F1 manual gearbox operation circa 1991

[gold333]

Not being a critic, just asking questions. I mention the Goodyear thing as up until recently I worked at Goodyear Racing, in the area of F&M data, and do not recall hearing of any data requests of this nature (not that the data necessarily exists anyway). I am aware there is some very brief bit of data in RCVD.[...]

Thanks! One of the reasons that we are developing this 1991 mod is that rFactor1 does not (yet) simulate the turbo unit very well. So a mod on 1985 cars (which does exist) to this level of detail would have been futile. But it's my favorite period aswell.

But we are making this mod "housing" independant. Whether it is run in rFactor1, rFactor2, or some other future application does not matter. It will be portable. The primary technical data of this mod is .xls based, it's just a very very large table of technical specifications and values, from tire slip angles, to airbox dimensions and pressures, dry and wet sump weights, etc. of all cars and models. These values (while constantly being added to) will never change, as 1991 is frozen in time. It's just that newer versions of, say rFactor, will use more of them. our job is to gather everything.

The way we are trying to overcome limitations are by getting actual (F1 and other) race drivers to test the sim so that it feels as right as it can.

Thats awesome that you worked at Goodyear Racing! I have no data at the moment of what information was exchanged and with whom, but I will find out. Do you still know anyone that might have various data (of the types I mentioned) on these old specification slicks at Goodyear Racing? That would be so cool. (I understand that the guys on our team who are researching tyres have not come across you yet).

On a road car I often change gear without a clutch (MUCH slower than a racing 'box would allow) and it is probably more than 90% feel at the lever to get it right.

That is such a coincidence, I have taught myself to drive like this everyday. Once you know the revs the gearlever just smooths in even smoother than a clutched shift

Perhaps for people with a normal gamers 'H' shift you will need to give them a slightly larger window for error, otherwise the skill level required may even be higher than doing the real thing?

Would the Logitech of similar gate shifter stand up to the rigours of changing gear at the speed you will need to make this accurate?

Well I personally use a G27. This is only about $100 and features a clutch, and gated shifter (and pedals and wheel). It definitely stands upto the rigors of abuse of F1 driving speed shifting.

We will give the choice of running it 100% accuracy. Obviously for getting used to it you can set easier gearchange options. The lack of feel you mention using a device such as the gated shifter on a G27 might neccesitate broadening the range slightly of input and output dog differences, this is what we are looking into. But if I understand William Hewland correctly the gearboxes were quite robust and able to deal with considerable dog speed differences.

[Caito]

Ultimately I'm just skeptical of everything - but it's my job to be. IMO, the tire model in rFactor is fairly poor. Even if you had a heap of good data I can't see how you'd make it into something reasonable given that limitation. From what little I've seen, it looks like rFactor 2 may have something more powerful, but still will be up to someone to make something meaningful of it.

I do think it's a cool thing to work on, and that era of car is one of my favorites - overshadowed only by the big turbo cars of the late 80's.

rFactor tire model really sucks, but, you can modify it and make it almost as complicate as you want.


As an example of how "basic" it is:


The slip ratio curve sucks. And the combined is even worse.

The pneumatic trail and force on the wheel is poorly simulated, etc.
Tire wear is plain bad. Wear curves suck.


Anyway, you can search in the net, there's people who simulated a pacejka tire model into rfacor.

[strad]

Weren't they non syncro crash boxes????

[Ciro Pabón]

Sorry for the interruption but, Caito, I don't know if Racer Free could do that better? http://www.racer.nl/dl_beta.htm (includes a Pacejka curve editor).

AFAIK, it also simulates diferentials and gearboxes, with some level of detail. You might find the documentation (I find it very complete) interesting. Go to "Home" in the previous link, the "Documentation", then Gear ratios, Shifting, Clutch, etc. The parameters for each section could give you an idea. Of course, I assume you must know this simulator, so I apologize for perhaps pointing to the evident (but that's one of my jobs here! ). Anyway, I like this simulator a lot because, apparently, to me (duh) the lone author is relatively serious about physics perhaps at the cost of game graphics.

What's a non-synchro crash box?

[strad]

They did not have synchroniser rings but rather had rings with large "dogs" to facilitate clutchless shifting without requiring much finesse in rev matching.

[autogyro]

They did not have synchroniser rings but rather had rings with large "dogs" to facilitate clutchless shifting without requiring much finesse in rev matching.

The syncronizer 'cones' were replaced by 'dog' rings, to increase the speed of disengagement/engagement.
The cones are usualy bronze, sometimes iron and were prone to wear and break.
The process of slipping the cones on the gear to be engaged takes longer than slidding a 'dog' ring to engage in gear cut outs.
The 'finess' in rev matching is a far more difficult skill to master than the rev assisted change of a 'baulk/cone ring' gear change. The revs have to be matched by the driver as there is no 'baulk/cone ring' to do the job for him.
The disengagement/engagement using a 'dog' ring with the slidder directly attached to the gear lever/linkage can be modulated using a wide variety of techniques, timing and other control input.
With an F1 barrel operated selection, this 'feel' is taken away, which leaves the gearbox to be operated by computer without sufficient modulation, or manualy by the driver, who with the barrel shift system no longer has any gear selection 'feel'.

[Muulka]

Just wanted to say that I have this mod, and from what I've played of it, it's amazing! THanks- i have an idea of the work that goes into making something like this- I've been involved in mods for other games in the past (not racing games though...)

[PlatinumZealot]

Hi guys,

We are busy working on building the world's first 1991 F1 simulation software. (the reason for choosing that year is that it was the last year that F1 cars were using predominantly manual gearboxes and passive suspension).

From an engineering standpoint, you know what you have to first? Call an F1 Team and major gearbox manufacturer and ask about it, then see if you can arrange some form of meeting/deal. That is the best thing you can do - no point in re-engineering what already exists.

[Callum]

It sounds like a brilliant project, sadly I know nothing of your questions but good luck anyway!

[gold333]

@n smikle: Yes we have done so. We are in contact with various sources (drivers, teams, manufacturers) but some data we are still missing.

Getting back on track: but what is the technical reason that a driver would not want to damage or wear out the dog surfaces?

Our assumption is that the damage will simply make it more likely that the gearbox will tend to miss shifts more (perhaps amongst other nasty effects? if so which?). If so is it a fairly linear decline with increasing wear or does it reach some threshold and then suddenly become a complete dog's breakfast? I also assume that once you start missing shifts it just compounds the problem, is this correct?

In a crude cumulative damage model, could we relate dog wear after n shifts to a sum of [torque across dogs] * [relative speed of dogs] for each 'poor' shift? "Poor" being when either torque or rev-matching is outside some tolerance provided by Mr. Hewland?

Clearly the speed of the shift also plays a big part in real life but our shifters are just switches so we can't include that (yet).

I'd love it if some of you technical guys could offer advice on the above.