paddle shift mechanism for my steering wheel. It will use two momentary limit switches on each side to give a nice solid feel and a bit of redundancy. I'll bridge the connections so that either switch will trigger to ground.

This design is the simplest I could come up with that requires the least amount of work for a good function and adjust-ability. It's basically a couple of aluminum blocks bolted either side of the mounting plate, with just the paddle itself needing a bit of machining. I'll bolt the handle to the main body of the paddle so that i can easily readjust if the position isn't working.

https://cimg4.ibsrv.net/gimg/www.mia...ee91555c0f.jpg

One made... the shift feel is brilliant! By using two switches the paddle has a nice solid click to it and a feel similar to a mechanical keyboard. I still need to make a moon shaped carbon fibre handle.

https://cimg3.ibsrv.net/gimg/www.mia...2f131fb893.jpg

You work faster on this all custom stuff. Thread is 100% DIY Win. :likecat:

Very cool project. Are you using any debounce code for the paddle shifters? I see you have two switches, is that so you can have them in parallel for redundancy, or in series to act as hardware debounce?

I'm really impressed, you are getting stuff DONE! And it's looking solid.

It's winter here in Australia and my entire family has been sick, so I haven't done much in the last few weeks other than coding and wiring. I'm also waiting for my 3 position air cylinders and 3 way valves to rock up before I can start to finalise things and test it in my car. I have ordered 5 x magnetic reed switches which I'll use to detect successful gear changes. These work on air cylinders with a ferrous strip around each piston so that the magnetic sensors pick up the movement of the piston. My current code relies on these sensors to display the gear as well as enable the next shift, so I can't really test it all until everything rocks up.

I have also ordered a air cylinder that I'll mount directly to a secondary clutch master slave. It's single action (return spring) 50mm bore air cylinder so it should be able to put out around 180lbs of force, which I think is enough to punch the twin plate clutch. That should rock up in a couple of weeks and then I'll be able to fire that off on down shifts.

Here is a 3D mockup of the various parts I'll be running. I already have the compressor (I'm nicking it from my 4x4) and I have just received my 3L (1Gal) ally air tank. The 3 position cylinders directly connect to the shift cables making for really nice packaging. My old system of piggy backed cylinders meant the pneumatic air tubes all jump around with each shift which isn't really ideal for long term use. I've calculated that the air tank will hold enough air for most of my events, but the compressor can keep topping it up as the pressure drops if I need.
https://cimg3.ibsrv.net/gimg/www.mia...5946414687.jpg

I had a bit of a play with my car's ECU and managed to get shift cut working in principle. So basically I earth an analog input and the ECU stops ignition so I should get a nice pop once the gear shift completes. I'll test this with the arduino next time I start up the car.

I also managed to figure out how to connect all 3 switches on my paddle shifter to my ecu with only two wires. This means I can use the inbuilt horn connections in my quick connect steering hub to transfer all my button presses to the shift computer. Not only that but I can also test for connectivity and short using the same process. So basically I have 3 different resistors hooked up in parrallel, plus a pull up and an inline resistor. This ends up giving me the following:

0v - no connection
~1.5v - Button 1
~2.5v - Button 2
~3.5v - Button 3
5v - Shorted connection

This is a really nice elegant way of connecting multiple buttons to a single analog input, plus I can add more if I want. I can even query multiple button presses, eg button 1 + button 2 to enable launch control...



Really the last component now is the auto blip. I'm thinking I'll use a generic 4-wire stepper idle control valve found on most modern cars which can connect to my ecu via dedicated inputs. I'm pretty sure they open and close quickly plus I'll be able to use it for idle control which is nice as it's hard to drive through the pits at low speeds as it is. I might even hook up a switch on my center console to idle up the engine when cruising in the pits.

Just a random thought: with the compressor and air tank in place, you could also put a bit of compressed air into the IM for it to blip.

Cool project btw, I am curious how it turns out :)