Idea

[Carbix]

Coming from a noob <---me: Well i was looking at the posts on of a dogbox or crashbox for shifting. I was thinking. Could an auto tranny be made of a dogbox... what would have to happen for it to work.

Now lets just keep an open mind about it. Lets not think that it cant be done, but how could it be done.

Some of the ideas i had was setting up a pc to control the rpm of the car to match the spin of the gears.

Next im thinking to make it more like a sport auto. something like steering wheel shifter pads. So u still pick when to shift. But it shifts for u...

If it could work what would be the perks? Im thinking this would work vary well for down shifting… could shifting speeds be lowerd?

well im just a noob with a funny idea i want to know what u guys think and how it could be done.

[Subtle]

I've been running a built 4eat for the last year. It also has the stall converter, so it quick off the line.

Also have a manual shifter for it that does push-button shifts, which isn't installed.

But I imagine that the shifts would be instantanious--particularly as on the shifts it goes to full line pressure.

One of the features of the 4eat is that upon acceleration in either first or second the tranny computer sends more torque to the rear. Viscous limited slip center diffs wait for the front wheels to spin before directing torque to the rear.

Haven't been able to find out what proportion the TCU sends to the rear

[greg donovan]

it has worked.

jonathan ryther, the guy behind knight racing, had a pair of open class in imprezas that were manually shifted automatics.

when they worked they were awesome. when they didnt...


...well they didnt.

[THAWA]

I think you guys are talking about different things.

Carbix was describing using a non-synchro manual transmission, and shifting it without using a standard lever. Pretty much a sequential transmission.

You guys are talking about using an automatic transmission, and shifting it with using a regular auto shifter, or a paddle/lever. Like a tiptronic.

Carbix, you need to figure out some way to move the shift forks without doing it by hand. I don't know that you could do this cheaply, or even with the standard transmission case. Take a look at this: http://auto.howstuffworks.com/sequential-gearbox.htm

[Carbix]

I was more thinking that the movment of the forks being by hand or by paddle would not matter. but that when u shifted u would not have to move clutch... or that there would be no clutch and synchro.

Make it so that AFTER 1st gear u would not have to push the clutch to shift or there would be vary lil pull away from the flywheel with the cluch. No synchros, the idea is to have the RPM automaticly right where it needs to be via computer controling the RPM. Being that much of the fork and cluch movement would have to be controled by computer a tiptronic type paddle system, i feel, would work best.

I guess one could mod a tiptronic system to work in this manner. Any one have a mini cooper they would loan out lol.

[THAWA]

It's generally accepted that tiptronic means an automatic system with paddles, or buttons to control shifting. If what you're asking is to shift a manual without using the standard H pattern this would be considered a sequential transmission. How it's done depends on the internals, but a non-synchro setup sounds the best.

You could just keep track of the vehicle speed and gear position to get a semi accurate RPM of what the engine needs to be at for a downshift. Or you could install a speed sensor on the driven shaft or something to get a more accurate reading.

Here's a link with a bunch of info on motorcycle shifters to give you an idea of what you're in for:
http://www.dansmc.com/gearshifters.htm
http://www.dansmc.com/gearshifter2.htm

[Carbix]

Cool. So the system would be a tiptronic system. Your right on the money with the installing a speed sensor on the driven shaft. the idea is to have the tranny work like an auto but have no need for a clutch, synchros and torque converter, at least a clutch only when starting to move in first. A computer would have to control both movements of the tranny and the rpm of the engine when the shifting is taking place. It could go as far as to make the computer calculate the exact ratios of the teeth in real time so they meat up seamlessly. A computer can track the teeth to where they are where they are going to be and when they will all line up. Think of it like making a perfect shift every time with a dogbox or crashbox assisted by a computer. So right on the money that u would not even have to push on the clutch to shift down or up because the computer would place it in gear right at the perfect time.

I think the best way to control the movements of the fork is with a hydraulic system. This way the forks could be almost fired into place within a split seconded.

Your shift time would almost be nothing. Plus mix this with a really loud BOV and WOW u have one sexy sounding shift.

"!!!!!BUSHHHHH STICKA!!!!!!!"

[THAWA]

I'm confused now. If it's a manual tranny based transmission it's going to be a sequential. If it's an automatic tranny based transmission it's going be tiptronic. Which does it start life as? If Auto it can be done much easier.

