Fastest gearchanges were pre-1990's

[machin]

Williams F1 began work on its semi-automatic gearbox several years before it raced for the first time in 1991. The team’s Director of Engineering, Patrick Head, talks us through its evolution.

“There are several advantages to the semi-automatic gearbox. First, the driver gets to keep both hands on the wheel, so it’s easier for him to get the maximum from the car – particularly through fast corners, when the steering gets very heavy. Second, you can change gear in 30-50 milliseconds as opposed to 200-250 milliseconds, so it’s significantly faster.

We started work on such a gearbox in about 1986. James Robinson was very keen on the idea of servo valve-operated gear changes, so he started looking into it and drew out quite a few of the bits.

“Some people tell me that F1 would be better if the drivers still used stick shifts, but that’s a bit like saying, “isn’t it a pity we don’t still walk around in clogs!”

Full article (although its very brief) here:

http://www.racecar-engineering.com/tech ... gearboxes/

[Richard]

Thanks Machin, the bit I found interesting is:

There was a discussion in ’93 about banning semi-auto ’boxes because people like Alain Prost viewed changing gear with a lever as part of a driver’s skill-set.

“But we decided to keep semi-auto ’boxes because 95 percent of engine failures in the old days were caused by a driver changing down too early and over-revving the engine. The moment you have gear-shifting under electronic control, over-revved engines are out the door and life is much cheaper!”

[autogyro]

Thanks Machin, the bit I found interesting is:

There was a discussion in ’93 about banning semi-auto ’boxes because people like Alain Prost viewed changing gear with a lever as part of a driver’s skill-set.

“But we decided to keep semi-auto ’boxes because 95 percent of engine failures in the old days were caused by a driver changing down too early and over-revving the engine. The moment you have gear-shifting under electronic control, over-revved engines are out the door and life is much cheaper!”

Williams had been working on a CVT Van Doorne transmission which Coulthard tested before it was banned by the FIA.

All CVTs use to much energy to keep the cones or disks engaged and driving and to operate the ratio change. Semis and full auto layshaft shifting was common knowledge to us in 1976.

Bt 1986 we we had discounted the layshaft concept all together and were developing planetary gear sets, both manual and automatic. They were and are lighter, more torque efficient and much stronger.

[munks]

Personally I think today's drivers have it tougher. No they don't shift gears, ( but the cars are quicker which means more physical loading on the body)but they must have presence of mind to use all the buttons and knobs on the wheels, and activate all the systems on a modern GP car, at full speed, listening to someone in your ear. Also, tactics are much better understood today, with everything calculated down to the second, where they expect the edriver to punch out times like a machine.

I think this is a good point. The drivers are at least as good today, but it's a slightly different set of skills. Just thinking about tire management these days, it's not enough just to avoid wearing the tread off. You now have to try keeping them in a specific temperature range. But that's a bit OT ...

The question is do we all agree that the inertias are the ultimate limiting factor? I think that's a reasonable point of view. But if you take it to the extreme, slamming into the next gear close to instantaneously and allowing the elasticities in the driveline to take up the slack and hoping nothing breaks, then you would actually get a small *increase* in acceleration during the shift, no? You're slowing down the engine and speeding up the car to match speeds.

While the traces from the seamless boxes don't necessarily show an increase, they don't appear to show any decrease in acceleration either. I have yet to see a manual shift without a decrease in acceleration, so it's just hard for me to accept that they are ever equal to or faster. If autogyro wants to count all the effects of engine management, clutch usage, etc., then what he really wants to see is equal or better acceleration. Is this not true? If it is true, please show us the evidence of a manually shifted car accelerating as fast as a seamlessly shifted car (or faster, as the thread title claims). If it is *not* true, please explain why.

[Tommy Cookers]

Do the current 'throttle' systems modulate the torque quicker on shifting (due to having a dedicated strategy for this ?) than the pre-90s driver could do ?

Current tyres (ie flat bottom era) are quite 'soft' and should tolerate a torque perturbation below a certain duration without critical loss of lateral grip ( act as a low pass filter to torque impulses) ?

[hardingfv32]

Current tyres (ie flat bottom era) are quite 'soft' and should tolerate a torque perturbation below a certain duration without critical loss of lateral grip ( act as a low pass filter to torque impulses) ?

