61 replies to this topic

[gricey1981]

Whilst driving to work today i was pondering the fact that manual transmissions suffer less power loss than automatics.

For example my 540 puts down about 230hp to the wheels where it is rated at 282hp. If I swapped in a manual it would put down nearer 250hp at the wheels.

I wondered what kind of power losses f1 cars are subjected to through the driveline. I am sure they are extremely efficient but any one have any ideas?

[24gerrard]

The epicyclic geartrain in an automatic gearbox is more efficient than the geartrain in a layshaft manual gearbox.
The problem is the torque converter, although modern lock up converters are more efficient and take out little torque when locked.
Conventional automatic gearboxes with converters are not well suited to high performance ic engines with narrow torque bands because of the need to use a torque stall condition for initial start to achieve inertia. The rpm to achieve brakes on stall for the converter has to be matched and usualy means a smaller converter diameter and a higher loss through heat in the oil.
The layshaft geartrain is less efficient than an epicyclic stepped train because the torque has to go off to one side onto a layshaft and then back onto the mainshaft. The epicyclic train sends the torque in a strait line with no offset burst loads, which also means the caseing can be lighter for the same strength.
Dual layshaft or dual clutch transmissions use extra shafting and are even less efficient than a conventional (F1) layshaft geartrain. They achieve high shift speeds using electronics and auto selector actuation but are still only as fast as a racing dog box with a good driver. The silly term 'seamless' is marketing hype and untrue. No layshaft stepped gearbox can be seamless, there has to be a gap no matter how small between one gear and another. Zero shift is another term that is untrue.
Torque loss from an F1 gearbox will depend on the gear design and the oil windage plus bearing losses. It is usual to add up the components transmitting torque in each gear.
Of course in top gear a layshaft box is still driving the layshaft in oil where as in an epicyclic train top can be the whole assembly rotating with only losses from support bearings and slight oil windage.

[bigleagueslider]

gricey1981,

A longitudinal F1 transaxel is likely about 98% efficient input-to-output. The primary sources of loss are gear mesh frictions, with the loaded spur gear mesh contributing around .30% loss and the bevel mesh contributing about .50% loss. The other sources of loss are bearing frictions, seal frictions, and windage. With high gear pitch line velocities, windage losses can become very significant. So a gearbox that has the gears mounted low in the casing, like in F1, must employ a dry sump lube system.

Conventional automotive automatic transmissions tend to have lower efficiencies due to their use of torque converters and the need for a hydraulic pump to constantly provide hydraulic system pressure. Consider this: if the AT's hydraulic pump is using 2HP, while the engine is producing 20HP to drive the car down the highway at 60mph, that amounts to a 10% loss in fuel consumption.

I would disagree with 24gerrard's comment about an epicyclic gear set being more efficient than a spur mesh of equivalent ratio and torque capacity. While the epicyclic might be lighter weight and more compact, the spur mesh would have better efficiency. The epicyclic has two mesh contacts producing loss versus just one for the spur gears. The epicyclic meshes also tend to produce more contact sliding than the spur gears, which hurts efficiency. 24gerrard is partially correct in his comments about the epicyclic not having the unbalanced radial bearing loads that the spur mesh has, but he also neglects the fact that the epicyclic has planet pin radial bearing loads that are 2 times the planet mesh tangential loads.

Modern automotive transmissions are gradually moving more towards the dual-clutch architecture automated manuals (AMT's) and away from the conventional concentric epicyclics with multiple clutches and bands, and torque converter. The dual clutch AMT's are more efficient, lighter, and less expensive. But until the advent of sophisticated digital controls, their shift quality was not acceptable for general automotive use.

slider

[munks]

They achieve high shift speeds using electronics and auto selector actuation but are still only as fast as a racing dog box with a good driver.

I've seen you say this before, but multiple sources indicate F1 shifts are on the order of 10-50ms depending on various factors. I don't have any telemetry showing a good driver on a dog box, but having shifted some in anger myself, it seems unlikely to me that one could do it in less than 100ms.

[24gerrard]

I've seen you say this before, but multiple sources indicate F1 shifts are on the order of 10-50ms depending on various factors. I don't have any telemetry showing a good driver on a dog box, but having shifted some in anger myself, it seems unlikely to me that one could do it in less than 100ms.

I would love to do a test program on dog shifts verses the latest gimmicks.
I had many years of experience with racing dog boxes and today I am still working on and setting up vintage gearboxes with all kinds of weird shift methods.

Also concerning the comparisons between a single layshaft box, dual clutch/shaft layshaft box and epicyclic box.
How many gears are running in engagement depends on the configuration of the torque path and in an epicyclic this can be a lower number than in the set number in the other two types.
In direct top gear in an epicyclic this might be zero gears moving in engagement.
Dual clutch/shaft gearboxes cannot be more efficient in torque transfer than a single layshaft box because they are driving two layshaft gear paths instead of one and windage and bearing losses will still exist even in the shaft not transfering torque in any particular gear.

Edited by 24gerrard, 22 April 2011 - 07:50.

[nodrift4me]

I've seen you say this before, but multiple sources indicate F1 shifts are on the order of 10-50ms depending on various factors. I don't have any telemetry showing a good driver on a dog box, but having shifted some in anger myself, it seems unlikely to me that one could do it in less than 100ms.

