Cheap, simple, and cool: paddle-shift Subie

[fbh]

Sup folks,

 

I'm having a lot of really crazy ideas lately

 

The latest one is "robotizing" the gearbox on my 89 GL (EA82) using heavy duty servos or pneumatic drive, and a PIC16F84A microcontroller.

 

Basically, connecting some "servos" (let's call them that for now) up to the shifter, or its linkage along some point, the servos being controlled by said microcontroller, linked to two microswitched paddles mounted on the steering column. None of this will be too hard to make - I've done gearlever-related microcontroller exercises before, including building an H-style gearlever for PC racing games using the same microcontroller.

 

Basically, the idea for now is just to make this system control the gearlever - I don't know where to find actuators strong enough to manipulate the clutch - so it goes sort of like, depress clutch, pull paddle, let go of clutch. Maybe later when I find some kind of actuator I'll be able to do an automatic clutch as well - hook up a torque/rev sensor and make it automatically do the clutch for takeoffs.

 

I'm budgeting on less than NZ$500 for this system. If I can get it going, I'll put some plans up for those that are interested - some tips will be very much appreciated!

[75subie]

or you can buy an automatic sube for $500:rolleyes:

[mikeshoup]

A manual gearbox is kinda one of those things that needs to be done by feel... If it can be done though, still sweet.

[NorthWet]

Others can question the "why?" ("because you can" is good enough for me ).

 

My thought on this goes back to my teen years, when there were issues with shifter and linkage design, especially in remote-trannied (rear or mid engine) vehicles. The shifter often had little or no "feel" to it, making quick and accurate shifting difficult.

 

Your idea would completely remove "feel" from the system. Unless the driver is mechanically insensitive, the shifting will likely be too slow and/or too noisy.

[Phizinza]

... And what is wrong with a H pattern shifter? John Bowe (race car driver here is Aus) said that he'd much prefer a H in his 360 Modena then it's silly computer controlled shifter...

 

As for the Auto idea.... Autos won't shift when you want them too dude, at least this would. Driving a auto is like driving a go kart, you may as well not have a gearbox. It's so boring.. (time to get flammed by auto owners across the world)

[fbh]

True - I had a thought of using a tunable two-way ratchet type mechanism to move the gearlever, so it can be calibrated to move the right amounts to get in gear. As for the clutch, switching from neutral to 1st would signal a takeoff in this scenario (or a seperate clutch paddle...?), when the controller would take the RPM's at which you chose to "release" the clutch (or take off), and choose a specific engine speed, at which point it starts releasing the clutch if the RPMs drop too much below that engine speed - could even use fuzzy logic.

 

The whole idea behind this is to combine the best of both worlds (auto/manual) - the comfort of being able to let the gearbox automatically handle shifts (auto) while still having efficiency, and total control if/when you want it (manual).

 

Getting pretty excited... just got to scrape enough money together to get the parts!

[Syonyk]

I've got a PDF, somewhere, of someone who did something like this as a thesis project. I'll see if I can find a link to it or get it posted again.

 

-=Russ=-

[fbh]

... And what is wrong with a H pattern shifter? John Bowe (race car driver here is Aus) said that he'd much prefer a H in his 360 Modena then it's silly computer controlled shifter...

 

As fo the Auto idea.... Autos won't shift when you want them too dude, at least this would. Driving a auto is like driving a go kart, you may as well not have a gearbox. It's so boring..

 

True

 

The idea isn't to completely replace the shifter though - it's only to add an automatic control to it - as soon as you move the gearlever the computer will be able to detect this (by means of contacts on the cables moving the shifter) and stop pulsing the servos, so they can turn freely.

 

I thought of doing this as I was driving from uni in the GL, in the stop-and-go traffic flicking through gears, wishing at that time to have an auto, but as soon as the road opened up I was thankful for having a manual again.. so this idea popped up

[TheSubaruJunkie]

Still wouldnt be automatic if you had a clutch peddle. And I doubt a servo could shift as fast as a human arm with 24" of leverage.

 

Sounds good on paper though.

