How many computer programmers, goes the joke, does it take to change a light bulb? Can't be done: it's a hardware problem. That touches on two ultimately related things: the remarkable variety of information cultures among different automotive enthusiast communities, the way, say, early Ford fans sit on a far more voluminous base of far more detailed information than, say, late-model Mercedes-Benz fans do; and the qualitative difference in the thinking of a generation raised on ubiquitous computing, compared to a generation grounded in physics only. Between them these two factors bear very heavily on the amount and kind of information an internet search is going to turn up on any given automotive subject. In short, information which exists in communities of older people about older stuff comes more from having taken things apart and measured them. Information which exists in younger communities around newer stuff derives more from circulation of digital code: maps, command files, look-up tables, etc. The difference is coming to be reflected in approaches to life: one seeks answers to questions which might never have been asked before they themselves asked them; the other seeks answers to questions posed by a puzzle-setter, and struggles to conceive of there being no puzzle-setter in any given situation. Being aware of that I was surprised that there is as much information about ZF automatic transmissions as there is out there, especially about the hardware — and equally unsurprised that the information doesn't go to the level of detail I want it to. This was yet another exercise in my ongoing quest for the perfect gearbox. I've been fascinated by some of the weirder gearboxes of the '30s for some time. The Wilson and Cotal preselector gearboxes were both based on epicyclic gear trains, and in the process of recapping I discovered that so was the de Normanville Safety Gearbox, made by the overdrive people and of which it is believed six remain in the world. It is therefore to be expected that my perfect gearbox would tend towards an epicyclic manual transmission with a torque converter and a clutch pedal. I had considered an overdrive as a splitter behind a manualized classic American three-speed automatic, but a ratio span of ±3.5 is really still too narrow. I had considered a 4-speed installed backwards behind a manualized automatic as a divorced triple overdrive, but that weight adds up. Considering a ZF, losing the pan-mounted TCU and feeding the shift solenoids directly from a kind of rotary switch was an interim step: but if what I'm after is an epicyclic manual, why not lose the valve body ("mechatronic" in ZFese) entirely? A 6HP would suit me because they're relatively common in my part of the world, probably much more common than '60s/'70s/'80s American stuff. Beefing-up strategies are known, and high-performance torque converters exist. A lot of that good stuff seems to come out of Australia. It'd give six relatively close ratios plus converter multiplication, and a ratio span around 6. But nobody is going to do with one exactly what I want done — or even give me the information I want. As I said, I was surprised at what I could find: ... but zilch about how long the thing is, even approximately. Now this is useful. Considering how close these holes are to the clutch cavities, and the absence of any other hydraulic signal points, I can only conclude that the clutches exhaust pressure back down these holes. Whatever I do would have to switch between line pressure and exhaust. Still no dimensions, though; nothing but a wild guess at scale. I know what I want the ergonomics to do. Without a computer, shift quality would be controlled by your left foot via a "clutch" valve. The way I drive I often find myself barely brushing the clutch pedal on upshifts. With an epicyclic gear train that could make for extremely fast, unmissable manual shifts. It would also allow for downshifting of which Piero Taruffi would have approved. The clutch would obviously be redundant for starting from a standstill, due to the torque converter. The converter would lock up with the 2-3 shift. A ratchet shifter shifting through P-R-N-1-2-3-4-5-6 becomes practical if there is a sideways degree of freedom available in 3, 4, 5, and 6, which shifts straight back to 2. Back for upshifts and forwards for downshifts feels intuitive for me. I keep wanting to load the shifter towards my body, so the sideways "2" shift would be away from my body. It occurred to me that all this could be done with a line pressure regulator, a "clutch" valve of some kind, and a barrel valve with a degree of axial motion. No need for any part of the stock "mechatronic". Lacking an actual transmission to tear into, or even any hard information about dimensions and fluid displacement volumes, I reached for my old stand-by of 3D modelling. Unfortunately it looks far more definite and worked-out than it is. It is in fact completely notional. Though it is probable that the complex shapes would lend themselves to some kind of 3D printing:
Sounds like an updated version of the old B&M Clutchflite. Also affectionately known as the Clutchbomb!
This would be the opposite, or complementary counterpart if you like, of a Clutchflite. This would be a manual with a torque converter rather than an automatic with a clutch.