I think you should be able to - with a dog-type change it's possible to make a gear change simply by lifting off the throttle a little to help unload the gears and then bang the gearstick hard into the next gear. That's pretty quick.
From the logging of the throttle position on one of my cars that has a conventional synchro-type gearbox, I take about 0.25 seconds to make a conventional change using the clutch.

[munks]

I think you should be able to - with a dog-type change it's possible to make a gear change simply by lifting off the throttle a little to help unload the gears and then bang the gearstick hard into the next gear. That's pretty quick.

Like I said, I know. I've done it myself, just not at an expert level. 100ms is awfully quick, I would say 10ms is impossible (for a mere 5cm throw, you'd have to accelerate the gear stick at about 100 Gs to accomplish this).

DSG can apparently upshift in 8ms at FULL throttle: DSG operation. I think the claim that any human can match something like this is really far-fetched (not to mention the consistency factor).

Edited by munks, 23 April 2011 - 02:37.

[24gerrard]

Like I said, I know. I've done it myself, just not at an expert level. 100ms is awfully quick, I would say 10ms is impossible (for a mere 5cm throw, you'd have to accelerate the gear stick at about 100 Gs to accomplish this).

DSG can apparently upshift in 8ms at FULL throttle: DSG operation. I think the claim that any human can match something like this is really far-fetched (not to mention the consistency factor).

Apparantly upshift in 8ms at FULL throttle.
Exatly apparantly! and they also claim no break in torque between gear ratios.
IMPOSSIBLE.
No stepped geartrain can change from one gear ratio to another without a break in torque from the engine, even if it uses two clutches and shafts.
The engine has to decelerate to the lower rpm of the next highest ratio during upshifts and the reverse of this during downshifts. There has to be a break in transfering torque.
If not then how does the engine supply torque when it is decelerating during an upshift?

The human does not have to match the 8ms operation of the gear selectors.
All the human has to do is move a gear lever with direct connection to the engagement components in the synchroniser assemblies.
It does not matter if this manual human connection uses dog rings, as in a single layshaft racing gearbox, a dual clutch shaft actuation, a so called zero shift actuation or even an epicyclic gear change mechanism, the speed of gear shift will be dictated by the mechanical (or hydraulic) mechanisms in the system. Human skill is the additional factor with the direct dog ring method but there is no reason why it cannot be as fast.
There are other factors at work that dictate gear shift speeds.
One important one is the flywheel and over all mass of the rotating engine components.
The higher this force the slower the gear shift because of the need to change the speed of this rotating mass to match the next gear.

[kikiturbo2]

Apparantly upshift in 8ms at FULL throttle.
Exatly apparantly! and they also claim no break in torque between gear ratios.
IMPOSSIBLE.

we had this discussion once...

the twin clutch gearbox on the new 997 turbo will go from 1st to 2nd with absolutely no break in traction, by virtue of having 1st and 2nd engaged at the SAME TIME for a short while.. so not only it is perfectly possible, it is in the stores for a while now.. .. You have to remember that DSG gearboxes do not change the gears when you shift, just open one clutch and close the other one... the gear changing itself is done in the meantime, while you are accelerating in the selected gear..

Edited by kikiturbo2, 23 April 2011 - 13:53.

[24gerrard]

we had this discussion once...

the twin clutch gearbox on the new 997 turbo will go from 1st to 2nd with absolutely no break in traction, by virtue of having 1st and 2nd engaged at the SAME TIME for a short while.. so not only it is perfectly possible, it is in the stores for a while now.. .. You have to remember that DSG gearboxes do not change the gears when you shift, just open one clutch and close the other one... the gear changing itself is done in the meantime, while you are accelerating in the selected gear..

No it is not possible to have two gears engaged at the same time.
What is in the stores as you put it is a gearbox system sold on hype.
Opening one clutch and closing another results in a break in traction through slip and the generation of heat.
No gearchange is made while accelerating in the selected gear, the next sequential gear is simply selected by the mechanism without transfering torque.
It is a pre-selector system.
The Wilson epicyclic pre-selector gearbox designed and built in the 1930s was a much more efficient method of both transfering torque and in its speed of actual shift, it simply suffered from being to heavy at the time.
The first ERAs won many races using the Wilson, until it was changed for a lighter and less efficient layshaft gearbox.
Racing cars have laboured under the layshaft gearbox either single shaft or DSG ever since.

Edited by 24gerrard, 24 April 2011 - 23:39.

[gruntguru]

Like I said, I know. I've done it myself, just not at an expert level. 100ms is awfully quick, I would say 10ms is impossible (for a mere 5cm throw, you'd have to accelerate the gear stick at about 100 Gs to accomplish this).

I don't see 100G as being impossible, even with manual shifting. There are many mechanisms that can accelerate several orders of magnitude faster than this. Consider also that the actual shifting occurs somewhere in the middle of the 5cm travel, so the mechanism can start accelerating befor the shiftt begins.

No it is not possible to have two gears engaged at the same time.
What is in the stores as you put it is a gearbox system sold on hype.
Opening one clutch and closing another results in a break in traction through slip and the generation of heat.
No gearchange is made while accelerating in the selected gear, the next sequential gear is simply selected by the mechanism without transfering torque.

