437 replies to this topic

[Gold]

Hi guys,

We are busy working on building the world's first 1991 F1 simulator. (the reason for choosing that year is that it was the last year that F1 cars were using predominantly manual gearboxes and passive suspension).
I'm a part of the team doing the research.

We are stuck on the types of gearbox used.

Some of the questions we have are for example:

1. Were the gearboxes:

Synchromesh, dog or half-dog engagement? (95% sure it was dog, but I'll ask anyway) Also spur gears or helical?

2. Was clutch use optional (or only on upshifts or only on downshifts)?

3. Did revs need to be matched to enable downshifts?

4. Did revs need to be matched to enable downshifts even when using the clutch?

The only information we have are these 2 footwell video's from Katayama driving a manual Tyrell in '93

One slow lap where he uses the clutch almost everywhere:

http://www.youtube.c...eature=r­elated

and one insane qualifying run where he sporadically uses it, predominantly on upshifts(?), and once on downshift entering spoon.

http://www.youtube.c...eature=r­elated

We have the weights and dimensions of some of the gearboxes of the teams to use in modelling the physics of the simulator, but really need some more info. Can anyone help?

Edited by Gold, 29 April 2011 - 12:36.

[nodrift4me]

Hi guys,

We are busy working on building the world's first 1991 F1 simulator.

Second, actually http://www.rfactorce...m?ID=F1 1991 LE

[Gold]

Second, actually http://www.rfactorce...m?ID=F1 1991 LE

Nope first (and only), actually ;)

That's us (F1SR), the LE was the first version (League). This is for the HE 2.0 version (Historical).

[CSquared]

and one insane qualifying run where he sporadically uses it, predominantly on upshifts(?), and once on downshift entering spoon.

http://www.youtube.c...eature=r­elated

Really cool video. It looks to me like he uses it for every shift. It's hard to tell, though, because the clutch pedal travel is so short. He also doesn't engage the clutch for every gear while going down through multiple gears. Nice left-foot braking, too (looks like, at least).

[24gerrard]

The method of gearchange is a power on racing change upshift and a heel and toe downshift to match the shaft rpm.
The heel and toe change uses the side of the foot rather than the heel.
It is used to allow pressure and control on the footbrake at the same time as operating the throttle.

Most of these gearboxes used a dog ring engagement and a slight helical gear shape to reduce gear noise and back lash judder with no measurable reduction in torque transfer.

I did a large number of tests to compare this kind of 'proper' racing and rally driver gear change skill (which the modern driver rarely has), with the two pedal automatic racing gearbox systems we were developing that only requires a clutch for initial take off.
Left foot braking increased the drivers control because there was no longer a need to move the right foot off the throttle to apply the brakes, saving time.

[Gold]

Really cool video. It looks to me like he uses it for every shift. It's hard to tell, though, because the clutch pedal travel is so short. He also doesn't engage the clutch for every gear while going down through multiple gears. Nice left-foot braking, too (looks like, at least).

You're welcome

No he doesn't use it for every shift. Both video's are of the same car on the same weekend. The slow run is where the full clutch travel can be seen as he uses it on almost every gearchange.

The fast run (which you reference) does not use the clutch on many downshifts and only very, very lightly uses it on upshifts and rarely (also very lightly) on downshifts. To be honest it looks like he is just doing it out of habit but knows he shouldn't use it.

As for left foot braking, I believe he only does it in 2 corners, at the exit of 130R, while still on the throttle, and the exit of Dunlop, also while on the throttle.

Edited by Gold, 01 May 2011 - 01:21.

[Gold]

The method of gearchange is a power on racing change upshift and a heel and toe downshift to match the shaft rpm.
The heel and toe change uses the side of the foot rather than the heel.
It is used to allow pressure and control on the footbrake at the same time as operating the throttle.

Most of these gearboxes used a dog ring engagement and a slight helical gear shape to reduce gear noise and back lash judder with no measurable reduction in torque transfer.

I did a large number of tests to compare this kind of 'proper' racing and rally driver gear change skill (which the modern driver rarely has), with the two pedal automatic racing gearbox systems we were developing that only requires a clutch for initial take off.
Left foot braking increased the drivers control because there was no longer a need to move the right foot off the throttle to apply the brakes, saving time.

Thank you but your post confuses me somewhat. I know exactly what heel and toe downchanging and left foot braking are. My questions were more about the gearbox operation.

You mention that the gearboxes were dog ring and and slight helical? As far as I know a gear is either helical (slanted) or spur (straight). What do you mean with slightly helical?

In the video he appears to lift from the throttle when changing up.

1. Was clutch use optional on these boxes? It appears so.

3. Did revs need to be matched to enable downshifts? It appears so, but I wonder if someone has a definitive answer.

4. Did revs need to be matched to enable downshifts and upshifts even when using the clutch? I guess this is the primary question.

Edited by Gold, 01 May 2011 - 01:19.

[Catalina Park]

Thank you but your post confuses me somewhat. I know exactly what heel and toe downchanging and left foot braking are. My questions were more about the gearbox operation.

You mention that the gearboxes were dog ring and and slight helical? As far as I know a gear is either helical (slanted) or spur (straight). What do you mean with slightly helical?

In the video he appears to lift from the throttle when changing up.

1. Was clutch use optional on these boxes? It appears so.

3. Did revs need to be matched to enable downshifts? It appears so, but I wonder if someone has a definitive answer.

4. Did revs need to be matched to enable downshifts and upshifts even when using the clutch? I guess this is the primary question.

have a read of this page on the Hewland website... http://www.hewland.c...vga/advice.html

[24gerrard]

have a read of this page on the Hewland website... http://www.hewland.c...vga/advice.html

William covers most of the gear changing techniques for dog ring racing boxes very well and there is little to add. Although I have no idea how you will reproduce this on a simulator.
Interestingly, it is possible to achieve a racing change on a syncromeshed 'cone' gearbox but the shift has to be very fast. It is also possible to change through all gears without the clutch.
Working on gearboxes for many decades results in circumstances where this has to be done with clutch failures and I am certain William Hewland has done this as well many times.

