53 replies to this topic

[Lee200]

No clutches Bob for the Cam and Pawl.  The cams and pawls do all the locking.

Edited by Lee200, Dec 20 2018 - 08:46 PM.

[Robert Fleurke]

First, thanks again Lee for your extended explanation. It's fun to learn more about this, and how it works from an engineering standpoint. I basically understood how the diff worked in GPL and how to adjust it to my style, car and track. But this certainly helps in my understanding, and it's also fun to visualize what the diff does actually.

Lee200, on Dec 20 2018 - 01:55 PM, said:

By using different combinations of ramp angles and number of clutch disks, the differential designer can control how much the two wheels lock together under power on and off conditions.  It is possible to achieve complete lock which effectively is a locked differential, no lock which effectively is an open differential, or any amount of lock in between.  In the last case, the two wheels can still rotate at different rates if the reactive torque difference from the two tires exceeds the locking friction.

I wonder if differentials would be changed during race weekends back in the day (70s), would there be an engineer changing the ramp angles (cage halfs) and amount of clutches (clutch disks)!? (ofcourse as you mentioned before, in 1967 they wouldn't use this type of differential)

My guess is they wouldn't, also since you speak about a differential designer...

Edited by Robert Fleurke, Dec 22 2018 - 07:24 AM.

[Lee200]

AFAIK, about the only thing they could change on a transaxle with either a Cam and Pawl LSD or a Salisbury LSD during a race weekend were the rear drive ratios.

Back in the shop, the mechanics could change the plungers (pawls) on the Cam and Pawl and the ramps and disks on the Salisbury.

Someone else may have different info.

[Robert Fleurke]

I guess it would be a lot of time consuming work in a race weekend to make such a change, and also I suspect most teams/drivers would use the same diff for most tracks.

I've learnt within nowadays F1, they can adjust the diff settings (locking) on the wheel. That's really handy with changing griplevels and changing balance, to compensate or to change tire temps for certain tires. I gathered this hearing onboard communications.

[Michkov]

I imagine if the teams changed the diff settings they'd do it between sessions, same as gear ratios. Not commonly done as far as I'm aware but certainly nothing that couldn't be done in the field if needed.

[mcmirande]

I was testing some values of differentials and I just don't feel the car as behaving "the same" with 30/30:3 and 60/60:6 (power/coast:clutches), as it should be according to the first post of the thread...

Did somebody test it?

Cheers, Marcos.

[Robert Fleurke]

30/30/3 is about 17% locking on/off, and 60/60/6 is about 18% locking on/off (you can see this in GPLRA, the accurate percentage numbers I can't find ATM, John Roberts posted them before)

So Marcos, there's a slight difference, and I'm impressed you feel the difference. Both settings cause a lot of rotation on the thottle, with 60/60/6 slightly more.

Edited by Robert Fleurke, Dec 25 2018 - 02:01 AM.

[mcmirande]

Robert Fleurke, on Dec 25 2018 - 01:33 AM, said:

30/30/3 is about 17% locking on/off, and 60/60/6 is about 18% locking on/off (you can see this in GPLRA, the accurate percentage numbers I can't find ATM, John Roberts posted them before)

So Marcos, there's a slight difference, and I'm impressed you feel the difference. Both settings cause a lot of rotation on the thottle, with 60/60/6 slightly more.

Yes... my point is that 30/30:3 and 60/60:60 should behave almost the same. I'll do further tests with the original gpl cars. My tests were with the 67x and don't know if somebody went that deep in the code to modify some of that calculations...

Thanks and Merry Christmas!!

Marcos

[Lee200]

While researching differentials for this thread, I recently came across a very technical and detailed source for how differentials work.  On page 1487, the formula for locking torque for a Salisbury LSD is given (Equation 27).  Among other factors such as the number of clutch disks, the cotangent of the ramp angle is used.  This make sense if you look at how the cage half moves sideways when the planet axle moves fore and aft.