As a self proclaimed expert in transmissions - including autos - you should be able to envision a shift usings clutches where torque is transferred without interruption. In the middle of a DSG shift, both clutches are slipping and torque is being transferred to the final drive through BOTH layshafts simultaneously.

[rachael]

No it is not possible to have two gears engaged at the same time.
What is in the stores as you put it is a gearbox system sold on hype.
Opening one clutch and closing another results in a break in traction through slip and the generation of heat.
No gearchange is made while accelerating in the selected gear, the next sequential gear is simply selected by the mechanism without transfering torque.
It is a pre-selector system.
The Wilson epicyclic pre-selector gearbox designed and built in the 1930s was a much more efficient method of both transfering torque and in its speed of actual shift, it simply suffered from being to heavy at the time.
The first ERAs won many races using the Wilson, until it was changed for a lighter and less efficient layshaft gearbox.
Racing cars have laboured under the layshaft gearbox either single shaft or DSG ever since.

In F1 two gears ARE routinely engaged at the same time and the upshift IS made whilst accelerating in the lower gear. Post upshift the engine indeed must slow to drive the higher gear but it is still transmitting positive torque - the torque transducer on the input shaft between gearbox and engine does not record a negative torque during the shift.

[kikiturbo2]

What is in the stores as you put it is a gearbox system sold on hype.
Opening one clutch and closing another results in a break in traction through slip and the generation of heat.

while Gruntguru has elegantly answered your post allready, I'll just add my 2c...

While DSG is not my gearbox of choice, I find manual gearshifting to be a part of driving enjoyment, we can not run from the fact that it does work..

looking at torque transfer at the wheels, and that is only thing that actually accelerates the car, 997 turbo gen2 with PDK gearbox will suffer NO interuption in torque at the wheels going from first to second.. In fact, as far as the goon behind the wheel is concerned, you feel as someone has shoved you in the back.. it does work.. The fact that actual "gearchange" or as you put it "gear preselection" is done while you are accelerating in a different gear, makes no difference to the fact that the system works as advertised..

as far as I am concerned, gimme a H pattern dogbox for my rally car, sequential hewland for my trackcar, ad a ZF 6 speed auto for wafting around..

[24gerrard]

In F1 two gears ARE routinely engaged at the same time and the upshift IS made whilst accelerating in the lower gear. Post upshift the engine indeed must slow to drive the higher gear but it is still transmitting positive torque - the torque transducer on the input shaft between gearbox and engine does not record a negative torque during the shift.

Two gears may be trying to transfer torque at the same time Rachael and to do so in the regulation restricted F1 gearbox designs a shift mechanism similar to Zero shift must be used to do so because F1 boxes are AFAIK all single layshaft gearboxes.
However, if two gears are partialy engaged during the gearshift, then neither is fully engaged and therefore there is a condition during gearshifting where no gears are fully engaged.
Using slip to allow two sources of input torque access to output will indeed speed up the shift sequence but to do so torque will be converted to heat and lost.
The other main torque loss inbuilt in the layshaft box and even more in the DSG gearboxes, is the fact that all the gears are rotating in mesh and in oil at all times, even in a direct top configuration.
It just so happens that I have a seven speed stepped gearbox design for F1, which is also a KERS harvesting and apply unit all in one. It does have 'seamless' shift but it achieves it using electro magnetic force to do so (CV shifting). In each gear ONLY those gears are transfering torque and in direct top the WHOLE unit rotates as one with only ONE bearing for support. There is no clutch direct or otherwise dry or hydraulic in the powertrain. It allows full electric engine off use in the pits (or anywhere else), full electric IC start and the geartrain goes into a box size of 84mm.
All it needs for KERS or road hybrid use is an electrical storage and a control system. No need for extra gearing or difficult packaging (interested Adrian?)
Three F1 teams have already said it deserves a feasibility study and two universities would like to do a doctorate on the project if commercial funding can be found.
Being retired I no longer travel around the world like some cap in hand travelling salesman, which is what is expected of you in the current badly balanced F1 industry, which has stayed stagnant to new ideas other than model aeroplane design for well over three decades IMO.

Oh yes and I discused a twin clutch dual layshaft geartrain with Tony Rudd and Alec Stokes way back in the 1970s
It was not developed because the electronic control was not available at the time to operate it.
It is amazeing what comes from engineering machinery drive systems.
I think the idea of dual clutched input torque to single output can be seen in some vintage vehicles back in the 19th century.
I will take a look at the 1908 Panhard we are rebuilding, I think one of our old cars has a cone system that does this.

Edited by 24gerrard, 25 April 2011 - 08:18.