The strait cut 'spur' gear is the most efficient at transfering torque.
However it is noisy and subject to backlash due to the wider gap of tooth engagement and under no load conditions and dependent on the over all geartrains number of gears and other drive connection types in the train, it can result in a judder on overrun rangeing from annoying and capable of upsetting the car to outright gear and or component damage.
The answer is to use a helical gear with a shallow angle of engagement. These gears are less noisy more positive in torque transfer and very close to strait 'spur' gear efficiency.

[mariner]

At the risk of being Nerdy and telling what you have aleady done there are two references that may be useful.

1) The anatomy and development of the formula one car from 1975 - by Sal Incandela - ISBN 0 85429 320 5

This coveres the older Hewland type manual boxes in some detail, Sal Incandela was a professional F1 mechanic.

2) Formula One Technology by Peter Wright - ISBN 0 - 7680 - 0234 - 6. Enough said.

This covers up to 2000 and has a very detailed chapter on manual and semi auto transmissions. You also get lots of illustrations by Tony Matthews!

BTW Peter Wright quotes 0.2 seconds to do 12 shifts on a semi auto seven speed box.

[24gerrard]

At the risk of being Nerdy and telling what you have aleady done there are two references that may be useful.

1) The anatomy and development of the formula one car from 1975 - by Sal Incandela - ISBN 0 85429 320 5

This coveres the older Hewland type manual boxes in some detail, Sal Incandela was a professional F1 mechanic.

2) Formula One Technology by Peter Wright - ISBN 0 - 7680 - 0234 - 6. Enough said.

This covers up to 2000 and has a very detailed chapter on manual and semi auto transmissions. You also get lots of illustrations by Tony Matthews!

BTW Peter Wright quotes 0.2 seconds to do 12 shifts on a semi auto seven speed box.

Probably the best source material anywhere.
However Peter's quote on the time taken to do 12 shifts on a semi auto seven speed would depend on where the time was measured.
I once measured the shift time of an early hydro pneumatic shifter working a dog engagement against the same dog engagement manualy shifted.
The actual shift time was identical.
Also with a manualy race shifted instalation, the driver is balancing the forces of acceleration, braking and gearchanging in a skilled ballet of actions.
It results IMO in better car control and the speed of gear shift is but a small part of this.
When Nigel Mansell tested the first Ferrari auto/semi auto boxed car he was for ever spinning it as the gear shift took away his fine vehicle control.
Perhaps todays cars are closer to computer games and simulators, so the drivers can do it better.

[Tony Matthews]

However Peter's quote on the time taken to do 12 shifts on a semi auto seven speed would depend on where the time was measured.

1-2, 2-3, 3-4, 4-5, 5-6, 6-7, 7-6, 6-5, 5-4, 4-3, 3-2, 2-1. 12 shifts. I'm not sure what you mean by "where the time was measured". He told me it was like standing next to a Gatling gun.

[24gerrard]

1-2, 2-3, 3-4, 4-5, 5-6, 6-7, 7-6, 6-5, 5-4, 4-3, 3-2, 2-1. 12 shifts. I'm not sure what you mean by "where the time was measured". He told me it was like standing next to a Gatling gun.

Sorry Tony a Gatling gun varies its rate of fire relative to the barrel pack rpm.
Or in the case of the Vulcan cannon and similar, the breech rpm.
Modern versions of these weapons have staggering rates of fire and can be used to down incoming missiles.

Mechanical, hydraulic, pneumatic or electrical shift systems using dog engagement in layshaft boxes can be better compared to a blow back breech weapon.
Of course such systems can be set up with instant selection on a test rig to give a shift speed for 6 upshifts and six downshifts close to the figure Peter quoted.
However in the car the actual speed of shift depends on whether it is fully automatic (downshifts only perhaps) or manualy selected (paddle or buttons).
The shift not only has to change ratios it also has to do it without upsetting the car balance.
This is part of the problem in F1 downshifts that provoke the drivers into requiring a lock up differential to stabalise the car under braking/downshifting.

I still maintain that the actual shift of the selector assembly can be the same speed in either case (manual or auto) and is not dependent on the method of operation.
After all the same selector components can be used in either case. (use a sledge hammer if you like).
My ESERU design uses electro magnetics for the actual shift and the mechanical operation is CVS, that is a constantly variable shift.
It takes longer than a car destabalising automated dog shift but transferes torque throughout the shift in a smooth transition not a big jerk.
The result is a better controlled car.

[Tony Matthews]

Sorry Tony a Gatling gun varies its rate of fire relative to the barrel pack rpm.
Or in the case of the Vulcan cannon and similar, the breech rpm.
Modern versions of these weapons have staggering rates of fire and can be used to down incoming missiles.

Yes, we know all that, it was an off the cuff comment about the sound, is all.

[24gerrard]

Yes, we know all that, it was an off the cuff comment about the sound, is all.

My comments were off the cuff also Tony.
I was just pointing out that to equal the shift speed of the current semi auto F1 layshaft systems.
A manual gearbox driver simply has to move his wrist fast enough.
I am sure there are drivers today who are well skilled in that art.

Edited by 24gerrard, 01 May 2011 - 15:21.

[Gold]

Thanks everyone. I'll try to get those books.

But the question remains; did the revs have to be matched even when using the clutch or not? And how strictly did the revs have to be matched.

Does anyone know anyone on these boards whose driven a 1984-1991 F1 car?

@24gerrard, reproducing on a simulator is not difficult, you just enter the gearbox shape and dry and wet sump weight and the gearing type and some other parameters and you have a 1:1 copy.

Driving the sim you can use a 200$ G27 with clutch, brake throttle and 6 speed shifter on a playseat evo alcantara or a $20,000 simconmotion, whatever you choose.

Edited by Gold, 01 May 2011 - 16:59.

[24gerrard]

Any dog engagement gearbox needs the rpm to be equal on the two assemblies being joined.
Slight differences in rpm, i.e, 200 rpm or so will result in a noisy shift and hopefuly just component wear.
Any rpm difference more than this will damage the components in most cases.
It is essential during the shift, to either lift off the throttle or lift off the throttle and dip the clutch, or simply dip the clutch very rapidly under full power.
This skill has to be undertaken in milliseconds in balance with other control movements.
This is why in almost all shift attempts where the drivers timing is off, it results in an application of engine torque at the wrong time.
This is what causes the damage, not so much the difference in rpm.
If the driver is very alert and has an extra millisecond, he may be able to abort the shift from the feel of the gear lever,
(which may tighten or feel unusual) or engine sound. He can then curse himself for a missed gear change and rush to catch up the time lost.
Much of the gearlever feel needed from a dog box to improve the shift capability of the driver is lost with a mid mounted gearbox using long linkages.
Learning the dog gearbox techniques in a direct lever car before going to an F1 car was always useful.
So using a dog box in any race car with extended gear linkage could be described as partly by guesswork.
With a missed shift the driver might of course simply find himself off the road.
I still find it hard to believe that an accurate simulation can be made of this.

