Gear Control Unit

[Eryk Sokolowski]

Hi,


I am designing a Gearbox Control unit for your car, I am using a micro-controller and electro-penoumatic valves, a pinball air tank and a pneumatic ram.

I got my system working ok but it is not great!

Currently I am running a spark cut for 15ms any thoughts on this time??? (I read in some SAE paper that the minimum time for the gears to become disengaged is 100ms, surely that is an awful lot, F1 gear change is 50ms)

Currently aiming to get the all shift done in under 80ms (The SMG in a BMW E46 M3 can do that time) ...

Am I being too Optimistic??

[Landon Kelley]

Probably not. Could be really difficult without being able to use auto throttle blips. Be careful not to ram any gears around. The gearbox wouldn't last long, though. Remember that F1 gearboxes only have to last 5 or so races and that they cost something like 80+ grand.

[Some Guy]

Can you elaborate on what "is not great" means? That could mean any number of things.

[wweissin]

An F1 gear box compared to an FSAE gear box is apple to oranges. And 15ms may work on a dyno but I can tell you with near certainty that will not work on the road. Typically what we see in the AMA super bike program is around 50ms.

One thing you may want to mull over is power reinstatement. Will you have the ignition come back on in full all at once so you have 100% torque again? Maybe be you could look in to adding the ignition back in degrees per engine cycle. Think about how it affects the chassis if you have all the power hit the tires again right after a shift.I know its not something normally mentioned around here but like I said just something to think about.

[Carrington]

We use Motec for controlling the shift and clutch solenoids. A 100ms cut is required for the 1-2 shift, but the 4-5 shift only needs a 30ms cut. But that's what works for us. The speed of your shift system, and your engine and transmission will all affect what you need for your setup. Solenoids from even the same manufacturer will vary in their activation speeds, and piston speeds will be affected by your pneumatic setup, for example.

[Eryk Sokolowski]

Originally posted by Some Guy:
Can you elaborate on what "is not great" means? That could mean any number of things.

around 40% success rate in shifting

[Tim.Wright]

I think your targets are a bit optimistic. What is your logic for applying force? Timed pulse?

Sounds like the shift is not strong enough or held for long enough. Ive seen many times people finding the shift force by loading up the shifting lever with weights until it shifts, and then calling that the shift force. Then they wonder why their actuator wont shift.

To get a good fast shift, you need a pretty brutal actuator.

A hint, the cut time should be the last thing to optimise. You cant set the cut time, until you know how long the shift actually takes! In anycase, start at 500ms and work your way down. If it doesnt work at 500ms, then the hardware needs a rethink.

Tim

[Bemo]

Well 500ms is quite pessimistic. But 15ms is definitely too optimistic. Try it with longer ignition cuts.
In lap times there isn't such a big difference if it takes a couple of ms longer until you got power at the wheels again. But if your driver never can be sure, if the gearbox really shifted that will cost you quite a lot of time.
And as mentioned before actuation force really shouldn't be too careful. Gearbox actuation is designed quite robust at most motorbike engines as they should last a while and some drivers hit the lever really hard.

[Eryk Sokolowski]

Originally posted by Bemo:
Well 500ms is quite pessimistic. But 15ms is definitely too optimistic. Try it with longer ignition cuts.
In lap times there isn't such a big difference if it takes a couple of ms longer until you got power at the wheels again. But if your driver never can be sure, if the gearbox really shifted that will cost you quite a lot of time.
And as mentioned before actuation force really shouldn't be too careful. Gearbox actuation is designed quite robust at most motorbike engines as they should last a while and some drivers hit the lever really hard.


Yeah but the acceleration event...

[Eryk Sokolowski]

Originally posted by Timo:
I think your targets are a bit optimistic. What is your logic for applying force? Timed pulse?

Sounds like the shift is not strong enough or held for long enough. Ive seen many times people finding the shift force by loading up the shifting lever with weights until it shifts, and then calling that the shift force. Then they wonder why their actuator wont shift.

To get a good fast shift, you need a pretty brutal actuator.

A hint, the cut time should be the last thing to optimise. You cant set the cut time, until you know how long the shift actually takes! In anycase, start at 500ms and work your way down. If it doesnt work at 500ms, then the hardware needs a rethink.

Tim

I doubt that we suffer with force (250N+), the problem was the pneumatic ram it took a massive 150~200ms to fully extend even at 180psi (its rated at 120psi)

We changed it to an voice coil actuator and got the actuator to push the leaver up in 80ms http://fsae.com/groupee_common/emoticons/icon_smile.gif the whole shift is done in 100ms...

I am quite happy about it, tomorrow we are going to do some dynamic testing http://fsae.com/groupee_common/emoticons/icon_smile.gif

[Tim.Wright]

We used a pneumatic system one year. The shifting was very slow on the test rig until I drilled out the inlet ports of the cylinder to allow faster filling of the cylinder. It made an immediate and large difference.

Also, sizing of the cylinder is also critical to speed.

[Eryk Sokolowski]

Originally posted by Timo:
We used a pneumatic system one year. The shifting was very slow on the test rig until I drilled out the inlet ports of the cylinder to allow faster filling of the cylinder. It made an immediate and large difference.

Also, sizing of the cylinder is also critical to speed.

we have tried two sizes of the cylinder ram... both where very slow...

Did the drilled ports help?

[Eryk Sokolowski]

After testing disaster... Drive shaft failure... My guess is that the spark cut was too aggressive and they have seen too much torque reversal...

[Carrington]

we have tried two sizes of the cylinder ram... both where very slow...


Throw a gauge on your system between your cylinder and valve. This will show you if the problem is restricted flow at the cylinder, or restricted flow from the low pressure regulator (which I'm assuming your are using). Some react quicker than others.

I also recommend checking the specs of your cylinders. The manufactures all provide speed/load charts. Are you within ballpark of the specs? If not.... the problem may be elsewhere...

[Eryk Sokolowski]

We are using FESTO pneumatic ram's, Yes we do have a intermediate regulator,

We have looked at the data sheet and the best response time we have got out of it was half the quoted value in the data sheet. (This was going almost to double the rated pressure of the ram)

I think voice coil actuators are the way to go...

[Tim.Wright]

Originally posted by Eryk Sokolowski:

Did the drilled ports help?

Yea, made a massive difference in the speed. The ports were about 1mm diameter and I opened it upto about 3mm from memory, so in terms of cross sectional area, it was a massive increase.

Tim

[Carrington]

We are using FESTO pneumatic ram's, Yes we do have a intermediate regulator,

We have looked at the data sheet and the best response time we have got out of it was half the quoted value in the data sheet. (This was going almost to double the rated pressure of the ram)

We also use Festo pneumatics. We have had problems with slow moving pistons, but it was never the actuator; it was always somewhere else in the system (regulator, binding in linkages or mounting etc). We have not modified port sizes.

Since your cylinders are not moving as fast as spec, I'd suspect the cylinders are not your constraint.

How close are your actuators to your valves? This affects response time also.

Are you sure the force required to accomplish shifts is really what you think it is? The transmission under load won't shift as easy as with no load.