[rachael]

Two gears may be trying to transfer torque at the same time Rachael and to do so in the regulation restricted F1 gearbox designs a shift mechanism similar to Zero shift must be used to do so because F1 boxes are AFAIK all single layshaft gearboxes.
However, if two gears are partialy engaged during the gearshift, then neither is fully engaged and therefore there is a condition during gearshifting where no gears are fully engaged.
Using slip to allow two sources of input torque access to output will indeed speed up the shift sequence but to do so torque will be converted to heat and lost.
The other main torque loss inbuilt in the layshaft box and even more in the DSG gearboxes, is the fact that all the gears are rotating in mesh and in oil at all times, even in a direct top configuration.
It just so happens that I have a seven speed stepped gearbox design for F1, which is also a KERS harvesting and apply unit all in one. It does have 'seamless' shift but it achieves it using electro magnetic force to do so (CV shifting). In each gear ONLY those gears are transfering torque and in direct top the WHOLE unit rotates as one with only ONE bearing for support. There is no clutch direct or otherwise dry or hydraulic in the powertrain. It allows full electric engine off use in the pits (or anywhere else), full electric IC start and the geartrain goes into a box size of 84mm.
All it needs for KERS or road hybrid use is an electrical storage and a control system. No need for extra gearing or difficult packaging (interested Adrian?)
Three F1 teams have already said it deserves a feasibility study and two universities would like to do a doctorate on the project if commercial funding can be found.
Being retired I no longer travel around the world like some cap in hand travelling salesman, which is what is expected of you in the current badly balanced F1 industry, which has stayed stagnant to new ideas other than model aeroplane design for well over three decades IMO.

Oh yes and I discused a twin clutch dual layshaft geartrain with Tony Rudd and Alec Stokes way back in the 1970s
It was not developed because the electronic control was not available at the time to operate it.
It is amazeing what comes from engineering machinery drive systems.
I think the idea of dual clutched input torque to single output can be seen in some vintage vehicles back in the 19th century.
I will take a look at the 1908 Panhard we are rebuilding, I think one of our old cars has a cone system that does this.

There are two 'seamless' shift strategies in current F1; twin barrels and over-run ratchets - I've posted previously describing how they differ. In both systems the gears are fully engaged during the shift - only during downshifting is the clutch slightly opened to allow a bit of slip.

[Greg Locock]

and to be boring, and getting back to the OP's question, the losses in hypoid diffs tend to be rather enormous.

At low torque levels they generate a lot of friction, at high torque levels they distort the housing, so they generate a lot of friction. So they are re-aligned to generate less friction at high torque levels so they generate even more friction at low torque levels.

Measuring the torque loss in a diff is not easy, there is a good SAE paper on it by Dana or Ford, they use IR thermography. In a typical city drive cycle the diff of a small car absorbs a significant fraction of the total energy generated.

[24gerrard]

There are two 'seamless' shift strategies in current F1; twin barrels and over-run ratchets - I've posted previously describing how they differ. In both systems the gears are fully engaged during the shift - only during downshifting is the clutch slightly opened to allow a bit of slip.

I am well aware of this.
You are describing two gears 'selected' at the same time, not two gears 'engaged'.
I will say it again, It is impossible for any stepped ratio gearbox to have two gear ratios 'fully engaged' at the same time.

If as you say both gears are "fully engaged", then why would there be a need for a gear shift?

I am also dubious about the 5 and 8ms shift times claimed for F1 and DSG type boxes.
In fact with a road car engine with a heavy flywheel and clutches I do not think it is true.
Dont forget that during the timed shift the engine rotating components have to be speeded up or slowed down by a minimum of around 500rpm.

Edited by 24gerrard, 26 April 2011 - 12:09.

[rachael]

I am well aware of this.
You are describing two gears 'selected' at the same time, not two gears 'engaged'.
I will say it again, It is impossible for any stepped ratio gearbox to have two gear ratios 'fully engaged' at the same time.

If as you say both gears are "fully engaged", then why would there be a need for a gear shift?

I am also dubious about the 5 and 8ms shift times claimed for F1 and DSG type boxes.
In fact with a road car engine with a heavy flywheel and clutches I do not think it is true.
Dont forget that during the timed shift the engine rotating components have to be speeded up or slowed down by a minimum of around 500rpm.

This sounds like arguing over semantics but if both gears are capable of transmitting drive torque we would say they are both engaged. In the ratchet system the lower gear is engaged and supplying drive torque up until the point it is over-sped by the higher gear and is then disengaged by the ratchet - ditto the twin barrel system. If the system doesn't work correctly you very quickly get a box full of shrapnel.

The driver can only select one gear at a time so there is not the possiblity of 'selecting' two gears.

I've always wanted to understand how the Wilson works - got any pictures that show the internals?

[munks]

I am well aware of this.
You are describing two gears 'selected' at the same time, not two gears 'engaged'.
I will say it again, It is impossible for any stepped ratio gearbox to have two gear ratios 'fully engaged' at the same time.

If as you say both gears are "fully engaged", then why would there be a need for a gear shift?

I am also dubious about the 5 and 8ms shift times claimed for F1 and DSG type boxes.
In fact with a road car engine with a heavy flywheel and clutches I do not think it is true.
Dont forget that during the timed shift the engine rotating components have to be speeded up or slowed down by a minimum of around 500rpm.

Semantics aside, you are (obviously) correct that if clutches are slipping, energy is being lost. But there's also some portion of the energy that can be stored in shaft compliances and clutch springs.

In other words, if one desired, a system could be designed where both gears were "fully engaged" at the same time, but they're both winding up torsion springs or similar. You'd want to complete the shift before the torsion springs broke, but at a 500rpm difference between gears, it's only winding it up at 3 degrees per millisecond. Electronics could coordinate such a shift, humans probably couldn't.

[24gerrard]

Semantics aside, you are (obviously) correct that if clutches are slipping, energy is being lost. But there's also some portion of the energy that can be stored in shaft compliances and clutch springs.