Edited by 24gerrard, 02 May 2011 - 08:52.

[rachael]

My comments were off the cuff also Tony.
I was just pointing out that to equal the shift speed of the current semi auto F1 layshaft systems.
A manual gearbox driver simply has to move his wrist fast enough.
I am sure there are drivers today who are well skilled in that art.

Possibly a manual gearbox can be shifted as fast as a semi auto but definitely it cannot avoid the dog-on-dog engagements that result in a miss-shift or dog damage that the current F1 boxes avoid with clever electronics.

[mariner]

Just to add some confusion (!) here is a drawing of the infamous Lotus queerbox and so comments by Colin Chapman on the benefits of a synchro box and no clutch.

http://www.haynes.co...4413spread3.pdf

The queerbox was truly tiny, later Lotus tried to use the principle for the first F1 semi auto box with ( in true Lotus style) use of the starter motor in reverse to drivethe clutch. Of course they never got it to work reliably.

[24gerrard]

Possibly a manual gearbox can be shifted as fast as a semi auto but definitely it cannot avoid the dog-on-dog engagements that result in a miss-shift or dog damage that the current F1 boxes avoid with clever electronics.

The driver avoids the missed shifts and dog damage by being a skilled fully rounded F1 racing driver.
Most things an F1 driver does today can be done on a simulator or computer game, the only truely F1 requirement is the fitness.
If it was not for this the job could be done by any skilled computer gamer with sufficient guts to accept the very small chance of injury.
Compared to a few decades ago F1 has been emaciated.

[24gerrard]

Just to add some confusion (!) here is a drawing of the infamous Lotus queerbox and so comments by Colin Chapman on the benefits of a synchro box and no clutch.

http://www.haynes.co...4413spread3.pdf

The queerbox was truly tiny, later Lotus tried to use the principle for the first F1 semi auto box with ( in true Lotus style) use of the starter motor in reverse to drivethe clutch. Of course they never got it to work reliably.

I think Williams might be having similar problems this season.
Making layshaft geartrains small and compact has many drawbacks.

[Canuck]

Compared to a few decades ago F1 has been emaciated.

Indeed. Watching the Senna era versus today - no contest. Today's racing is boring by comparison but then that's a result of what the teams have been forced to work within.

[24gerrard]

I do have a lot of respect for those who put together computer games and simulators.
It is just that I fail to see how anyone can reproduce a manual gearchange system, that has many differences between cars, was usualy set up for individual drivers and for which the controls would bear little resemblance to the ergnomics of those in the actual car.
To even get close it would need at least one expert on the design team with direct experience of driving such systems.

I also believe that the majority of people using such a 'game' would soon become frustrated with missed gears and would be unable to master the skills needed.
The result would be the usual modern principle, to simplify the thing for the benefit of swift profits and to meet the low capability of most modern users.
What is needed from computers is a far higher level of education products instead of kids games and for those too reliant on 'virtual reality', to get out and (what is the modern term), get a life.

[munks]

I do have a lot of respect for those who put together computer games and simulators.
It is just that I fail to see how anyone can reproduce a manual gearchange system, that has many differences between cars, was usualy set up for individual drivers and for which the controls would bear little resemblance to the ergnomics of those in the actual car.
To even get close it would need at least one expert on the design team with direct experience of driving such systems.

An excellent point. Even if you got the ergonomics right, it's nearly impossible to produce the physical cues that a driver feels and uses to shift. Even for 'simple' g-forces on the car, all but the most expensive simulators are severely lacking, at least in terms of driver training. The feedback through the wheel can be made pretty accurate, but as far as manual gearboxes I'm not even aware of anybody trying to simulate the precise sensations that a driver might detect. (Well, okay, there was that old arcade game Ferrari 355 Challenge which would block off gears in its H-shifter if you didn't match revs ...)

[24gerrard]

An excellent point. Even if you got the ergonomics right, it's nearly impossible to produce the physical cues that a driver feels and uses to shift. Even for 'simple' g-forces on the car, all but the most expensive simulators are severely lacking, at least in terms of driver training. The feedback through the wheel can be made pretty accurate, but as far as manual gearboxes I'm not even aware of anybody trying to simulate the precise sensations that a driver might detect. (Well, okay, there was that old arcade game Ferrari 355 Challenge which would block off gears in its H-shifter if you didn't match revs ...)

Most road Ferraris of earlier eras than the current flap lever gearshifts, used a gated gearshift lever with a slotted plate.
They did this in the mistaken belief that it would add to the cars sporty appeal by reproducing the look and operation of either an Italian style rear racing gearbox mounted at the back axle 'linkage', or a mid engined F1 'linkage', both of which required such a gated shift plate to give the driver any chance at all of selecting the rquired ratios.
Ferrari missed the whole point of the exercise because this gated shift plate was forced on the design by the gearbox instaled position in the car.
Many performance cars in the UK retained a conventional gearbox position with a much more direct lever action.
The result in almost all Ferrari road cars of this type was to take away ALL of the gearlever feel from the driver and reduce the drivers control over the car.

Reproducing this 'Ferrari' interpretation of gear shifting should be easy to simulate as like F1 cars with mid engined manual gearboxes, the shifts are undertaken less by feel and rather more by guess work.
To drive such mid engined F1 manual gearbox cars at the top level, IMO it is necessary to first master the dog gearbox in a car with a more direct lever actuation, to feel the shifts.

[24gerrard]

I am still interested to know if it is considered possible to simulate this gearchange skill in a computer game.
I believe that comparing reality to what can be done with 'tools' like computers is very very important for the future.

[cheapracer]

I am still interested to know if it is considered possible to simulate this gearchange skill in a computer game.
I believe that comparing reality to what can be done with 'tools' like computers is very very important for the future.