By definition the tires should be at the limit at all times, thus the tires operating at there 'limit' will not 'tolerate a torque perturbation'.

Brian

[machin]

Current tyres (ie flat bottom era) are quite 'soft' and should tolerate a torque perturbation below a certain duration without critical loss of lateral grip ( act as a low pass filter to torque impulses) ?

I see what you mean Tommy... referencing my earlier quote:

The total variation of torque [with the seamless shift] is actually smaller and can easily be absorbed by the wind up in the drivetrain.

....with the tyres being part of the overall transmission system some of the tyres' elasticity will inevitably be part of the "wind up" that is referred to above by the Xtrac engineer.

I also agree with Brian though; if the tyres are already right at their limit any torque fluctuation will cause slip -I think the answer is not to change gear when you are right at the limit of lateral grip! Although the above quote is interesting as it says the torque fluctuations with seamless shift are less than previous gearshift techniques/systems... so with the seamless shift you can be closer to the limit when shifting...

[autogyro]

I also agree with Brian though; if the tyres are at their limit any torque luctuation will cause slip -I think the answer is not to change gear when you are right at the limit of lateral grip! Although the above quote is interesting as it says the torque fluctuations with seamless shift are less than previous gearshift techniques/systems... so you can be closer to the limit when shifting...

The torque fluctuations during any shift are a result of the amount of torque supplied by the engine and the change in ratio to the output. If there are less fluctuations in torque during a stated 'seamless' shift it can only be because the available torque has been reduced during the shift to soften the gear shift.

If efficient control is applied to this 'reduction' in torque transfer then the shift can be undertaken with little if any gap in acceleration at the output. I agree.

However this acceleration will not be as high as would result from a rapid shift at the rotational inertia shift time limit of the input components without any form of torque reduction surrounding the shift overlap.

The shift in this case would be a 'sledgehammer' shift if undertaken automaticaly of course, or it could be a manualy triggered racing upshift undertaken by a skilled driver. The actual time taken for shift overlap could be the same. Radial dampers can be fitted on the output of either to absorb most of the shock, although this also absorbs some torque. Which is chosen would depend mainly on the level of aerodynamic downforce.

[andylaurence]

I can gaurantee a manual upshift down to 40ms or so on a medium sized geartrain.

I would dearly love to know how. I've never come even close to that. What am I doing wrong?

[Tommy Cookers]

Why should the automated change be a sledgehammer ?

Surely the technology is in place to have modulation of engine torque on an almost microsecond-by-microsecond basis ?
Surely there has been much fiddling of the controlling software on an empirical basis to get the best result overall, maybe with a different setting for the wet ?

The driver wants the quickest way round the track, not the smoothest.
My (track) gearchanges were rubbish, but they only spun me once.

[machin]

Why are Autogyro's posts appearing 12 hours or so after he must've written them? Makes following this thread difficult...???

But what about the problem of the inevitable torque spike when the inertia of a fast spinning engine is slammed into a set of dogs connected to some sticky racing tyres? Phil Roper, principle engineer, explains that it has not proved to be a problem. The total variation of torque [with the seamless shift] is actually smaller and can easily be absorbed by the wind up in the drivetrain. The drivers say it is very smooth and actually unsettles the car less in corners

The only way the 'variation of torque' can be 'actually smaller', is if the torque from the engine is reduced by modulating the shift.

But reducing the torque by modulation is a variation in torque is it not? And Xtrac are telling us there is less variation in Torque with the seamless system.

Here's an analogy to explain how it is possible to have less variation whilst maintaining torque transfer:-

First the "conventional shift":

Imagine you are holding a heavy book at arms length in your right arm. The book represents the gearbox output shaft with a number of gears attached to it. Your right hand represents say 1st gear on the input shaft which is currently engaged. The force on the book is constant (your holding it at a specific height without moving it), this represents the torque from the engine. Now you want to change hands (change gear). You have to completely release the book from your right hand and as quick as you can bring your left hand up to catch it once your right hand is out of the way and then restore it to the original height at arm's length. Naturally in that instant the book is going to fall slightly as the force (torque) on it drops to zero and then gets restored again as you catch it (engage second gear). So there's a big torque fluctuation with a conventional gearbox, whether its manually or pneumatically operated.