[greg donovan]

what it sounds like he is describing is the transmission from a subaru world rally car.

that thing costs a gagillion dollars and is shifted w/out a clutch by paddles and hydrolics.

[Carbix]

yep i think F1 cars use it too. But they all still use a clutch. with F1 cars u need to replace the hole dame thing after each race. but dont quote me on that

Start life would be a manual. controled by a hydrolic actuator, controled by a computer (controling RPM), controled by padles that u push.

Im more thinking of a way to make a cheap one. modding in the system from a bike..... as posted above could make this work... I guess the hard part of this is making a tranny that can shift fast enough that a computer could control. something with a linear action.

[THAWA]

Yeah, what you need to do is replace the shifter rod inside the transmission with a grooved cam or a plate that will select between the gears. Then you need a motor or actuator to control the moving of the cam. Then monitor the rpms to make sure you won't downshift past redline, and you're pretty much set. Just machining the parts is going to be the cost part of it.

Of course if you really wanted to get ghetto about it, just install three acuators with three positions, and hav the computer control which position each actuator would be in for each gear change. But you wouldn't like it if an actuator got stuck and you have two gears engaged at the same time.

[Subtle]

A couple of years ago I talked to someone at Knight and they were using the 4eat and had it set up "like a sequential shifter".

The cost for the build was about 25 large.

On the launch the driver held it in second--built some revs and boost-- then at the actual start dropped into first and hammered it.

I'll see if I can find the article.

[Carbix]

cool now we talking
see all this i can do and so far i think for cheap
my budy has a steal shop that i can get just about anything made for the price of the steal needed.

The big part now that i know what im going to do with the tranny is making a program that can control about 12000 calculation a min. Now that it think about it. Q basic should do just fine.

From this looks of this i could build this into a mod system.

If this was done right would the shifting be smooth or chopy?
I think smooth? Cus we are talking about the RPM being right on the money for shifting.

[greg donovan]

A couple of years ago I talked to someone at Knight and they were using the 4eat and had it set up "like a sequential shifter".

The cost for the build was about 25 large.

On the launch the driver held it in second--built some revs and boost-- then at the actual start dropped into first and hammered it.

I'll see if I can find the article.

they were modified by Rancho Transmissions. forgot to mention that in my earlier post. heat was a big issue in them. they added huge transmission coolers in the trunk with vents in the quarter panels.

they also chose to turbo charge ej25s as well.

steve gingras really liked driving it on the start. he was prety fast in it but the car kept breaking.

ryther also tried using one piece driveshafts as well. they worked fine untill you got to the high speed straights at wild west. then they vibrated apart.

[BAC5.2]

Or you could install a speed sensor on the driven shaft or something to get a more accurate reading.

That's where the speed sensor is. It's driven off of the output shaft (specifically located on the 3-4 fixed cluster).

It sounds like what you want is a manual transmission that has a computer control engine RPM when you yank on the shift lever.

Having paddle shifters and having a hand shifter are very different things. A hand shifter requires you to move through the gears like you are using the clutch. With a dogbox, you don't have to use the clutch. I can shift up through 4th gear, and back down to first without using the clutch, and I can do it at almost any RPM.

For paddles, you'd need to have something move the shift rails for you. Hard to do in the small space in the transmission. There is actually almost zero space.

You also have to worry about HOW you have the gears shift. You have to have it so only ONE rail will move at a time. Otherwise, you'll be locked out of gear. Not fun. A cam wouldn't do it, because you still have 3 rails in the case that sit parallel to eachother, plus the amount of time it will take to rotate the cam to engage the gear. When you shift a dogbox clutchless, it is so amazingly fast, you don't even realize. Ask Rallysam.

It's likely that any automatic modification to a manual transmission will result in a slower shift than a dogbox is capable of providing. And an automatic transmission manually shifted would be pretty easy. A dogbox, shifted clutchlessly, is almost as fast as a built automatic. All you have to do in a dogbox is move the shift collar less than 1 inch.

The best automatic available on the market today belongs to the Audi TT and the VW GTI (and probably a few other of the VW/Porsche/Audi family) in the form of the DSG trans. No torque converter, which is awesome, and under full gallop, it's about .2 seconds between shifts.

I have been meaning to work on an automatically shifted Subaru dogbox, but I haven't had the time.

I think Subaru's WRC boxes are nitrogen shifted, and cost to the tune of $30k-usd from Prodrive.