Was being done many years ago since the introduction of Grand Prix Legends.

http://www.google.co...w...mp;aql=&oq=

[24gerrard]

Was being done many years ago since the introduction of Grand Prix Legends.

http://www.google.co...w...mp;aql=&oq=

Thanks cheapy for the link.
All I saw was the flat hat man comparing the circuits with the game, presumably to promote RBS before he parted company when they scewed their customers.
No description of the gear shift comparison with the real cars.

[Gold]

Thanks cheapy for the link.
All I saw was the flat hat man comparing the circuits with the game, presumably to promote RBS before he parted company when they scewed their customers.
No description of the gear shift comparison with the real cars.

Gerrard,

I apologise for not having much time at the moment, I will reply at length later.

But the way we are trying to simulate this is as follows:

-We are not making a "game" or a product for mass market production.
-We are making a simulation, if you have no patience don't run it.
-It is made by enthousiasts and will not be for "sale" it will be freely downloadable by everyone interested.
-You can run it on any PC with a wheel (and if you choose to run it in authentic detail, you will need a clutch and 6 speed shifter, about $150 from logitech)
-It features authentic physics and authentic sounds of every car from 1991
-It features all models of cars from 1991 (inc. for example the 6 different engine specifications of the RA121E and chassis models of the MP4/6, just to mention McLaren)
-The research used are a variety of publications from the era (autosport, motorsport, Heinz Pruller, Autocourse, etc)
-The real F1 teams are being contacted to provide data
-It has been tested by real F1 drivers who raced in 1991 (Mark Blundell for example has been testing) and we are trying to contact other drivers (Eric van de Poele, various others, etc)
-The way we are simulating the gearbox is similar to the way we simulate the entire physics, from "porpoising" of the early 90's F1 cars down to the correct use of the bump stops in setups, or down to the fact that the McLaren was 13kg lighter in Hungary specification as opposed to the previous race, just as an example of the detail.
-The gear box is simulated as e.g. position of the clutch (%) to allow for a successful change, delta or difference between rotational speeds of the gears and axle to allow for succesful change or misshift, max and min rpm difference to allow for or refuse a shift, etc.
-Our aim is developing the software to be 100% accurate, the hardware depends on your budget. For a $150 g27 the gear will simply not engage when you slot in a missed gear change, the stick will shift but the car will be in neutral and you wil hear grinding, if you have a $20,000 simconmotion the gear will physically not engage and you will feel the grinding. Up to your budget really, our aim is to get the software right.
-Our aim is to get a very correct simulation of the cars and the tracks (2 tracks are already finished to -exacting- 1991 specification, down to the kerbs.)
-Our aim is to then plug in actual setup data (received from teams) of Alain Prost, Ayrton Senna, Nigel Mansell, Jean Alesi, Michael Schumacher, etc, so that we can distil or extrapolate and really feel what the driving preference of these gods was like, by actually feeling it in the car we can feel where their specific talents lied. To our knowledge this has never been done before.

Here are some video's of the sim's v1.0 using a Ferrari (where you actually have to lift the throttle with every semi-auto gearchange, just like in the real car.) We are now working on version 2.0.




(you'll notice in that second vid some attention to detail, if you for example choose to show the drivers hands on the wheel in the sim, it will actually use the right hand positions the drivers used, in this example Alesi's famous 11-1 hand position. Just a small example of the level of detail we are aspiring to).

That is a summary of the sim, I'll get back to replying about the gearbox soon. Thanks for your help, it is by understanding the knowledge of people who know (and in addition to publications we can find) that we can build this sim accurately. It is simply the sum of every little bit of data we gather.

Edited by Gold, 12 May 2011 - 22:50.

[24gerrard]

Impressive.
A nice job on the 91 Ferrari.
The shifts need tuning though and having F1 drivers to 'road test' it will be invaluable.
The Ferrari was very sensitive to throttle during shifts (or so I am led to believe).
I am not sure the sim reproduces this from the video.
The earlier manualy shifted cars are going to be far more difficult to simulate acurately, as you obviously know.
I do think your project is of great importance.
I hope you can continue to improve it.

[Gold]

Impressive.
A nice job on the 91 Ferrari.
The shifts need tuning though and having F1 drivers to 'road test' it will be invaluable.
The Ferrari was very sensitive to throttle during shifts (or so I am led to believe).
I am not sure the sim reproduces this from the video.
The earlier manualy shifted cars are going to be far more difficult to simulate acurately, as you obviously know.
I do think your project is of great importance.
I hope you can continue to improve it.

Thanks Yes you are right, in this version the throttle position during shifting is not yet accurately modelled in the Ferrari, that is for version 2.0 we are working on at the moment.

The project itself will be scalable to future applications. Right now it uses rFactor, when rFactor2 comes out it will easily be portable to that (much better graphics, etc.) The "code" of the sim is irrespective of the generation of the racing software which houses/enables it. That is why we are trying to get it perfect, this is (and probably will remain) the only detailed simulation of the last year of manual gearboxes in F1, ever.

Perhaps it is an idea to recruit knowledgeable (and interested) people from these boards to take a look at our development when we have concrete issue or development lists. Ideas and any bits of info are always welcome.

As for the manual gearbox cars; what we are missing now are concrete values (or approximate ranges) of differences in rpm allowed when shifting (both when using and when not using the clutch), and how the position (in %) of the clutch affects this range to enable succesful shifting.

You mentioned a value of 200rpm when matching revs to enable a succesful shift, that appears to me to be a very very small margin, are you sure? The second video above of Katayama (on the qualifying run) he appears to change gears down without using the clutch merely by blipping the throttle. The speed at which Katayama's feet move seem to be too fast to maintain accuracy of 200rpm.

Edited by Gold, 13 May 2011 - 03:50.

[24gerrard]

I was baseing the 200rpm figure on my own experience with dog gearboxes in various race vehicles including super saloon and F3 cars.
The figure is from my memory of the rev counters of the day (which were hardly accurate) and my experience of such boxes on the bench and on test rigs. F1 boxes of this type used a stronger gear set, which began heavier but was reduced in size and weight over the years resulting in geartrains little larger (eventualy smaller) than the lower formula.
This put many gearbox companies off getting involved in F1 because of the fine limits on reliability through strength.
However this shows a range of development that would compare with other dog boxes showing a progresive increase in shift speed and a lowering of shift acceptable rev band with time.
Dont forget, this is not the rev drop between ratios, it is the mechanical capability of the gearbox in rpm range to undertake a shift without excess wear or damage (or clunk or miss shift).
There were never any detailed under load test rigs for the dog boxes in my experience.
The later auto and semi auto boxes have been tested under load to a much higher degree, because of course there is a need to accurately set up the shifting mechanism that is no longer controlled manualy.
Most layshaft design has been understood for decades, so geartrain testing has little use.
There were of course test rigs to confirm the strength of components and some load testing but AFAIK no complete rigs.