Now the Seamless shift:-

Now the shift mechanisms for odd and even gears are independantly operated so you can move your right and left hands independantly, so whilst you hold the book out steady in your right hand you can raise your left hand and gently transfer the load seamlessly from one hand to another. Naturally you won't be able to keep the book perfectly stationary as you do this so there is inevitably some force (torque) variation on the book (output shaft) with this method, but far less than with the conventional system where the gears are selected with just one barrel selector.

But what about the difference in speeds of the various components? Won't this damage the gearbox? The Xtrac engineer tells us that this speed differential is catered for in the "wind up" (or torsional elasticity) in the drive train. The acceleration/deceleration of the various transmission parts will inevitably result in additional torque variation, but what the Xtrac engineer is telling us is that the total torque variation is still less than a conventional (single shift barrell) shift mechanism.

Xtrac tell us that not only are the torque spikes less

The drivers say it is very smooth and actually unsettles the car less in corners

But also that the average torque accelerating the car is higher:-

[seamless shift] is typically worth 3 tenths a lap

Win-win for seamless shift.

[MrEngineer]

Reading Autogyros post earlier, perhaps I misunderstood, but...

Actually, the largest contribution to torque fluctuation is due to the change in engine speed during the shift.
Torque disturbance due to inertia torque of the engine can be over 1000Nm at the layshaft - these engines are "decelerating" (or accelerating for the purists) at over 150,000rpm/s.

Where you are not traction limited however, there is no need for torque cut or clutch modulation. XTrac \ Machin are 100% correct, the compliance of the entire drivetrain and tyre is sufficient to dampen the torque spikes, preventing damage to the gearbox and reducing risk of tyre slip.

[RH1300S]

Guys - can someone help me understand something. There is a lot of talk about torque spikes upsetting the car. I don't quite get it.

Let's assume we are changing UP

When you go from a lower gear to a higher one the torque at the wheel is surely less by a greater or lesser degree depending on the step between ratios (and of course where in the torque curve you change from and to! ).

So the torque spike wouldn't be all that important unless the gearchange was very slow and the power dropped away markedly between one ratio and the next (i.e. dead time between gears). And presumably the closer the ratios the less the problem.

Also, I had always assumed that the major cause of instability when changing gear whilst cornering was the load transfer away from the rear of the car as the torque drops in the dead time between gears. Obviously a potentially big problem on a slow synchromesh style gearbox (road cars), but increasingly less so as the gearchanges get faster. With gearchange speeds being talked about here (manual or mechanical) the interruption in torque happens so quickly the car would hardly be upset at all.

I do understand that the jolt from a harsh gearchange could also cause a spike at the wheels, but surely this isn't the whole of the problem?

BTW as an aside I can remember flat-shifting when I raced MX. The Maico gearboxes could take that (I didn't do it very often though) - there was a distinct kick as the box went from one gear to the next. From that limited perspective I would say it was better to roll back a fraction and get a smoother gear-change. The less grip you had the smoother you had to be. Which probably takes us to something autogyro has been saying - when you have more spare grip you can mask the downside of a harsh gearchange.

It is also worth adding that no matter how much grip a car starts with, when it's near the limit it's near the limit, so the tyres can't take much more.

[autogyro]

The torqe spike is an increase of torque to the output of the powertrain caused by converting the input rotating inertia into torque at the engagement dog or ratchet.

[autogyro]

Do the current 'throttle' systems modulate the torque quicker on shifting (due to having a dedicated strategy for this ?) than the pre-90s driver could do ?

Current tyres (ie flat bottom era) are quite 'soft' and should tolerate a torque perturbation below a certain duration without critical loss of lateral grip ( act as a low pass filter to torque impulses) ?

Current seamless shift systems cannot look ahead on the race track and decide on the level of modulation and speed of shift needed at the next racing dynamic situation. They cannot undertake shifts to match all the variables encountered that a skilled driver using a manual gearbox can.
Programming can be used to meet most variations in acceleration, deceleration and corner forces. In a high downforce car this is sufficient.
In a low downforce car it is not.