[Subtle]

[BAC5.2]

But you wouldn't like it if an actuator got stuck and you have two gears engaged at the same time.

The trans won't let you select two gears at once. The factory interlocker locks out the other rods when you are in gear.

In a stock transmission, the synchros preload the hubs. So when you shift from 2nd to 3rd, this preload prevents the 1st gear rail from shifting into first before you get the selector into 3rd gear. The factory interlocker is designed to work WITH this preload.

It does the job pretty well, and we have only seen 1 or 2 synchro gearsets shift into first before the selector got into 3rd.

With dogboxes, this preload doesn't exist, and you CAN shift back into 1st before you can shift into 3rd. It'll do this without any problem too, so it's scary. That's the purpose of the Interlocker that PPG makes. We won't sell a dogbox without an interlocker.

I was at 4700RPM in 3rd gear, and was going to clutchless shift into 4th. I had my wrist tweaked kinda funny, and so when I yanked back on the shifter, I slammed it into 2nd gear and not 4th. No grinding, just an immediate ~7000 RPM's on the tach and the brake light flashing. It's easy to shift a dogbox, so you really gotta be careful. I would have any automatically shifted dogbox attached to a motor you don't care about thrashing while you work out the kinks.

[THAWA]

Or you could install a speed sensor on the driven shaft or something to get a more accurate reading.

That's where the speed sensor is. It's driven off of the output shaft (specifically located on the 3-4 fixed cluster).

I was thinking more along the lines of a hall effect sensor, not a worm gear setup.

You also have to worry about HOW you have the gears shift. You have to have it so only ONE rail will move at a time. Otherwise, you'll be locked out of gear. Not fun. A cam wouldn't do it, because you still have 3 rails in the case that sit parallel to eachother, plus the amount of time it will take to rotate the cam to engage the gear.

Why wouldn't a cam work? If it have it sitting perpendicular to the shift rod I'm sure you could make a pattern that would allow shifting through all gears. Externally mounted would be the easiest I think, because as you mentioned, there's no space inside the tranny.

But you wouldn't like it if an actuator got stuck and you have two gears engaged at the same time.

The trans won't let you select two gears at once. The factory interlocker locks out the other rods when you are in gear.

Right, but if the actuator is strong enough it could break the detents/ball/spring and shift it into another gear while the other is still engaged. Not likely, but neither is the actuator failing anytime soon if it's setup properly.

[THAWA]

Or you could make three new shift rods with a setup better made for a cam or someething.

[Carbix]

Neto....... i love this. im taking it all down on paper.

I like the cam idea. To turn a cam using a hydraulic system would be EZ. and for the area of time it would take to shift. a spin would be much fast. i would have to think. I have a few thoughts on how to make it spin. one u would make it so it locked into place after each turn. When released for up or down. the cam would become unlocked. now at one end of the cam u could have a geared end. With a hydraulically driven sprocket, that could be controlled "hydraulically" to spin one way or the other from valves opening and closing, under high presser. Now back to the locking after each spin. it would automatically stop after one gear shift, via the locker that keeps it in place. say this locker as soon as it is first released will slide along the circumference of the cam until it locks its self again into a grove, notch, clip what ever.


_--___--___--____!```! <- thats the cam lol
-----_-----_----------|, , , ,|

what i want to know. Where does the heat come in... I would to build this for every day driving hahaha

Tell u guys what. I have a 95 sunfire that needs a new tranny.( my revers is fugged) So how about u all help me build this thing "idea" over time. well as long as the tranny i have in holds out.

So what do i need. lol
what the best kind of tranny to start with. being that we will be moding one and making one from scratch is too much.
Lets keep in mind its for every day driving. So i think things like really count. But im still hard on my cars. At a later date i want to turn my 93 SS into a manual (its an auto right now).

So lets get idea, draw them out, add them all up, and build some plans. Even if we fail we will succeed.

For the hell of it lets not make this thing cost a lot. See how cheap we can build one. I dont mean cheap as in crap. Mean cheap as in used parts and finding stuff from wreckers that would work.

[BAC5.2]

This doesn't seem like a daily-driveable idea to me. The violence that a dogbox shifts with would be too much for a stock motor, and the time it would take to control timing and fuel would be nearly impossible to correctly calculate.