[murpia]

200rpm sounds like a good dog delta speed for a smooth shift, although a little slow as you might need 50ms to get the dogring rotation necessary to put the shift in.

The trick is not to attempt the shift with no or very low dog delta speed: the risk is a dog-dog clash and no shift. Greater than a few hundred pm and you'll still get a shift, just a harsher one, so faster = safer / more reliable (reliability of getting a shift, not of the gearbox...).

Nowadays with shift systems communicating with engine ECUs it's feasible to target an exact dog delta speed by applying an engine rev-limiting cut and matching the engine revs to calculated gearbox shaft speeds.

Regards, Ian

[24gerrard]

200rpm sounds like a good dog delta speed for a smooth shift, although a little slow as you might need 50ms to get the dogring rotation necessary to put the shift in.

The trick is not to attempt the shift with no or very low dog delta speed: the risk is a dog-dog clash and no shift. Greater than a few hundred pm and you'll still get a shift, just a harsher one, so faster = safer / more reliable (reliability of getting a shift, not of the gearbox...).

Nowadays with shift systems communicating with engine ECUs it's feasible to target an exact dog delta speed by applying an engine rev-limiting cut and matching the engine revs to calculated gearbox shaft speeds.
Regards, Ian

Thanks Ian.
I agree.
It does show that there is no such thing as a seamless shift with modern layshaft boxes though, doesnt it?

Edited by 24gerrard, 15 May 2011 - 21:15.

[Gold]

Thanks Ian.
I agree.
It does show that there is no such thing as a seamless shift with modern layshaft boxes though, doesnt it?

Guys I am the amateur in this discussion of that there is no doubt, but looking at this video:

http://www.youtube.c...eature=r­elated

Ukyo's feet are tapdancing away on the pedals of that Tyrrell 021 like there is no tomorrow. His feet are literally going nuts on those pedals.

I happen to know that car went from CV joints to Spicer-type joints, it had a 72degree V10 750hp, 310 lb/ft.

More info:

http://www.histomobi...Tyrrell_021.htm


I am trying to find values which will enable a shift with (little) damage, and values that will physically engage but cause some damage, and possibly rpm differences that wont even engage and just grind.

That video appears to me to indicate that the gearbox will shift down as long as he heel&toe's somewhat accurately (seeing as his feet are moving at 1000mph) and you guys are saying 200rpm max delta.

Which is it!

Edited by Gold, 16 May 2011 - 02:34.

[murpia]

Thanks Ian.
I agree.
It does show that there is no such thing as a seamless shift with modern layshaft boxes though, doesnt it?

The so-called 'seamless shift' is designed to eliminate the torque interruption during an upshift so the car accelerates continuously ('seamlessly').

With a twin selector barrel, the new gear is engaged exactly as the old one is released, rather than the single selector barrel method of release then engage. This is in part achieved because the gearbox sensors & software can calculate where the dog 'windows' are, such that the new gear engagement is guaranteed to occur in the minimum feasible time. I think the teams have 200ms to play with. i.e. the software can wait up to 200ms for the gearbox to be ready to do the seamless shift.

During the upshift the engine inertia must be decelerated to the new engaged gear revs, tranferring some additional energy to the car. The ignition cuts are setup to make this process smooth, so the shift feels 'seamless' to the driver (and longitudinal accelerometer). I don't know whether or not the clutch also gets actuated (or maybe just slips a little bit) to achieve this.

I don't believe downshifts are 'seamless' as there's less (or no) performance advantage to do so. I may be wrong.

Regards, Ian

[dav115]

I don't know whether or not the clutch also gets actuated (or maybe just slips a little bit) to achieve this.

I seem to recall there being an article on F1 clutches in Racecar Engineering a while back that addressed this, and basically said it was down to each team's preference.

[24gerrard]

Guys I am the amateur in this discussion of that there is no doubt, but looking at this video:

http://www.youtube.c...eature=r­elated

Ukyo's feet are tapdancing away on the pedals of that Tyrrell 021 like there is no tomorrow. His feet are literally going nuts on those pedals.

I happen to know that car went from CV joints to Spicer-type joints, it had a 72degree V10 750hp, 310 lb/ft.

More info:

http://www.histomobi...Tyrrell_021.htm


I am trying to find values which will enable a shift with (little) damage, and values that will physically engage but cause some damage, and possibly rpm differences that wont even engage and just grind.

That video appears to me to indicate that the gearbox will shift down as long as he heel&toe's somewhat accurately (seeing as his feet are moving at 1000mph) and you guys are saying 200rpm max delta.

Which is it!

The speed of the drivers feet would not be noticably different with a 'proper' race/rally driver using any dog gearbox of similar type.
It is just unusual that it has been filmed in this case.
In a road car the foot movement would include noticable delays as the driver waits for the inertia of components to be overcome.

An F1 engine will change its rpm to match the next gear very very rapidly, as there is hardly any flywheel or clutch mass.

The 200rpm we are talking about, is not the difference in engine rpm between one gear and another at the same output shaft rpm (road speed), The 'rev drop' (defined by the closeness of ratios).