In other words, if one desired, a system could be designed where both gears were "fully engaged" at the same time, but they're both winding up torsion springs or similar. You'd want to complete the shift before the torsion springs broke, but at a 500rpm difference between gears, it's only winding it up at 3 degrees per millisecond. Electronics could coordinate such a shift, humans probably couldn't.

Both gears could be 'selected' not engaged.
This is a very important point.
Unless full input torque is being transfered to output with just system losses, then the gear is not fully engaged.
I have often used radial springs in clutch plates and drive line components to both 'wind up' shifts and to take out gear change 'shock', when replacing torque converters with direct engagement systems or designing rapid shift systems for various stepped ratio gear boxes.
It may be possible to use wind up springing as you suggest but there is still an energy loss winding the springs up which is a loss just as slipping clutches in DSGs is.
In anycase in the 19th century layshaft geartrain either single or twin shaft, all the gears even those not engaged are still churning around in mesh in an oil bath and multiple bearings wasting energy.
My design of planetary gearing only has the gears in use rotating in mesh and in direct top non are rotating in mesh at all. It has a shift overlap like any other stepped gearbox but it is a constantly variable torque shift NOT so called seamless.
(I thought this term was only used in garment manufacture, it means nothing in a gearbox).

Edited by 24gerrard, 26 April 2011 - 15:03.

[24gerrard]

This sounds like arguing over semantics but if both gears are capable of transmitting drive torque we would say they are both engaged. In the ratchet system the lower gear is engaged and supplying drive torque up until the point it is over-sped by the higher gear and is then disengaged by the ratchet - ditto the twin barrel system. If the system doesn't work correctly you very quickly get a box full of shrapnel.

The driver can only select one gear at a time so there is not the possiblity of 'selecting' two gears.

I've always wanted to understand how the Wilson works - got any pictures that show the internals?

I do not believe it is a question of word semantics.
It is a case of the marketing goons getting hold of high level engineering and polluting it with their hype.
What is the mechanism used to move the gears. The selector mechanism, be it a twin barrel selector fork system, a ratchet (sprag) system, a leverage system like the zero shift or a twin clutch/shaft system. It is not called the engagement mechanism is it.
The selector mechanism moves the gear components needed for engagement and the gear is only engaged when torque is transfered from input to output with only system losses.

The Wilson was a pre-selector gearbox and was fitted in the early 1930s ERAs as well as a number of road cars including the Armstrong Siddeley.
It had a number of planetary sets in line with the sun gear of the first driven by the engine and each planetary set driving the sun gear of the next in line.
The annulus of each set formed a drum upon which a brake band operated.
In neutral all these bands were off and each planetary was 'engaged' by applying its brake band.
The planetaries all had ratios for a different gear and first was applied by slipping its band.
'Selection' was achieved through the use of a mechanical linkage connected to a gear lever and the clutch pedal. The next gear needed by the driver was 'selected' by the driver with the lever (similar to an F1 driver using a flap paddle or button to select the next upshift). This 'sets' the linkage.
When the clutch is pushed down and up it would move the linkage to disengage the engaged gear band and 'apply' the new one selected. Two gears were NOT 'engaged' at the same time but a very fast constantly variable shift was achieved (not seamless), with torque transfered from input to output throughout. 1930s ERAs won many races until the Wilson was replaced with a lighter 19th century technology layshaft gearbox similar to modern F1 geartrains.

http://en.wikipedia....elector_gearbox

Edited by 24gerrard, 26 April 2011 - 14:50.

[rachael]

I do not believe it is a question of word semantics.
It is a case of the marketing goons getting hold of high level engineering and polluting it with their hype.
What is the mechanism used to move the gears. The selector mechanism, be it a twin barrel selector fork system, a ratchet (sprag) system, a leverage system like the zero shift or a twin clutch/shaft system. It is not called the engagement mechanism is it.
The selector mechanism moves the gear components needed for engagement and the gear is only engaged when torque is transfered from input to output with only system losses.

The Wilson was a pre-selector gearbox and was fitted in the early 1930s ERAs as well as a number of road cars including the Armstrong Siddeley.
It had a number of planetary sets in line with the sun gear of the first driven by the engine and each planetary set driving the sun gear of the next in line.
The annulus of each set formed a drum upon which a brake band operated.
In neutral all these bands were off and each planetary was 'engaged' by applying its brake band.
The planetaries all had ratios for a different gear and first was applied by slipping its band.
'Selection' was achieved through the use of a mechanical linkage connected to a gear lever and the clutch pedal. The next gear needed by the driver was 'selected' by the driver with the lever (similar to an F1 driver using a flap paddle or button to select the next upshift). This 'sets' the linkage.
When the clutch is pushed down and up it would move the linkage to disengage the engaged gear band and 'apply' the new one selected. Two gears were NOT 'engaged' at the same time but a very fast constantly variable shift was achieved (not seamless), with torque transfered from input to output throughout. 1930s ERAs won many races until the Wilson was replaced with a lighter 19th century technology layshaft gearbox similar to modern F1 geartrains.

http://en.wikipedia....elector_gearbox

It is not called the engagement mechanism but since it engages the gears it could be. On a regular manual gearbox gear lever movement left to right is 'select', forwards or backwards is 'shift' so you could argue that it's the shift mechanism that causes the engagement not the select. Enough already.