Not sure if that barrel valve is right, come to think of it, given my unhealthy preoccupation with bistability. The barrel valve is a kind of One Spool Valve to Rule Them All, and as such it would have radial clearance which would render it slightly leaky. Something involving poppet valves actuated by cams would suit me better, and might even be easier to make. That would allow the gearbox to be left in a gear, and the poppet valves to hold pressure in the relevant clutches. I know that this obsession doesn't really make any sense. Of course nobody has had any reason to post how much ATF gets displaced applying any of the five clutches in a 6HP online. My closest educated guess is around 30cc, but I could be way off in either direction.
Ned; There's a saying about "thinking outta the box", w/you, there isn't a box in sight, nor for that matter, not even a box-factory... . In fact, almost on another planet... . So, that's why I like reading your writings, although this one's way over my head. So I'm just reading n watching. Thanks for the mind-stretch. . Marcus...
I had a go at the poppet valve option: This is more of a model engineering or clockmaking exercise than the previous one. It's also more compact. Though for some reason this was something like a 27-hour render. There are seven compound poppet valves, each with two spring-loaded elements arranged so that the pressure supply element closes before the exhaust element opens. The valves are of two kinds: clutches B and E, brake D, and torque converter clutch apply have push-to-open valves. Clutch A, brake C, and torque converter clutch release have push-to-close valves. The valves are not in alphabetical order but arranged for the most direct connection to the respective pressure ports on the underside of the transmission case. The valves are operated by a camshaft driven by a bidirectional ratchet mechanism with a rod lever to the shifter via a pair of gear wheels. The camshaft assembly can pivot away from the valves but is kept in position by springs. A lever connected by a second rod linkage to the shifter pivots the camshaft so that the gear wheels are out of engagement when the shifter is moved sideways. The pinion-and-sector mechanism at the bottom of the illustration causes the cam to rotate to the 2nd gear position. Thus moving the camshaft away from the valves closes B, D, E, and the TCC apply, and opens A, C, and the TCC release — which is the shift logic for 2nd gear. When the shifter returns to its default position, the camshaft moves back onto the valves, only now in the orientation for 2nd gear. Because the poppet valves close tight, the transmission is pretty much bistable in any gear. As this means that the vehicle may be left in 3rd gear when parked, as you would leave a manual gearbox in gear, as a backup for the handbrake. That would obviate the need for a Park function in the transmission.
That is a whole lot of poppet valves, springs, and cams that can get out of time, or have a mechanical failure pretty fast. I can understand the desire to have a mechanical function, but I think electronic switching and electronic solenoids would be more reliable then your current mechanical design. A rotating cam triggering electronic switches carries much less load, can trigger instantly, and can have much lighter weight materials for faster, easier movement of the fluid. Think pressurized transmission fluid, through electronic switch activated port injectors, similarly to how electronic fuel injection operates. Just a thought from a guy that worked in factory maintenance in a factory that had a lot of mechanical poppets, springs and cams that failed often, as they grew older.
Nedd's got the kind of genius mind that could conjure up a computer using mousetraps powered by a water wheel. I have trouble keeping up!
&, it's *never* boring in his "classes". . You either think, follow along, or get a headache. . Marcus...
Thinking about this some more, and learning a lot in the process. All automatic transmission valve bodies, as far as I know, use multiple hydraulic valves in series, and the obvious reason for that is that it allows all the control to be done in low-pressure, low-flow pilot circuits before finally activating the valves which feed actual line pressure to the clutches and brakes. A shift in a typical automatic transmission might involve three or four spool valves activating each other. That means I've still got some scope for gratuitously adding spool valves and still ending up with something significantly simpler than a stock valve body! I looked at poppet valves in my quest for bistability, but that can be achieved by applying line pressure to each clutch/brake/TCC state via a check valve and exhausting it via a hydraulically activated poppet valve, a net simplification. That would mean, if you remove all external feed pressure, the clutch/brake/TCC would stay exactly as you left it, for months or even years. Each of the six control assemblies would comprise a spool valve feeding line pressure either to a check valve or to the activation end of a poppet valve, except the TCC control assembly, which includes two poppet valves and two check valves. The spool valves are actuated by a barrel valve or rotary spool valve as I 3D-modelled for Post #1, but it can be made much smaller as it would only need to flow enough to activate the spool valves. There are only six control assemblies because the TCC control has only two states which alternate, and so can have the same spool valve controlling both the supply and release circuits. All six spool