However as explained before, GPL uses the cosine of the ramp angle in calculating its locking percent.  I now believe this is a minor bug in the code and Papy should have used the cotangent of the ramp angle instead.

The attached graph displays GPL's locking percent if the cotangent is used instead of the cosine in the calculation.  It shows that the same locking percent can be achieved with lower ramp angles.  This may explain why real world power on ramp angles are typically lower than what we need for GPL.

I could easily make a patch that fixes this bug, but it's not worth doing as essentially it's a cosmetic fix.  As stated before, locking percent is what is important for handling regardless of the ramp angle, number of clutches, or trigonometric function used to determine it.

Just FYI.

Attached Files

•  ijaerv13n2_89.pdf   1.64MB   17 downloads
•  GPL Salisbury Differential Locking Percent Using Cotangent.jpg   52.73K   48 downloads

[Cookie]

Thanks Lee for this interesting find.

I do allways use the locking percent in my head for my setups

As far as I understand, we must not change anything

I do vary my setups from 8 to 10 %  on curvy tracks 12-13% for eg Monza...

[Robert Fleurke]

Thanks Lee. I've heard before the ramp angles weren't correct in GPL. This explains why I suppose

So, if I do understand this well, 30/30/3  (cosine ramp angles calculation) should have shown as 30/30/1 (cotangent calculation) in GPL (about same locking percentage). Indeed it would be a cosmetic fix, as long we understand it's about locking percentages, it's no problem

Edited by Robert Fleurke, Jan 04 2019 - 08:05 AM.

[Lee200]

Robert Fleurke, on Jan 04 2019 - 08:04 AM, said:

So, if I do understand this well, 30/30/3  (cosine ramp angles calculation) should have shown as 30/30/1 (cotangent calculation) in GPL (about same locking percentage). Indeed it would be a cosmetic fix, as long we understand it's about locking percentages, it's no problem

Yes.  

I wonder if whoever was writing this piece of code simply made a typo and typed in cos instead of cot.

[John Woods]

Thanks Lee for this thread and great info.

Still I am an idiot.

What is the GPL drive-ability difference between 45/xx/1 and 75/xx/6 when both are 10 percent +/-.

Is it time?
Rotation/time, time/rotation?
Or what if any difference is there?

[Lee200]

As far as I can tell from the code, there is no difference.

[Iestyn16]

There is no spoon!

On a more serious note, I still think in terms of the ramp angles and clutches rather than the locking percentages as shown in GPLSM, which I should probably try and migrate towards in future.

Great thread Lee; it's interesting to still find there are bugs in the code after 20 years!

Rob - that really is one of the areas that could be explained in more detail about modern F1, but there is increasingly less of a focus on the actual driving in the media coverage of it and moreso on the social/storyline side..

[mcmirande]

John Woods, on Jan 04 2019 - 09:41 AM, said:

Thanks Lee for this thread and great info.

Still I am an idiot.

What is the GPL drive-ability difference between 45/xx/1 and 75/xx/6 when both are 10 percent +/-.

Is it time?
Rotation/time, time/rotation?
Or what if any difference is there?

I still would like some talented drivers if they really feel the same comparing those different schemes producing the same lock. I don't feel them the same...

Cheers. Marcos

[Remco Hitman]

Niels Heusinkveld did an experiment years ago, sending out three setups with varying diff settings for blind testing to a bunch of volunteers. Much discussion ensued as to which was which, as opinions varied which was the most aggressive setup.

The diff settings were all equal.

Edited by Remco Hitman, Jan 05 2019 - 04:10 AM.

[Remco Hitman]

Great thread by the way, love the easy to digest intel on differentials!

[TvO]

Who is mister Salisbury?

[Lee200]

TvO, on Jan 05 2019 - 06:37 AM, said:

Who is mister Salisbury?

Robert Michael James Gascoyne-Cecil according to this:  

https://en.wikipedia...ss_of_Salisbury