 

-Brian

[NorthWet]

... As for the clutch, switching from neutral to 1st would signal a takeoff in this scenario (or a seperate clutch paddle...?)...

Porsche's "SportMatic" used a microswitch on the shift selector to signal clutch usage. (Their design used a clutch and torque converter in series on a manual gearbox.)

[bgd73]

I have daydreamed of it at a minimum much shorter throw than oem- (that has been in several posts already) It seems like my 2wd ea82 finds a nice range of power in one gear and I'm takin forever like a tractor trailer going to next gear. I don't have a turbo lag on my n/a and still don't like losing the "head of steam". I experimented by shifting as fast as I could - it really does hang onto some noticable momentum of previous gear and rpm achieved.In fact, I thought my little sedan was fast. The exhaust almost lets out a snap of stubborn and the gains keep on going.Dare I say exotic car sounding ... if compression were any higher... wow.

[75subie]

As for the Auto idea.... Autos won't shift when you want them too dude, at least this would. Driving a auto is like driving a go kart, you may as well not have a gearbox. It's so boring.. (time to get flammed by auto owners across the world)

 

 

uhh, yea you can dude thats why autos have a lever that says drive (boring) or 1,2 (selectable) you can also kick them down by flooring it.

 

for this idea, reverse would be a problem cause they don`t always go in. shifting while making a turn in 4wd would also be a problem. down-shifting would also be a problem.

 

good luck with it, if it comes out, more power to ya;) i`ll stick with shifting/clutching by hand.

[DaveT]

Since Formula 1 race cars use this kind of (indirect) control for shifting & clutching, it can't be all bad...

 

This link talks more about the controls, but mentions how and why they went to electronically controlled shifting:

 

http://www.f1technical.net/articles/30

[fbh]

Still wouldnt be automatic if you had a clutch peddle. And I doubt a servo could shift as fast as a human arm with 24" of leverage.

 

Sounds good on paper though.

 

-Brian

 

You'll still have the clutch pedal and the gearlever, they'll just move by themselves ;-)

 

But you're right, electronic servos (most of them anyway) don't shift very fast at all. Allright for model planes, and maybe generic driving, but not spirited driving

 

that's why i'm either trying to look for electronic servos with high amounts of torque that can shift very fast (200ms for a 60-degree rotation with 10kg/cm of torque is the best I found so far.....), or go for something else, like pneumatics.

 

It'll suck quite badly having your arm broken by the gearlever though

[mikeshoup]

Was talking with a buddy, and it turns out one of the mechanical engineers a few years back used a hydraulic setup with paddles to shift in his honda for a senior project. Perhaps I'll go through the archives at the library.

[GeneralDisorder]

The problems I forsee are both the feel, and the interface.

 

The feel part others have covered.

 

The interface is going to be difficult - how do I select 1st gear from 5th? Having to smack the down paddle 4 times would suck, and having to operate the clutch by foot? No I think if you are going to do it you will at least need to eliminate the clutch completely. Retrofit a hydraulic clutch on there, and servo that too. I've seen some toyota's with small hydraulic actuators that should be easy to retrofit. And instead of paddles, go with a seperate button for each gear - one for each finger on an ergonomic grip somewhere....

 

To get the kind of speed you need, you will probably have to go hydraulic with a TON of psi. Probably well over 1000 psi to get the system to move fast enough.

 

GD

[LPGsuperchargedBrumby]

 

 

To get the kind of speed you need, you will probably have to go hydraulic with a TON of psi. Probably well over 1000 psi to get the system to move fast enough.

 

GD

 

use over size fittings and hoses... the less restiction you can put in the lines and system the faster your actuators/rams will move

[fbh]

True, true...

 

I've been wondering though, what about pneumatics, possibly powered off the exhaust?

 

As for the interface, good point... some more thinking to do! maybe the paddles, but with a seperate 1-5 selection? alternatively just manually move the gear lever

 

for a first concept i'd like to automate the movement of the gear lever, maybe later do the clutch...

[Phizinza]

uhh, yea you can dude thats why autos have a lever that says drive (boring) or 1,2 (selectable) you can also kick them down by flooring it.