valves are loaded by light spring pressure in the apply position. This means that, if the pilot circuit leaks down, all open poppet valves would close. Because the pilot circuit would be fed via a pressure regulator from a line pressure reservoir or accumulator, this would only happen once line pressure itself has significantly leaked down. Here is a control assembly, as for all the clutches and Brake C: Applying pressure from the barrel valve causes the lower piston in the spool valve to push the upper piston up, cutting off line pressure to the clutch or brake, and opening a poppet valve which exhausts the circuit, releasing the clutch or brake: I took a day's daydreaming allocation to consider a literal, physical, mechanical clutch independent of the transmission hydraulics. A clutch behind the transmission would work in theory but it would need to hold 3000lb.ft or more, and if it was going to be a little thing that would fit in a transmission tunnel it wasn't happening. A clutch between the engine and the torque converter is not unprecedented. Mopar tried something like that in 1939 with their Fluid Drive. Practically it promised actuation and rotational inertia problems. It turned out there is a way to achieve the desired results using the transmission hydraulics. The "clutch" pedal was inspired by the dirt-track trick of using a ball valve to dump line pressure when using e.g. a Powerglide without a torque converter. That wouldn't be quite practical here, as it would be necessary to void circuits selectively, which implies several separate poppet valves again controlled by a pilot circuit: so why not use the same poppet valves? To this end the spool valves have two pistons to create a chamber between connected to a master cylinder and pedal, just like a conventional hydraulic clutch. Pressing the "clutch" pedal pushes the two pistons apart, activating the poppet valve: This happens regardless of whether the barrel valve has the spool in the clutch-apply or clutch-release position, simply by making allowance for a bit of redundant stroke: This is so that "clutch" pedal pressure remains the same regardless of which gear is engaged. According to the 6HP's clutch logic, only Brake D is applied in Neutral. That makes it possible for the Brake D control assembly to be simplified: As noted above, the TCC can be controlled by a single spool valve, unaffected by the "clutch" pedal: That was another research frustration. Sources will tell you how a lock-up torque converter works in theory, or they will tell you how to service a specific sort in detail. Nobody tells you how a lock-up torque converter works in detail. It took a deep dive for me to learn that the way the lock-up clutch actuates is a lot simpler than I thought. I was wondering if there was some kind of hydraulic cylinder somewhere, and couldn't see anything on published technical drawings that explained it satisfactorily, until I realised that the entire torque converter is the cylinder. It's the fluid pressure inside the converter which applies the TCC. Ordinarily, fluid is pumped by the pump, down the stator support into the gap between the front face of the converter and the TCC plate. It flows around the edges of the TCC plate and into the main body of the converter, where it does its torque transmitting and multiplying thing. Pressure equalizes both sides of the TCC plate. Fluid is bled out of the converter, through the cooler, back to the pump. To apply the TCC, the flow is reversed: fluid is drawn out of the gap between the converter body and the TCC plate, the pressure there drops, and the pressure in the main converter body pushes the plate against the converter body with considerable force. It's actually extremely simple! but they don't tell you that. It turns out, moreover, that there are 2-path, 3-path, and 4-path TCCs. Finding out what kind any given converter has isn't straightforward. I eventually learned that there were versions of the 6HP which used 2-path TCCs and versions which used 3-path TCCs. As far as I could gather, 2-path TCCs cut off fluid feed from the converter to the cooler when the TCC is applied, and the idea behind the third path is to cool the fluid during the partial TCC engagement conditions a lot of TCUs are asking for. As I'm contemplating a simple on-off TCC engagement with either the 1-2 or 2-3 shift, that wouldn't be an issue. I was originally thinking 2-3, but there might be a use for having the TCC engaged in 2nd. I can't decide.
It occurred to me that the reason Brake D is applied in Neutral is to prevent the kind of planetary overspeeding issues which cause unwisely set up Torqueflites to explode. I can't imagine that ZF would rely on electronics to prevent the driver from revving the engine in Neutral, when applying Brake D achieves the same thing more easily and less intrusively. As Brake D is released in gears 2-6, it would in that case be prudent to have the "clutch" pedal apply Brake D while releasing everything else. The tandem piston arrangement in the previous post won't work for this, so I tried a concentric sleeve arrangement with a single piston: It responds to pilot pressure exactly as before: Depressing the "clutch" pedal moves the sleeve up, but this has no effect as long as the piston is in the apply position: If the piston is in the release position, depressing the "clutch" pedal moves the sleeve to keep the apply circuit pressurized: It is possible to control the three clutches and Brake C by the same arrangement, except that the "clutch" pedal moves the sleeve down:
The Jalopy Journal