It is the rpm 'difference', that is acceptable between dog shifting components in the gearbox to achieve a racing shift.
As Ian pointed out (and I missed), if there is no difference between input component and output component rpm
( because the driver matched input to output exactly), then there is a good chance that the dog ring will not engage properly and there will be a 'bulked' (blocked shift). The lever will jump forcefuly out of gear position.
There needs to be an rpm difference to achieve a smooth 'dog' engagement.
If the rpm 'difference' is to much, then a higher load is taken by the 'dog' components, which will result in a jerk in engagement with excess wear, actual damage to the components or a rebound miss engagement.
With a direct gear stick (linkage) this can result in physical injury to the driver (I have experienced this with a sprained wrist )
I have used 'wind up' shock absorbing in power trains, rangeing from rubber joints on drive shafts, to radial springs in clutch plates, these are fitted as standard in most road cars with single friction plates.
Such shock absorbing is often used to assist the gearshifting by fine tuning the engagement loads.
It is good to see your attention to detail Gold.
However the 200 rpm maximum delta has to be an educated guess, because there are so many variables and compromises that the gearbox and car will be subjected to from developements and set up.
These changes will alter, engine inertia, loads on engagement components, braking loads and many more.
Which is why a 'good rounded F1 driver' capable of adapting his driving style and supplying good 'feed back' is essential to the whole process.
On the Tyrell video, it may be possible to calculate a figure for this car at the time the video was made but it will not give a full picture for the racing life of that vehicle.

It does show that modern drivers have life very easy by comparison, because almost all of this process is done for them.

Edited by 24gerrard, 16 May 2011 - 10:39.

[Gold]

Gerrard,

I kind of understand your rpm difference between the components of the gearbox.
You see I thought that with 200rpm you meant that a 1991 F1 driver had to hit a certain rpm to within 200revs to enable a succesful up or downshift.
Imagine while heel and toeing on downshifts that he had to rev up the engine (heel) to within 200rpm of a certain value for the lower gear to engage.
That simply seemed impossible at the speed which Ukyo's feet seemed to be moving.

I see now that this understanding I had was wrong.

I understand your explanation of the inner workings of the gearbox, and that e.g. 2 gears having the same rpm will likely just rub the teeth together and not shift, or will bulk.

If you are patient with me I'd like to ask some questions on the "practicality" of driving these manual F1 gearboxes (more than the inner workings of the gearbox).


You see in building an accurate gearbox for the simulator we can alter values such as:

-If you clutch past x% the gear will always engage (not sure if that is correct but that is the current mode).

-If you do NOT clutch past x% then the gear will only engage if the difference in speeds of gears is less than y%.

These two effects are computed independently, x is customisable generally or by car, y is customisable generally or by car and can be different for upshifts or downshifts.





If you will allow me I'd like to ask some (very basic) questions on the practicality of driving a manual F1 car (sorry that I formulated these questions so formally, but I believe it was the clearest way).

What would practically happen / what would you experience in the examples below for a circa 1991 manual F1 car.

Some questions may seem very stupid but I want to make sure we have not missed anything that we didn't know.?


Upshifting

1. a. (Fast Clutch) If you just do a normal upshift, dip the clutch very fast while lifting the throttle it will just shift up I presume. Can anything go wrong with an upshift if you use the clutch like this?

1. b. (Fast No Clutch) What about when you do a fast upshift not using the clutch? If you just quickly lift slightly / or fully off the gas and engage the higher gear. What could go wrong, or will it be 95% foolproof? Does it depend how much you lift off the gas?

1. c. (Slow Clutch) What would happen practically if you upshift like in a roadcar. Lift off the throttle slowly, press the clutch slowly, (revs drop), engage the higher gear. Let clutch out slowly.

1. d. (Slow No Clutch) What if you do this upshift (again) slowly without using the clutch but by trying to match rpm's while in neutral (like you can in a family car)? I.e. hold stable medium rpm so you can ease into neutral, then hold slighly lower rpm until the sweetspot so it gently slots into the higher gear.


Downshifting

2. a. (Fast Clutch blip) If you just do a normal fast downshift, press the clutch fully very fast while braking, heel and toe so you hear the engine (to an unspecified rpm), engage the lower gear and release the clutch. Can anything go wrong with a downshift if you use the clutch like this?

2. b. (Fast Clutch no blip) What would happen in the above example if you did not blip the throttle, but tried to engage the lower gear and released the clutch pedal quickly without the "heel"

2. c. (Fast No Clutch blip) What about when you fast downshift without using the clutch (like in the Katayama video)? If you just do a fast clutchless downshift, while braking, heel and toe so you hear the engine (to an unspecified rpm) and engage the lower gear. Can anything go wrong with a downshift like this? Does the "heel" rpm rise matter in terms of rpm?

2. d. (Fast No Clutch No blip) What would happen in the above example if you did not blip the throttle, but tried to engage the lower gear quickly without using the clutch and without blipping. Ok this is probably the stupidest one. What would happen is probably the same as in your family car.

2. e. (Slow Clutch blip) What would happen practically if you downshift like in a roadcar. Lift off slowly, press the clutch slowly, (revs drop), heel to increase rpm, engage the lower gear. Let clutch out slowly?

2. f. (Slow Clutch no blip) What would happen practically if you downshift like in a roadcar. Lift off slowly, press the clutch slowly, (revs drop), no heel to raise rpm, engage the lower gear. Let clutch out slowly?

2. g. (Slow No Clutch blip) What if you do a downshift slowly without using the clutch but by trying to match rpm's while in neutral (like you can in a family car)? I.e. hold stable medium rpm so you can ease into neutral, then hold slighly higher rpm stable until the sweetspot so it gently slots into the lower gear.

2. h. (Slow No Clutch no blip) Ok this one is again stupid.


Thanks so much for bearing with me, I know it's presumptious to fire these questions at you like this, but I am not one to not try.

PS: I will invite our gearbox programmer to this discussion aswell.

Thanks again for reading all that!

[24gerrard]

Gerrard,

I kind of understand your rpm difference between the components of the gearbox.
You see I thought that with 200rpm you meant that a 1991 F1 driver had to hit a certain rpm to within 200revs to enable a succesful up or downshift.
Imagine while heel and toeing on downshifts that he had to rev up the engine (heel) to within 200rpm of a certain value for the lower gear to engage.
That simply seemed impossible at the speed which Ukyo's feet seemed to be moving.

Not only is it possible to achieve an engine rpm during 'heel and toeing' to within 200 rpm, there is a lot more involved.
The shift is achieved through 'feel' and a complete mastery of the particular car and gearbox.
The rev counter is just a quide.
There are other components with varying input and output rpm and loads that effect the shifts at any one time.
The engine rpm during a shift is not neccesarily the dog input component rpm, depending on it being a clutch or clutchless shift and clutch drag also has an effect.
The output dog components will be effected by wheelspin or wheel lock up under braking and this will vary driver control over the output rpm and varying loads.