Of cause... but this also adds a fair bit more wear on your gearbox (I have heard, and also seen in my sisters auto Audi.) If done a lot it can shorten the life of the box a lot.

And I don't know about you but every 80's auto I drive seems to change up and down at the wrong times and most seem like there on their way out.

 

Since Formula 1 race cars use this kind of (indirect) control for shifting & clutching, it can't be all bad...

This is a good point, but you have to also remember when you have a paper thin clutch you don't want some human controlling it. Also a few mil$ goes a long way in developing things.

 

 

Over all I'm not saying its a bad idea. I just don't get the point in it. It doesn't use much human energy to shift with a H pattern over just flicking a switch and from my expirance with old subie gearboxes you have to "persuade" them into gear. A mechanical device is only going to shorten the life of your box. Although it would be cool to have paddle shifters, I just think its about them same coolness as having 8000watts of subwoofer in the back of your car (pointless.)

But by all means.. Give it a go!

[bgd73]

Hire someone to shift real fast to the light on the dashboard that say "shift" from the back seat, -- as the map man in the front seat is yelling coordinates.The shifter guy could even change chanells on the radio and make coffees for the driver!

 

I took a gander underneath closely... it would be quite a presicion to get that stuff correct. An arm and hand feels to the bouncing engine, all terrains, tranny warm and cold. I see those gear paddles on bouncing rally cars and assuming "genius".

[Phizinza]

I took a gander underneath closely... it would be quite a presicion to get that stuff correct. An arm and hand feels to the bouncing engine, all terrains, tranny warm and cold. I see those gear paddles on bouncing rally cars and assuming "genius".

Another thing.. On race cars (like WRC cars) you don't need to care about a gearbox lasting 100,000+ miles. As they can now change them in 17mins and would run a different one at each event. So smashing the gears together wouldn't really matter too much.

[Syonyk]

Actually, Subaru found that on their rally cars, the cost of the electronic gearbox and associated wizardry was more than covered by the extended life of the gearbox.

 

The computer won't miss a shift like a human can. All it takes to require a transmission replacement is a missed shift or two at the power levels they're running in WRC, and the computer controlled stuff prevented that.

 

Also, for the "up/down" buttons... having to go sequentially through gears isn't as big of a problem as most people seem to think it is. I regularly drive a sequential shift vehicle, and it doesn't bother me at all (motorcycle). Coming in from 5th to a stoplight, I clutch, and tap the gears down as I slow down.

 

As a member of USMB, I think this is kind of a pointless project. It's expensive (I've got money that you won't do it for under $1500), fixes something that works just fine, and will probably make operation rougher.

 

As an engineer, I think this is an awesome project, and if I can help in any way, feel free to PM me - I've got a Cpr E degree with experience programming microcontrollers, and this sounds like a blast (similar to the solar powered air conditioner I was talking with another friend about).

 

-=Russ=-

[hooziewhatsit]

I think to do it right, you'd also have to monitor both the engine RPM, as well as the driveshaft/front axle output rotation.

 

Having those, you could control the engine RPM to perfectly match input/output rotation, and make all your shifts silky smooth. No synchros FTW!

 

A friend who worked on diesel trucks was telling me about a automatic transmission they've come out with that does just that. He said the transmission was incredibly smooth compared to a normal human driver.

 

-Dave

[NorthWet]

I think to do it right, you'd also have to monitor both the engine RPM, as well as the driveshaft/front axle output rotation.

Rather than engine RPM, you really need input shaft RPM (and, preferably, pinion shaft RPM). You would also want throttle control. Given these you could make silky smooth shifts... that would be slower than an average human would do on their own.

[hooziewhatsit]

Rather than engine RPM, you really need input shaft RPM (and, preferably, pinion shaft RPM). You would also want throttle control. Given these you could make silky smooth shifts... that would be slower than an average human would do on their own.

 

d'oh, that's what I meant. know & control the input and output shaft rotation speeds, and you'll know exactly when to shift.

 

However the benefit of such a system....