On the Wilson - I have previously read the words and looked in the open top cover on era's in vintage padocks, what I need is an exploded or cutaway view. ;-)

[24gerrard]

What are the forks that move the synchroniser assemblies called Rachael?
Would it be 'selector forks' by any chance? truce?
I have not got an exploded view of the Wilson box handy Rachael.
I do have piles of old workshop manuals where I think I remember seing one.
I will take a look.
We have a 1930s Armstrong in the museum being rebuilt that was once owned by Harold MacMillan but unfortunately it has a manual gearbox.

[munks]

It may be possible to use wind up springing as you suggest but there is still an energy loss winding the springs up which is a loss just as slipping clutches in DSGs is.

I'm not saying the idea is practical, but winding a spring up is not a loss "just as slipping clutches" are. In one case, the energy gets stored in the spring. In the other, it is truly lost forever as heat.

I guess you may be talking about the gear that is to be disengaged - yes the energy in that spring will get lost. But let's change my fantastic and impractical design such that only the new gear will wind up the spring. In that case, all the energy will eventually get returned, minus a tiny bit of internal material friction in the spring and perhaps some lubricating losses.

I don't really care if a system with compliance is called 'engaged' or not. There are always compliances, though.

[Magoo]

and to be boring, and getting back to the OP's question, the losses in hypoid diffs tend to be rather enormous.

At low torque levels they generate a lot of friction, at high torque levels they distort the housing, so they generate a lot of friction. So they are re-aligned to generate less friction at high torque levels so they generate even more friction at low torque levels.

Measuring the torque loss in a diff is not easy, there is a good SAE paper on it by Dana or Ford, they use IR thermography. In a typical city drive cycle the diff of a small car absorbs a significant fraction of the total energy generated.

Indeed. Given the active radiator inlets and all the other fine whittling they are up these days, you have to believe hypoids are not long for this world. What to replace them with, there's the rub.

[rachael]

What are the forks that move the synchroniser assemblies called Rachael?
Would it be 'selector forks' by any chance? truce?
I have not got an exploded view of the Wilson box handy Rachael.
I do have piles of old workshop manuals where I think I remember seing one.
I will take a look.
We have a 1930s Armstrong in the museum being rebuilt that was once owned by Harold MacMillan but unfortunately it has a manual gearbox.

Workshop manuals - they are great aren't they, here's a page from a Toyota gearbox one - can you see how the forks are labelled;



I believe in common parlance they can be called either but the correct term is shift ;-)

[benrapp]

I believe in common parlance they can be called either but the correct term is shift ;-)

Speaking only as a victim, I feel I should point out that every time I ham-fistedly damaged the gearbox in my race car when trying to shift faster, I was billed for replacing a bent "selector fork"...

Perhaps Porsche and Toyota differ on terminology.

[24gerrard]

Speaking only as a victim, I feel I should point out that every time I ham-fistedly damaged the gearbox in my race car when trying to shift faster, I was billed for replacing a bent "selector fork"...

Perhaps Porsche and Toyota differ on terminology.

Shift is an american word taken from their general use of automatic gearboxes.
Few Americans can use a 'stick shift' even today. (sounds like a candy bar) 'gear lever' is the correct name or 'manual gearbox'.
The Japanese have followed American English since WW2, after all it was the Americans that supplied their nuclear power plants that allowed the Japanese dominance of the motor cycle and most of the car industry.
(It is realy a shame that the nuclear disaster in Japan is now showing the cost for taking such risks in power generation, to achieve the 'dream'.)
I speak and use British English which was used in some of the original development of gearboxes as German and French was in Europe.
The word 'shift' to me is a bit like 'seamless' and makes me think of ladys peticoats and zips on jeans.

[bigleagueslider]

I suppose one could make the argument that all 7 sets of shift gears in an F1 manual gearbox are always "engaged". F1 gearboxes are "constant mesh" designs, which means that the components that are "engaging or disengaging" are just the dog ring teeth.

As for a friction clutch not transmitting torque if it is used during a shift, that is not true. Even though the clutch may be slipping, it is still transmitting torque. Of course, the clutch is really a speed synchronizing device. That is its function is to synchronize the speed between the crank and the transmission input shaft.

Most racing gearboxes don't use the clutch for shifting. The synchronizing is done by the driver controlling engine speed via the throttle, or by the ECU performing that same function.

As someone noted, there are transmission systems that use various overrun or ratcheting devices in place of conventional dog rings, such as Zeroshift. Such systems allow essentially uninterrupted torque transfer as one gear overruns the other.

In response to the OP, there are many sources of driveline losses besides the gearbox itself. Technically, one could include parasitic losses from powering the shift hydraulic system, friction losses at each CV tripod, wheel bearings and seals, or even the drag loss from the transmission oil cooler.

slider

[Tony Matthews]

[MatsNorway]

Nice!

Now.... how the hell do you pull that off? do you have to calculate the line length and angle and stuff? on the lines going "inward"

[Lee Nicolle]

and to be boring, and getting back to the OP's question, the losses in hypoid diffs tend to be rather enormous.

At low torque levels they generate a lot of friction, at high torque levels they distort the housing, so they generate a lot of friction. So they are re-aligned to generate less friction at high torque levels so they generate even more friction at low torque levels.