It is probably the lack of feel through a dog gearbox on downshift (in modern semi auto F1 cars without them) that is the main motivator for 'lock up' diffs under braking. This is an attempt to replace some of the driver control capability with a dog box under braking.
It achieves little benefit and if downforce were reduced most drivers would lose control under braking.

As you can see, this subject is at least as complex as any other in a racing car, of equal importance for a winning car and sadly ignored by most.
Modern auto and semi auto boxes have relegated real race driver skills to the history books.
Unfortunately the current high downforce and domination of aero covers most of this up.

I could answer all your questions but each one would make up a chapter of a book, which may already be written. (I have yet to find one)
In anycase my experience of actual F1 cars in anger is limited. I do have a very wide experience of many types of gearbox including early racing types and most dog boxes.
What this needs is a much wider debate, with input from both technical experts and of course, drivers with experience in racing the actual gearboxes under investigation.
My memory is short and it is easy to make a slip up.
Driving racing cars with dog gearboxes is very much closer to playing music than simply undertaking an engineering task.
Sadly that art is all but dead today.

Edited by 24gerrard, 16 May 2011 - 22:48.

[Gold]

Yes, I agree that it is a much more involving subject than it appears with e.g. the output dog components will be affected by wheelspin and car inertia even bumps I assume.
It is almost impossible for us to simulate these aspects in their entirety.

The most important thing we can do is establish what happens in a practical sense when you actually drive these things in reality, the different types of scenarios I mentioned. What happens if you do a fast change vs a slow change for example. Just really basic practical driving things, more so than the specific dynamics of the content of the gearbox but more the practical "feel".

I hope I find someone on these boards that has specific practical experience of F1 gearboxes still, I might also try the technical F1 forums. Thanks for the great conversation so far though!

Edited by Gold, 17 May 2011 - 07:57.

[bigleagueslider]

Dog rings are normally used in racing gearboxes because they are much more durable than synchro rings. The clutch is not normally used for shifting with dog ring gearboxes. This is especially true with the tiny carbon clutches now used in F1. These clutches are only good for a few standing starts during pit stops.

It is sometimes necessary to unload the dog ring teeth during a shift so that they can be disengaged. Dog ring teeth have a negative pressure angle that keeps them engaged under load. And thus the dog ring can be difficult to slide out of engagement by the shift fork when under load. The most logical design for dog rings is a 3 tooth arrangement, since that guarantees equal load sharing between the teeth.

Up shifts in a manual racing dog ring box are easier since the engine has little inertia and slows quickly to match the output speed. Downshifts require speeding the engine up to match the more uniform speed at the output shaft, thus the need for "blipping" the throttle. In the past, there have been "automatic" gearbox designs using overrunning clutches, such as the Lotus "queerbox" or the Wisemann design.

I would estimate that the typical speed delta between mating halves of a dog clutch during a shift in a circa-1991 F1 gearbox is probably greater than 200 rpm. The change in engine rpm at each shift is probably 2000 or 3000 rpm. But as noted, dog rings are very durable, so they can handle high impulse loads.

The issue with shifts is not so much the time required for the shift event itself, or the brief interruption of torque delivery to the rear wheels. It's really about the reduction in output from the engine as it moves away from its optimum rpm operating point during a shift. A shift will only require a few hundredths of a second. The engine will require several orders of that time delay to get back up to its peak power delivery point.

[Gold]

Thanks!

Dog rings are normally used in racing gearboxes because they are much more durable than synchro rings. The clutch is not normally used for shifting with dog ring gearboxes. This is especially true with the tiny carbon clutches now used in F1. These clutches are only good for a few standing starts during pit stops.

See below

It is sometimes necessary to unload the dog ring teeth during a shift so that they can be disengaged. Dog ring teeth have a negative pressure angle that keeps them engaged under load. And thus the dog ring can be difficult to slide out of engagement by the shift fork when under load. The most logical design for dog rings is a 3 tooth arrangement, since that guarantees equal load sharing between the teeth.

Ok, so can we assume that in:

http://www.youtube.c...eature=r­elated

Katayama very lightly uses the clutch on upshifts to do precisely what you say; unload the dogring teeth so they can be disengaged?
I also note he releases the throttle only very lightly (20%?) on upshifts.
The gearbox looks very robust, sometimes he lifts off 25% on an upshift, sometimes 5%. Same with this upshift clutch dip.

Up shifts in a manual racing dog ring box are easier since the engine has little inertia and slows quickly to match the output speed. Downshifts require speeding the engine up to match the more uniform speed at the output shaft, thus the need for "blipping" the throttle.

Hmm you say upshifts are easier in a dog ring box, yet in the video we see Katayama using the clutch lightly on precisely the upshifts (to unload the rings?). In the downshifts he does not use the clutch at all (apart from 1 corner). I also note he releases the throttle only very lightly (20%?) on upshifts.

I would estimate that the typical speed delta between mating halves of a dog clutch during a shift in a circa-1991 F1 gearbox is probably greater than 200 rpm. The change in engine rpm at each shift is probably 2000 or 3000 rpm. But as noted, dog rings are very durable, so they can handle high impulse loads.

I see. What would you anticipate the "accuracy" would need to be when blipping the throttle on downshifts to facilitate a "good" downshift. In the video he appears to blip quite lightly on the downshift, compared to the travel of the throttle pedal. I would estimate he blips it only around (15-20%) of the travel, very lightly.

The issue with shifts is not so much the time required for the shift event itself, or the brief interruption of torque delivery to the rear wheels. It's really about the reduction in output from the engine as it moves away from its optimum rpm operating point during a shift. A shift will only require a few hundredths of a second. The engine will require several orders of that time delay to get back up to its peak power delivery point.

Yes, but this the field of engine driveability and driver skill in gear change timing, this is outside the technical discussion of the practical operation of the transmission and gearbox.

Thanks for joining in on the discussion btw!

Edited by Gold, 18 May 2011 - 10:10.

[24gerrard]

From what BS says, the gear shift time for a modern semi automatic box must be in the hundreds of seconds and not the
milli seconds ("like a gatling gun") quoted by Peter Wright.
I still maintain that the actual 'shift time' of syncro components (working in the car under load) is the same for both the auto layshaft, the semi auto and the manual dog boxes. It is only altered by slight design differences, the component mass and inertia.
Peter Wrights bench demonstration was to show how fast the components can be made to operate off load with modern systems.
This is tending to prove that the only gain from modern semi auto F1 gearbox systems is to reduce the driver work load and to potentialy upset the cars balance more easily.

