Paddle Shifting for H-pattern gearbox

Hey all.

I'd like to share with you guys the project I was working on for the last 2 years. It's fully pneumatic paddle shifter system for any cable driven gearbox. This is at the moment mounted in VW GOLF 2, FIA E1 hill climb car. The ECU that's controlling the valves can also control the clutch so one could use this on syncro gearbox also. Average shift I get is 0.08s which I'm looking to improve.

System in action can be seen in this video:

.. or my FB channel https://www.facebook.com/mmemotorsport

Ideas & suggestions are more than welcome. I've dedicated big portion of my life to these paddle shifters and would love to optimize the system. I must say that the system on the video has not been optimized yes, it's overdimensioned because it's a prototype. I didn't want to deal with the details just yet, was too much going on :)

I'm running on 7 bar air pressure and with 7 bar I get average shifting time (up/down) of 0.08 seconds. That's 12.5 shifts in 1 second!
I believe I can get it to ~ 60ms (with proper gearbox)

Features I have at the moment (and still adding):
- Autoblip on down shift with sophisticated closed-loop algorithms for smooth shift
- Queued down shift (I can pre-select any gear at full throttle on a straight line (say in 6th gear) and when I start to brake, gearbox will automatically down-shift down to my pre-selected gear at perfect engine rpm. The harder you brake, the faster it will down-shift)
- Integrated G sensor for perfect timing of queued down shift
- Launch control - as long as I'm at full throttle, the gearbox will shift automatically at perfect rpm so I get perfect starts every time
- Two way clutch actuator (I can use this on a syncro gearbox with hydraulic clutch). I control the clutch with two different speeds. One has quick clutch release (for upshift) and one with slow release (for downshift).
- Auto retry on mis-shift
- 5 hour of datalogging so I can fine-tune this to perfection
- CAN BUS support with any external ecu or dashboard

p.s.: sorry about the video, i know it's not the best and my "shooting" road was really short.

More images:

Nice. I have had the opportunity to tune some awfully fast shifting systems, so there are gains to be made. Keep it up. Watching with interest.

Nice!

Doc Brown will arrive in 3...2...1...

I got distracted by the music of the straight-cut gear whine. Very nice system!

Nice. How much weight does the entire system weigh and how many shifts do you get before you run out of air pressure?

Hey. Thank you for your kind words. It means a lot :)

T.J. wrote: Nice. How much weight does the entire system weigh and how many shifts do you get before you run out of air pressure?

System weights around 9kg. I actually never run out of air because there's an air compressor that fills the pressure when running low.

Marko

The_Jed wrote: Nice! Doc Brown will arrive in 3...2...1...

Stuff like this makes me drool...

I love it.

This is awesome!!

Very Very cool!

Very cool. Pulled apart any gearboxes to see how the gears and synchros are reacting to the clutchless shifting? I've done clutchless shifting myself, but typically MUCH slower.

In reply to DILYSI Dave:

If I'm reading the above correctly, he's using the clutch for up and down, using a air cylinder to depress it.

Cool system.

My project is DIY ECU http://www.rusefi.com and I kind of always wanted to make a robot transmission!

What is the license for your software/hardware? Is there a build thread/forum/site anywhere with any specifics/drafts, like what actuators are you using, what is the cost of all the air components etc?

Good job!

Fantastic work! You've accomplished the hardest part -- making it work well. It's only going to get smaller, lighter, and faster from here.

Do you plan to commercialize it and sell as a kit, or release it as open source?

I'm curious as to why the mechanism is acting on the shifter cables in the cockpit and not the shift levers on the gearbox.

Hey guys. Thanks again for all the comments. I will try to answer individually :)

series8217 wrote: Fantastic work! You've accomplished the hardest part -- making it work well. It's only going to get smaller, lighter, and faster from here. Do you plan to commercialize it and sell as a kit, or release it as open source?

It's posts like this that make my day. It just makes everything worth it. Thanks. I'm not really sure what I will do with the system. I'm currently designing a case so I can cover this in nice enclosure, from there I don't know. We will see.

DILYSI Dave wrote: Very cool. Pulled apart any gearboxes to see how the gears and synchros are reacting to the clutchless shifting? I've done clutchless shifting myself, but typically MUCH slower.

The car on the video has a dog box (non-synchronized gearbox) so it can be shifted without the clutch with proper engine management. However, system supports full clutch operation, but it is not as fast as this, because synchros (even with clutch pressed) don't allow these speeds. I believe 150ms on a syncro gearbox is achievable. In this mode, car is still shifting with clutch, the only difference is that driver doesn't :)

MCarp22 wrote: I'm curious as to why the mechanism is acting on the shifter cables in the cockpit and not the shift levers on the gearbox.

When I started this idea few years back, this was my intention. But, it's near impossible to fit this in any car I've seen. At least with the dimensions I have at the moment. I'll definitely be looking in this direction as this would be TOP if it could be done.

russian wrote: My project is DIY ECU http://www.rusefi.com and I kind of always wanted to make a robot transmission! What is the license for your software/hardware? Is there a build thread/forum/site anywhere with any specifics/drafts, like what actuators are you using, what is the cost of all the air components etc?

Nice work man. I know how much effort was needed to get there. Thumbs up There's no build thread at the moment. I'm not really sure what I will do with the system when it's finished. Costs.. I'd rather not talk about costs, because they're very unrealistic at the moment :) It took me more than 2000 hours on the hardware/software/testing alone and atleast 10k+ for all the parts I bought and later realized they were no good :( But this is part of development.. I guess.

In reply to cmcraeslo:

The time and effort (not to mention money) you've put into this project certainly show.

Saw this on my FB feed this am. Cool project and good to see you here.

While im not the intended market for this i watched the video in facebook and was impressed. Technology scares me but some of its impressive!

Thanks guys. I will keep you updated with the latest status of the project :)

Marko

Just finished the re-design of valves setup and enclosure, to look it somewhat better & more compact. Here's the preview:

What do you guys think?

Now I actually like the application of this. A automated manual sounds pretty neat. Where as you still have to actually shift although with a paddle. This I could see really wanting on a race car but would even use it on the street more than a DSG from any manufacture out there. Don't care what anyone say's they suck in auto mode. I'm guessing there is still a long way to go to have a viable cost effective version for the masses?

Nice! I did an air over hydraulic set up to move the clutch pedal to a proportional switch on the gear lever (but could be wheel mounted). I still wanted to row so I didn't follow up on the paddle shift element. Have you looked at the AP racing (shiftec) fly-by-wire clutch system? They've got the packaging pretty small. Pricey though - $15,000 per unit with a 10 unit minimum - think spec series or F2

Hey.

Yeah, that shiftec unit has killer price. Cant imagine someone buying this, other than the F1 teams.

Can you explain your system a little more? I'm interested in seeing your solution to the clutch problem and different speeds.

Thanks