My old idea, "Overdriven Driveline" (q.v.) was duly fishboned, but nobody seemed to hit upon the true reason it was fishboneworthy. Oil drag, being a fluid drag, might reasonably be expected to vary with the square of the rotational speed involved. If one considers that for all intents and purposes the
power loss associated with an all-roller-bearing gearbox derives entirely from oil drag, we can therefore expect that power loss to vary with the *cube* of the relevant rotational speed.
[kevinthenerd]'s idea, "A New Chevy Nova" (q.v. also) got me to thinking about that. It seems that the requisite upgradability would ideally (i.e. without recourse to swopping gearboxes) require a gearbox equally capable of handling huge torque inputs and offering very little resistance to low-powered engines. Is that possible?
The thesis here is that there is an optimum ratio of engine and gearbox speeds where the effects of oil drag and the necessity of making the gearbox stronger and heavier to accommodate the multiplication of torque are balanced; and that that optimum will probably be found when the gearbox is turning more slowly in relation to the engine than would normally be the case.
Simply scaling a twin-shaft constant-mesh gearbox up by the requisite ratio might produce a gearbox that is far too bulky and heavy for use in anything that is supposed to be light and low-powered, however hardy also. It is, however, well known that epicyclic gear trains as found in automatic transmissions are capable of handling remarkable levels of torque, compared to conventional manual trains of comparable bulk.
The twin considerations of avoiding unnecessary proliferation of gearing and suitability for use with some of my prior suspension ideas suggests a transverse gearbox. So: a drive shaft from the engine and clutch drives a hypoid-bevel primary reduction gear, taking the speed down to, say, half engine speed. A hollow shaft from the ring gear takes the drive to a compound planetary as found in, say, a THM350, which gives three speeds and reverse. The output therefrom is also hollow, and drives a simple two-speed splitter planetary on the other side of the ring gear, through the hollow input. The output from the splitter is directly coupled to a differential, one output of which runs down the middle of the input/output concentric shafts, to the other end of the 'box.
So, we have a transverse barrel-shaped gearbox, with an input in the middle of the length and outputs at the ends. Different combinations of the available ratios allows six forward speeds and (potentially) two reverse. Final drive is by chains, the best of which, also, are capable of handling much more power than is commonly appreciated.
Now, the planetaries are somewhat larger in diameter than those found in the typical automatic. Because clutches and bands and their attendant hydraulic systems are dispensed with, the entire assembly is much more compact than an automatic, especially in length. Instead, the necessary couplings are effected by what are essentially large-diameter synchromesh mechanisms, controlled by multiple, motorbike-style axial cams (each controlling both planetaries: there are several only to obviate the need for shift forks to be heavy in order to handle the large diameters). A result is a motorbike-style quadrant shift pattern, which would require a reverse lock-out and warnings.
Of course, by adding perfidious electronickery and electric servos an efficient automatic may be had for a while. This may be important to those of you who demand gratuitous microchips.-- Ned_Ludd,
Dec 14 2007
Admittedly the aim here is somewhat different to that of the mainstream motor industry: I was very much thinking Wildebeest Not Chevrolet gearbox. But, given a finely-judged weight-to-drag compromise, this could also be of use in longitudinal-car-engined motorcycles and single-rear-wheel trikes, and all kinds of weird and worderful home-built contraptions.-- Ned_Ludd,
Dec 14 2007
I'm going to have to read this again when I have a pencil and paper handy, but I have two questions already. The first is your startng out with bevel/hypoid gears. Bevels are inefficient and very loose. They are usually a necessary evil, but I think having them right at power takeoff, I think would cause problems. The first of which is that by gearing down x2 at the onset to lower oil drag, you are forcing everything to accept twice the torque. The looseness of a bevel would give you a scary gear slap everytime you lifted throttle. I realize this is always (OK almost always, some transverse trans' don't have any bevels) in there anyway but having that "clunk" right at the top might be an issue.
The second point is the use of chains, which has been tried a lot and usually has issues. The two cases that come to mind are SAABs (900 era) and NP203 vs NP205 transfer cases. Chains are also loose, hate reversing torque direction and are large in high torque environments. SAAB had a 2" wide chain right at the engine with a couple of hydraulic tensioners to absorb shock load to handle the torque from a 2L motor. I think it went to 3" for the turbo. The NP203 & NP205 were transfer cases from the 70s. The 205 was all gears and legendary for strength. The 203 had a BIG chain (and a diff) and as I remember, it had issues with the chain breaking.-- MisterQED,
Dec 14 2007
[MisterQED] I don't know where you got the idea that bevels are loose and inefficient, 'cos it ain't so. I can't find a fundamental difference between the efficiency of a spiral bevel gear and that of a helical gear. Gear lash can be controlled equally well. However, they are very sensitive to misalignment; you've got to mount both members of the pair accurately.
[Ned], epicyclic geartrains are relatively inefficient. Driving three of them in series: compound planetary - horrendous efficiency - two-speed splitter and output differential (effectively an epicyclic laid out differently) will give poor overall efficiency, not to mention high cost. Hypoids are not efficient either, as the degree of sliding and hence heat generation is directly driven by the degree of offset of the pinion and wheel axes.
The box would work, but it would be complex, expensive and inefficient. Please don't take this as a personal attack; it's not intended as such. It's just that on this occasion I don't agree with the premise of the idea :(
I think that the point I disagree with is that the heat generation is primarily due to oil drag. I think you'll find that a lot of the heat that gets into the oil is caused by the oil being squeezed through the gear mesh; it is in effect acting as a coolant to prevent the gears overheating from sliding friction as well as a lubricant. Bearings can be considered in a similar manner. The upshot of this is that more gears = more heat.-- david_scothern,
Dec 14 2007
I have to agree with [david scothern] on the planetary efficiency - they can handle enormous torque loads because there are so many more points of engagement. The downside is that each point of engagement is another point of friction.
I'm totally with you on the surprising capacity of planetaries and chains, though. The original Toronado, all 5800lbs and 340hp of it, was driven through a chain from the engine to the trans.
Maybe the most efficient transmission would a really big derailleur system. 21 speeds in a car would probably obviate the need for a CVT.-- elhigh,
Dec 14 2007