The change in engine rpm is a direct result of the difference in gear ratios and is seldom 3000 rpm, more like 500 rpm to 2000.
A gear shift attempted with a difference in rpm between the input synco components and the output components of 500 rpm or above will almost certainly cause damage to the gearbox. It will definitely upset the cars stability.
To remind the reader, the 'rev drop' between each gear ratio, is NOT the difference in rpm between the engageing components during a gearshift, it can be widely different.

The dog rings are unloaded by taking off the throttle mainly, the slight slipping of the clutch assists, although modern F1 clutches will not last long if over done.
When in a clutched neutral condition, the engine revs can be changed much faster using smaller throttle changes than when loading the powertrain (obvious realy).

Edited by 24gerrard, 18 May 2011 - 10:40.

[Ross Stonefeld]

Thanks!



See below



Ok, so can we assume that in:

http://www.youtube.c...eature=r­elated

Katayama very lightly uses the clutch on upshifts to do precisely what you say; unload the dogring teeth so they can be disengaged?
I also note he releases the throttle only very lightly (20%?) on upshifts.
The gearbox looks very robust, sometimes he lifts off 25% on an upshift, sometimes 5%. Same with this upshift clutch dip.





Hmm you say upshifts are easier in a dog ring box, yet in the video we see Katayama using the clutch lightly on precisely the upshifts (to unload the rings?). In the downshifts he does not use the clutch at all (apart from 1 corner). I also note he releases the throttle only very lightly (20%?) on upshifts.








I see. What would you anticipate the "accuracy" would need to be when blipping the throttle on downshifts to facilitate a "good" downshift. In the video he appears to blip quite lightly on the downshift, compared to the travel of the throttle pedal. I would estimate he blips it only around (15-20%) of the travel, very lightly.




Yes, but this the field of engine driveability and driver skill in gear change timing, this is outside the technical discussion of the practical operation of the transmission and gearbox.

Thanks for joining in on the discussion btw!

I don't know enough about the engine/gearbox combo of Katayama's car from that year, but it looks like a fairly standard H-pattern.

With most racing gearboxes that use a handle, ie not electronic paddles, you can pull for all your might on the lever and nothing will happen if the throttle is at 100%. The quickest way to shift those cars is to pull back on the lever, and the moment you lift off the gas the tension in the gearbox is released and the lever goes into the next gear. Pretty much as quickly as it takes for you to get back to 100% throttle.

[24gerrard]

I don't know enough about the engine/gearbox combo of Katayama's car from that year, but it looks like a fairly standard H-pattern.

With most racing gearboxes that use a handle, ie not electronic paddles, you can pull for all your might on the lever and nothing will happen if the throttle is at 100%. The quickest way to shift those cars is to pull back on the lever, and the moment you lift off the gas the tension in the gearbox is released and the lever goes into the next gear. Pretty much as quickly as it takes for you to get back to 100% throttle.

Fine between 1st and 2nd and 3rd and top gear but not enough across the 'dogleg'.

[Ross Stonefeld]

So you wait a little longer.

[24gerrard]

So you wait a little longer.

Using your technique across the 'dog leg', would result in either a missed shift a damaged gearbox or a very slow gearchange.
It is possible to 'bounce' the syncro slidder' off the disengageing dog ring (or synco cone), it increases wear in the components and risks damage however.
The dog on the other side of the syncro slidder assembly will engage with an added force that will I agree overcome a higher rpm difference between components that would be the case in a slightly slower 'proper' racing change but the engagement will be with a noticeable 'bang'.

I think you will find that Katayama is using a technique that does 'partly' 'bounce' the syncro assembly off the disengageing dog ring on upshifts.
However he uses a 'slight' clutch disengagement as load comes off, to smooth the shift.
This will result in little or no clutch wear. The clutch is always fully engaged during load conditions.
The engagement of the syncro slidder with the next higher gear is therebye matched closer in rpm difference and little if any 'forced' engagement results.
(clutch engaged)
This is 'one' type of 'proper' racing shift possible with the gearbox shown. It is not the only method.

Edited by 24gerrard, 18 May 2011 - 12:38.

[F1Insider]

From what BS says, the gear shift time for a modern semi automatic box must be in the hundreds of seconds and not the
milli seconds ("like a gatling gun") quoted by Peter Wright.
I still maintain that the actual 'shift time' of syncro components (working in the car under load) is the same for both the auto layshaft, the semi auto and the manual dog boxes. It is only altered by slight design differences, the component mass and inertia.
Peter Wrights bench demonstration was to show how fast the components can be made to operate off load with modern systems.
This is tending to prove that the only gain from modern semi auto F1 gearbox systems is to reduce the driver work load and to potentialy upset the cars balance more easily.
...

I don't quite understand how this 'proves' anything, but reducing driver workload certainly isn't the only benefit of a modern F1 box.

[24gerrard]

I don't quite understand how this 'proves' anything, but reducing driver workload certainly isn't the only benefit of a modern F1 box.

I have continualy stated that the speed of the actual gear shift, that is the time it takes for the syncro slider assembly to disengage one gear and engage the next under powertrain load conditions, is limited by these engagement components and not the form or age of the control system.
The speed will be the same in any layshaft gearbox, manual, auto, semi auto or twin clutch/shaft.

Using a manual 'dog' layshaft gearbox in a racing car, is a skill that has to be learnt and makes up a major part of a race drivers ability to drive fast enough to win races.
This skill is far more than just moveing a lever from one gear to the next. It also includes balancing the various dynamic loads on the car.
The brakes, the steering, the engine power, the road surface changes and even the weather conditions.
Replacing a manual layshaft gearbox with one operated by computer, removes this crucial interface with the driver.
He no longer has sufficient feed back through the gearbox to judge such things as braking or throttle application, at least nowhere near as finely as with a manualy operated gearbox.
Most cars now have a lock up differential operated under braking in an attempt to regain some of this lost control under braking.
However the main factor that almost completely masks this loss of 'mechanical' control, is the current 'huge' levels of downforce allowed by the regulations. Until this 'problem' is dealt with F1 at least will remain emaciated and confined to history IMHO.