Measuring the torque loss in a diff is not easy, there is a good SAE paper on it by Dana or Ford, they use IR thermography. In a typical city drive cycle the diff of a small car absorbs a significant fraction of the total energy generated.

On a modified Falcon turbo that an aquaintace owns. He got talked into fitting a 9" diff 'because the BW will never handle the power [which it just has for several years] After fitted it he thought this feels lazy plus the car did not handle as well as before. Ratio was near the same 3.25-3.27. When dynoed it had lost 8kw at the rear wheels. He then sold the 9" again and has refitted the another BW and the car is all happy again.

Many years ago on my Sports Sedan I found similar losses but were offset by gaining better ratios compared with my previous GM Salisbury.

And yes thinking about the fact that any diff gear is trying to force itself apart makes you appreciate that somebody generally is doing something right in the engineeering. Modern oils help too,, a lot.

And the forces in an F1 car really must be something, though drag racing really loads them up!! And they never seem to bust them. Sprintcars are another case too, though they do break them occasionally, and often if the gears are not replaced every 8-10 shows [410] and those gears are fairly homely also. 70s Ford Freighter!
And a midget still uses Ford V8 gears.

[kikiturbo2]

Tony, that is just sick, in the best possible meaning..

[Greg Locock]

Peugeot used to use a worm drive diff. Those of us who had Meccano were always told the worm drives were nasty inefficient things, yet my machinery design book gives some equations that seem to indicate that very respectable efficiencies are possible. So perhaps we might go forwards by going back to the worm drive diff.

[Charles E Taylor]

Efficiency of the F1 Gearbox.


Some published detail from 2009.

With the advent of DLC coatings and Low Friction gear oils, overall transmission efficiency approaches 97%


Transmission efficiency losses can be characterized as.


Churning-----Seals-----Bearings-----Gears-----Pump-----Condition
40%---------6%-------0%----------50%------6%-------Non-Loaded
14%---------2%-------20%------- -62%------2%-------Loaded


I hope this helps.



Charlie.

[24gerrard]

I suppose one could make the argument that all 7 sets of shift gears in an F1 manual gearbox are always "engaged". F1 gearboxes are "constant mesh" designs, which means that the components that are "engaging or disengaging" are just the dog ring teeth.

As for a friction clutch not transmitting torque if it is used during a shift, that is not true. Even though the clutch may be slipping, it is still transmitting torque. Of course, the clutch is really a speed synchronizing device. That is its function is to synchronize the speed between the crank and the transmission input shaft.

Most racing gearboxes don't use the clutch for shifting. The synchronizing is done by the driver controlling engine speed via the throttle, or by the ECU performing that same function.

As someone noted, there are transmission systems that use various overrun or ratcheting devices in place of conventional dog rings, such as Zeroshift. Such systems allow essentially uninterrupted torque transfer as one gear overruns the other.

In response to the OP, there are many sources of driveline losses besides the gearbox itself. Technically, one could include parasitic losses from powering the shift hydraulic system, friction losses at each CV tripod, wheel bearings and seals, or even the drag loss from the transmission oil cooler.

slider

Constant mesh is one of the reasons layshaft gearboxes are inefficient in transfering torque.
It is overlooked mainly because for over 100 years the concept has been accepted as the norm
while most development has been undertaken on the now done to death ic engine.
Uninterrupted torque transfer does not have to include fully 'engaged' gears or drive components.
In fact during gear changes (shifts) in any layshaft gearbox or any stepped gearbox this is never the case.
With the current KERS and HYBRID powertrains there seems to be a race as to who can add on the most
parasitic loss to the gearing.
The Ferrari FF (as discused in another thread), reminds me of a block of flats with all its extra gearing.
Red Bull are struggling to get their KERS to work because Adrian has concentrated on aerodynamics and left little space for the KERS system.
The current instalations using a crank nose driven generator/motor is IMO far to complex and bulky.
My unit replaces the gearbox, is also the motor generator but with finer control and is smaller.
It also allows full engine off traction under electric only with the ice totaly disengaged from the powertrain.

[24gerrard]

Efficiency of the F1 Gearbox.


Some published detail from 2009.

With the advent of DLC coatings and Low Friction gear oils, overall transmission efficiency approaches 97%


Transmission efficiency losses can be characterized as.


Churning-----Seals-----Bearings-----Gears-----Pump-----Condition
40%---------6%-------0%----------50%------6%-------Non-Loaded
14%---------2%-------20%------- -62%------2%-------Loaded


I hope this helps.



Charlie.

How much is one support bearing in direct top Charlie? (input drives directly from crank in place of the 'not needed' clutch.
With no rotation of gears in mesh and the whole assembly rotating as a 'locked' unit.
No pumps or parasitic losses, no churning (only oil film).
That is much closer to 100 percent I beleive.
In the lower gears 'only' one gear set is in use, all the other five are locked.
Dont forget, it is also a KERS generator and motor.

[24gerrard]

What a superb drawing Tony, thank you so much for that.
I used to pride myself on technical drawing in my youth but I have not been near a board for years.
I am well aware of the huge skill needed to achieve such a drawing and the strain in eyesight.
Modern CAD users have never lived.
I run the odd art show and music gigs and there is definitely a very special artist in you Tony, I can feel that through your work.
No doubt another 'interesting' personality.