If a dogbox can be manually shifted in .15 seconds, that is less than 18 revoloutions of the crank at 7000 RPM. What you are asking for is a 2000 RPM drop in engine speed (accomplished by retarding timing and fuel) in 36 ignition events. That's tough to do.

You could do it more slowly, but then it defeats the purpose of having the box shift automatically.

Another minor inconvenience, you still need to use the clutch to get the car moving, and you can't automatically shift into 5th gear or reverse. You COULD do 5th gear, if you get a dogbox that replaces all 5 gears (which do exist, and they are quite expensive).

Finally, to use a single cam, you'll run into the problem of the lobes.

It would require 520 degrees of rotation to shift from first to second and back to neutral. How are you going to grind a cam to allow the 1-2 lobe to rotate 520 degrees, THEN allow the 2-3 lobe to travel IT's 520 degrees? You could use 2 cams, but then cam timing comes into the equation. When you use hydraullics, you run into the problems of line pressure and failure. If you use a pin or lobe-switching system that mechanically switches the cam lobe, then you run into exactly the problem you have with manually shifting the dogbox.

The best cam solution I could figure, would be to have a hollow cam shaft, with another shaft running through it, and you'd have to turn each cam lobe independently of it. Then you'd have to hope you time it correctly.

All in all, to convert a manual transmission into an automatically shifted trans, you'd have to do a lot of modification.

My idea involved solenoids, but space is a huge concern. It would be easier to build a custom rear section. Then you could set something up quite easily, but until then, there is so little room, it'd be tough to fit it all together, let alone make it operate correctly.

[BAC5.2]

One last thing I forgot about...

It's impossibly hard to pull a good dogbox out of gear while on throttle.

In order to upshift smoothly, you MUST double clutch.

Since you don't really want to subject your engine to instant RPM changes on every single shift, you would need the computer to cut fuel and retard timing so the cam could pull the trans out of gear, then it would have to add fuel and timing to increase engine RPM, then it would have to cut fuel and timing so that the cam could get back into gear. It'd have to do this all while you still have your foot down on the gas.

So when you drive, it would be something like this. Foot down part throttle and you want to shift.

You hit the paddle, and *POP* *POP* *VRROOM* *POP* *POP* *CLUNK* and continue accelerating.

It would take a LONG time.

I didn't think about that until I was driving the Legacy this morning. If you don't double clutch an upshift, you'll have a hard time driving a dogbox smoothly.

As far as doing it cheaply, I don't think there's a way really. You'd NEED a good gearset, you'd have to fabricate the cams or whatever, you'd have to build and program an ECU controller, and you'd have to use alllll billet internals. It'd be tough, no kidding. Your looking at 10 grand or more.

I don't really think this would be a possibility with a synchronized gearset, unless you really don't care about ruining synchros on a constant basis.

I'd suggest modifying an automatic to be paddle-shifted only. It would be a lot easier.

[Subtle]

For a built 4eat look up
www.ipttrans.com

and www.levelten.com

Both also list the same manual shifter.

For a developing paddle shifter check out

www.powertraincontrolsolutions.com/content-4.html

[Carbix]

man i love this forum. you guys are so good.

What i was thinking is what if the throttle was not cable drivin. Driven by something more along the lines of a full time computered crusis control?

If a dogbox is going to be so hard then skip it. would it work if i removed the syncs (and not my clutch) out of my manual for starting in first. And did the rest in the way we have been talking. (by the way loved the idea of 2 cams inside of each other thats so smart and i think would work much better) I dont see how it would be see how it would be so hard to get the timing down. Im not thinking of pree set shifting spots. More as a real time linking so that u dont smash your gears. Fuel mix and retarding was a thought of mine too but if your not having pree set shifting spots and its more for just making sure it goes in clean u wont need much of the retarding. Keep in mind i still want to have the feel of it being a manual and having control over whats going on.

Oh and for the hydrolic system insted of using a normal system that would have hoses. Being thats its a tranny i was more thinking of something casted in steel. This way u can seel it to a much higher psi it can be smaller, and the motor for it can be built right into it. in the end it would bolt right onto the tranny case to make it seemless. lol try and make it part of a tranny coolent system hehe.

Another idea could be that the tranny would only engage on the release of the padle. Say if the retarting system needed more time. I still think the human control of it should be a big part.



P.s I want to test this on a sunfire that i hate so... dont really care for the engine....... rebuilds only take 5 hours