Gear Box Rantings

[Nick]

Dohhhh!
Of course....damn you're good

Oh, I can be as incompetent as the next. Don't get me started on the pneumatic gear shift system I am helping to build. Blind leading the blind would be the description of that

[Max N]

With a 1:1 transformer, you can also do the GEC trick of adding coupling capacitors from input to output. Why bother? Because it can tame the transformer's HF resonance but you still have the advantage of large voltage swing.

I've always thought of bifilar windings as a very low distortion capacitor (rightly or wrongly)

[Max N]

Oh, I can be as incompetent as the next. Don't get me started on the pneumatic gear shift system I am helping to build. Blind leading the blind would be the description of that

Car or bike?

[Nick]

Car or bike

Car with a bike engine.

[Max N]

So does it have a 'bike-type' gear box - sequential shift rotating a barrel which controls the dogs?

[pre65]

Car or bike

Car with a bike engine.

Is this still the MX5, or have you a new toy ?

[Nick]

Its a long story, but to try and keep it slightly shorter. I have been using a local (Shiply) motorsport garage to help me with MX5 stuff (mods to the MX5, coaching and carrying sets of tires to places, and general don't do that, do this instead), and as seems to by my way, I can't leave well alone and just be a customer. They have a sports prototype (think Radical) that has had a GLXR engine and a MBE ecu that also controls the engine gearbox via a pair of pneumatic cylinders. The plan is to swap the GLXR for a Kawasaki H2 engine mainly because there are a lot more of them about. Also looking at swapping the MBE ecu for a Motorsport Electronics ECU as there is more experience of this ECU (its whats in my MX5) in house (and the chap in the next unit with the dyno). So that means the gear control is also needing to be done.

Now if it was only the one case, the £1k difference in the ECU would not be enough of an advantage to justify the change. But there is also a Westfield with a bike engine (CBR1000) that would also benefit from the same sort of gear control. And it would be a nice thing for them to have available to offer customers.

In the case of the H2, the engine cut and blip on change can be done via the ECU as its a DBW throttle. In the case of the Westfield its a cable throttle, so they have made a nice little pneumatic add on in the cable path to allow throttle blipping, and we can do the kill either via the DC feed into the coils, or by faking it via the engine kill input to the ECU.

Its just me doing my normal, "this looks like a fun problem", and it means I have to test the cars out . At the moment we have a pair of pneumatic rams and solenoid valves to push the gear lever, tach input from the engine, and a compressor and air tank with pressure sender. Next step is to connect it all up to probably an Arduino for the moment as its simpler than making a dedicated control board ATM.

If nothing else it will add to my credit account with the garage (and its fun) to get the winter stuff done on the MX5 for next year.

[Max N]

Car or bike

Car with a bike engine.

The reason I asked (and it seems like you are well past the stage where this will be news to you, but might be of interest to some on here...) is that it makes quite a lot of difference when applying auto-shift to an existing manual gearbox.

A typical 4-speed car gearbox will have two selector forks.
The 1st - 2nd selector can engage 1st gear or 2nd gear or neither. This is what is happening when you move the gear lever over to the left and then move it forwards or backwards.
The same is true of the 3rd - 4th selector when you move the gear lever over to the right and then move it forwards or backwards.
The H-gate restricts the movement of the gear lever so that only one ratio can be engaged at any time. As an example, when you shift from 2nd to 3rd, you first have to move the lever forwards to the middle (between 1st and 2nd) which puts the 1st - 2nd selector in the disengaged position. You then move the lever to the right, which means that the lever is no longer connected to the 1st - 2nd selector and is now connected to the 3rd - 4th selector. You can then select 3rd gear.

Early attempts at computer controlled electro-hydraulic gear shifts replaced the gear lever and linkage with an actuator that controlled the 1st - 2nd selector and another actuator that controlled the 3rd - 4th selector. (And so on for 5th - 6th etc)
You can probably guess where this is going.
Everything is OK until you introduce a bug or lose hydraulic pressure or indeed anything that results in two or more actuators selecting gears at the same time. The result is instant catastrophe.

Now contrast this with a typical bike gearbox. We'll stick with 4-speeds for simplicity.
The forks and selectors are the same as on a car. The difference is in how the selectors are moved. There is a cylinder (called a barrel) which has a 'wiggly groove' or channel running around it to control the 1st - 2nd selector and another channel for the 3rd - 4th selector. Each selector has a pin which runs in the appropriate channel.
To move from neutral to first, the barrel is rotated by say 15 degrees. The groove for the 1st - 2nd selector has a wiggle in it between 0 degrees and 15 degrees, which moves the selector from its mid-point so as to select first.
The groove for the 3rd - 4th selector is a straight line over the first 60 degrees of barrel rotation, so that selector will be held in a 'neutral' position for the first 60 degrees of barrel rotation.
The next 15 degrees of rotation will move the 1st - 2nd selector to disengage 1st and engage second. The next 15 degrees will deselect 2nd. From this point on, the 1st - 2nd groove is a straight line, so subsequent rotation will maintain that selector in neutral. The next 15 degrees of rotation moves the 3rd - 4th selector to engage 3rd, and so on.
So no matter how much you rotate the barrel, it is never possible to engage more than one gear ratio. And incremental 15 degree rotations result in the gearbox moving sequentially through the gears.
Now you just need a robust system for rotating the barrel in 15 degree increments.

[Morgan Jones]

Thanks for that; I'd always wondered about what was different inside a bike's gear box to produce sequential shifts.

[Nick]

Yep, it seems to be one of the benefits from fitting a bike engine to a "car" is that you get a sequential box FOC. In theory it all looks simple enough, but I think in practice it will be a matter of getting the timing right (in all cases). Because of course the point of all this is to give clutch-less shifting.

[Morgan Jones]

By the way, my experience is that Suzuki gear boxes have an agricultural feel whereas Yamaha are much slicker. I've no idea how you could race a GSXR - you'd never be able to get the gear change at the exact time you wanted it. My KTM690 had a false neutral between 5th and 6th and I found it as I was riding to the dealer not quite decided on whether to sell it; that decided me.

[Nick]

[Tony Moore]

Love that engine sound!

[Nick]

This project is still progressing. The prototype board actually got to move some things last week when it was connected to the pneumatic cylinder that's connected to the gear lever on the engine, and the throttle pull cylinder in the throttle cable (to do engine blips when changing down). Also controlling the compressor to maintain the operating air pressure. That went well and it did a few laps of Blyton Park messing with the timing of the various bits. Yesterday the next big unknown part seems to be working. Connecting the output from the existing ecu that thinks its driving the ignition coils into the board, and then using those inputs to drive some IGBT's on the board to actually drive the ignition coils. Doing this means I can cut the engine for 60ms (or some other time to be determined) when doing up shifts. The potential for chaos was high with the 400v or so of back emf from the coils and the >5A peaks current. But much to my amazement it all seems to work, engine still runs with the board in the way, and the board now can cut the ignition as required and also now knows what the engine RPM is so can start changing timings based on that. Test runs of that setup next week. The next thing is to get a wheel speed sensor connected, and then its all down to messing with the software to get things like finding neutral working better.

The actual board now is somewhat hacked about from this original picture, with opto isolators on the inputs added and the protection zeners removed from the analog inputs as it killed any hope of accuracy. Some missing decoupling caps will be added and a next version board will be the next step. Should be a bit smaller as well. Plan to get to standard 100*160 euro size.

[shane]

So I’m guessing that double de-clutch heel-and-toe gear changes are a thing of the past?