[24gerrard]

Tony, that is just sick, in the best possible meaning..

Unfortunately you cannot overrun a worm diff.

Sorry I replied to the wrong post, it should have been to Greg, on his suggestion to use worm diffs.

Edited by 24gerrard, 27 April 2011 - 11:32.

[Greg Locock]

You can certainly back drive a worm gear. After all the thing is just an inclined plane.

[gruntguru]

You can certainly back drive a worm gear. After all the thing is just an inclined plane.

Yep - just depends on the angle of that plane and the coefficient of friction. Once owned a Peugeot 403 that had to be push-started many times.

Interesting lubrication requirements for worm gears. A lubricant property called "slickness" (not required by other gear types) is useful for the "long", sliding contact zones that exist in a worm gearset, and tend to squeeze normal lubricant out.

[24gerrard]

Yep - just depends on the angle of that plane and the coefficient of friction. Once owned a Peugeot 403 that had to be push-started many times.

Interesting lubrication requirements for worm gears. A lubricant property called "slickness" (not required by other gear types) is useful for the "long", sliding contact zones that exist in a worm gearset, and tend to squeeze normal lubricant out.

http://en.wikipedia....wiki/Worm_drive

Not with a low ratio worm and wheel you cant.
They are not ideal for overrun braking due to excess wear in this condition.

Peugeot did a lot of high level gear development in the early years.
They had a gearbox with 'chevron' gears to eliminate side loads.
That one took some clever gear cutting to build.

Just for interest, the chevron gears developed by Peugeot were the inspiration for their bonnet badge.

Damn, I am an idiot, I got that completely wrong, sorry.
It was Citreon not Peugeot with the chevron gears and badge.
I always seem to get those French companies muddled up.
I think it is the lack of body styles.

Edited by 24gerrard, 28 April 2011 - 19:59.

[gruntguru]

Not with a low ratio worm and wheel you cant.

Correct - that would be a "steep" inclination on the plane Greg and I were talking about.

[cheapracer]

Not with a low ratio worm and wheel you cant.

I'm haven't played with Torsens at all but I believe thats the principle of those? - worm can turn the wheel but the wheel can not turn the worm.

[Tony Matthews]

- worm can turn the wheel but the wheel can not turn the worm.

I've always believed that, but it makes sense that if the angle of the plane is really 'shallow' that the worm could be turned. As to 'slickness' I have a can of 'Vigzol' bevel-box oil that is like black treacle, touch the surface with a fingertip and you can draw out an immensly long 'thread'. It was formulated to stick to simple crownwheel and pinion set-ups, with no other whirling components to fling the lubricant about. Not too good on toast, though.

[24gerrard]

I've always believed that, but it makes sense that if the angle of the plane is really 'shallow' that the worm could be turned. As to 'slickness' I have a can of 'Vigzol' bevel-box oil that is like black treacle, touch the surface with a fingertip and you can draw out an immensly long 'thread'. It was formulated to stick to simple crownwheel and pinion set-ups, with no other whirling components to fling the lubricant about. Not too good on toast, though.

I much prefer Marmite which as a side issue Morrisons sell under their own brand for half the price.

I remember 'Vigzol' and other graphite lubricants.
I used to use them on commercial diffs etc and also on some experimental gearboxes to mark up wear areas. The problem was not knowing how such sticky and thick lubricants would take to getting in contact with conventional oil. Would oilways block, or the film break down?

Edited by 24gerrard, 28 April 2011 - 20:01.

[bigleagueslider]

The diff in Tony Matthew's gearbox sketch is interesting. It's a planetary design, with friction clutch plates and a hydraulic cylinder controlling the degree of slip. Nice job of packaging. I wonder if this design is still used in F1.

[gruntguru]

The diff in Tony Matthew's gearbox sketch is interesting. It's a planetary design, with friction clutch plates and a hydraulic cylinder controlling the degree of slip. Nice job of packaging. I wonder if this design is still used in F1.

Going cross-eyed trying to find everything in the drawing. I can't see the extra gearset to equalise the torques L-R which would be necessary since a planetary diff cannot split 50-50.

Edited by gruntguru, 29 April 2011 - 04:47.

[Tony Matthews]

Un-cross those eyes, gg, before the wind changes...

[24gerrard]

Without the extra gearset, a 'spur gear' diff would have one wheel rotating in the reverse direction.
The torque split left to right can 'only be' 50 50 with both road wheels rotating at the same rpm.
This changes controlably during vehicle changes in direction and uncontrolably when one of the rear tyres loses traction.
In this condition the hydraulic cylinder applies pressure to progresively apply the wet clutch pack joining the left output to the right, reducing the differential effect and increasing tyre wear.
With traction limiting this can be finely controled along with the engine torque applied by using sensors to measure wheel spin and rpm difference.
In a non TC application it applies the clutch by mechanicaly operating the cylinder using a set point of difference between the output shafts rpm.
Clutch bite point and apply force will typicaly be adjusted by varying the clutch plate clearence.
I once designed a diff with two hydraulic pump halves that worked fully hydraulicaly.
Unfortunately it needed cooling.
There were a number of attempts using this method.

Thanks again Tony for yet another stunning (and there is no other word) drawing.

Edited by 24gerrard, 29 April 2011 - 09:31.