DIY 3-speed transmission

[Dauntless]

Back in the mid 80s when racing BMX i tried a 'Torpedo' rear hub was a automatic two speed dunno if you can find them still?
Was a coaster brake, was no good for me my little legs couldnt push the 2nd gear. Gave it to the side hack rider i swang for he loved it.

KiM

Ah yes, the Duomatic. Not to be confused with the Torpedo Boy coaster brake, but I did so when I came into possession of one. Never did figure out why it got so hard to pedal sometimes, or I should say I didn't the whole time I had it. Would have kept it if I knew.

In fact, wish I still had it to take apart and figure out how to design my own. Every beach cruiser should have one.

You needed a larger sprocket on your Torpedo, so you had both gears a bit lower.

http://www.3gang.de/3-gang/duomatic_expl.html

 

[bzhwindtalker]

You can get them from sturmey archer, they are caled kick back hubs. I had one on a city racing bike and I loved how simple the bike was with no cables, no brakes and no dérailleur. The coaster brake was a bit weak and leaked some grease, also the braking force was dependent on the speed you where in, wich ils always confusing.

 

[speedmd]

Great project. Love to see how light you can build it and how much power the drive will take. Love the cascading free wheel concept. Would be good if you could release the pull on the springs slightly to induce the engagement when you are nearing the automatic shift points. This way you can feather the throttle slightly simultaneously at manually induced shifts rather than letting them just bang up through the gears in auto and spill your coffee on the way to work. I think it would be called a "sportamatic".

$80 full retail will buy a Shimano Nexus 3 coaster hub

I have been considering such hubs for possible options on lower power builds. Would like something as compact and strong as possible and without the brake.

 

[turbo1889]

. . . Would be good if you could release the pull on the springs slightly to induce the engagement when you are nearing the automatic shift points. This way you can feather the throttle slightly simultaneously at manually induced shifts rather than letting them just bang up through the gears in auto and spill your coffee on the way to work. I think it would be called a "sportamatic". . . .

Figured out how to do that automatically. Adding a second layer inertial disk to the mechanism makes it so that under positive acceleration the tension is increased delaying the shift. So basically once your ready to shift you just back off the throttle (and/or pedaling) for a second and then it auto shifts:



This one has all the previous improvements (including the cam effect on the spring tension so that the tension is 30% higher holding the pawls in then when they are in the open position to ensure they "Pop!" open and closed and don't rattle in-between the two positions) and adds the floating disk to add inertial effect to the shifting so now the mechanism isn't just measuring RPM to know when to engage (and thus shift) but also takes into account acceleration and prefers to shift when not under heavy acceleration. Also the spring is now a torsional coil spring wound around the jack-shaft which allows the shaft clearance all the way through to allow multiple stacking instead of just two on the outer edges for three speeds max. The floating disk with the linkage rods to the pawls and the spring should be made of a very dense heavy metal, probably bronze or something along those lines to maximize the inertial effect on the shifting mechanism to delay the shift under heavy acceleration so it auto shifts so you can run it smooth by backing off for a moment at the shift points.

 

[fizzit]

Wow, that is really cool. The inertia disc is a great idea. I would be curious to see how well it downshifts if you slow down on a hill or something like that.

Edit: What exactly is the coiled spring used for? To pull in the pawls?

Also, wouldn't it be difficult to get both pawls to always engage simultaneously? It seems you'd have to control the pawl mass and the spring rate very precisely for that to happen.

I don't fully understand the design but I want to suggest an improvement anyways What if there is a disc, next to the disc with the pawls, and it is coupled directly to the shaft. The disc with the pawls is on a carrier bearing on the shaft. There are strong springs connecting the pawls to the disc that is coupled to the shaft, so that when the shaft is experiencing high torque, the pawls are pulled inwards to prevent engagement. But to keep the springs from stretching too far, the discs would be in some way loosely coupled, like a slot in one and a protrusion in the other, so that the shaft-coupled disc could not rotate more than a few degrees past the pawl disc when the motor was pulling hard.

 

[turbo1889]

. . . Edit: What exactly is the coiled spring used for? To pull in the pawls?

Also, wouldn't it be difficult to get both pawls to always engage simultaneously? It seems you'd have to control the pawl mass and the spring rate very precisely for that to happen. . . .

It all starts with the very basic and simple centrifugal clutch - like used on go-karts. Basic principles of such is a set of shoes mounted on a rotating disk inside a hollow drum. Spring force keeps the shoes tight towards the inside of the spinning disk until the RPM are high enough for the centrifugal force due to the spinning to over-come the spring force and the shoes make contact with the inside of the drum. For use in a bicycle transmission as a means of shifting up along a sereis of gears in a cascading manner this type of clutch in its simplest form has four main issues:

First, normally engagement RPM is usually in the thousands, usually about 2,000-RPM or so.
Second, engagement has a lot of slip in it until RPMs are double or more the engagement speed.
Third, disengagement occurs whenever RPMs drop below the threshold engagement whether the clutch is under power or not and is more likely to slip when under power; for a bicycle with pedals this can injure the rider if he/she is pedaling hard when the down-shift occurs and the pedal the rider is pressing hard down into suddenly gives. I've experience that a few times when I've blown out a worn free-wheel pedaling on a steep ascent. Twisted my ankle really good once.
Fourth, once the clutch engages it engages solid and won't freewheel in one direction, which if your going to set up a series of cascading clutches for more then two gears you do need. Possible to over-come by using a separate clutch and a freewheel assembly in series but doing so adds weight and complexity.

My idea was to use the same principles and build a hybrid unit that is half freewheel and half centrifugal clutch. Thus my first initial design:

"How it works" is very simple and clear to see on that one, centrifugal force tries to pull the pawls out to engage the freewheeling action (so it drives on direction and freewheels the other way) but the spring tries to pull them in so it spins freely in both directions. Once the RPMs are high enough the centrifugal force overcomes the spring force. A link bar ensures the two pawls open and close together. Once the pawls open and catch they won't release and close back up until there is a let-off in the torque being applied. Basically the pawls "torque lock" so you can't get "dumped out" by a sudden down shift into a lower gear when pedaling hard. Simple and direct but a clear problems was pointed out to me with it, namely that when the pawls are just starting to open they could float only slightly open and the tips of the pawls not fully engage but only grind against the edges of the mating notches in the outer ring. The solution to this was to relocated the spring connection points on the pawls so that a cam effect was achieved so that the spring force was strongest when the pawls where in their closed position and as soon as they start to open there is a let off effect on the torque due to the acting torque arm shortening as the pawl rotates into the open position. Basically this makes it so that once the centrifugal force becomes strong enough for the pawl to start to open it opens all the way right away and in the open position once you let up for a moment and allow the downshift to take place the pawls close all the way right away. Basically ensuring a clean shift without grinding both up and down:



Once again fairly easy to see "how it works". You do have to look a little more closely to see the cam effect on the spring tension and how the acting torque arm shortens as the pawls open. Look closely at the distance between center of the pivot pin on each pawl (the center of rotation) and the acting line of the spring. Notice how when the pawl is open the center line of the spring moves closer to center of the pawl pivot which shortens the torque arm (to about 70% of its length when the pawls are closed) which correspondingly reduces the torque force of the spring pulling the pawl closed against the centrifugal force. Once again though a problem was pointed out to me. Namely that there is no mechanism to prevent a hard sudden up-shift. Due to the way it "torque locks" a sudden down shift is taken care of but there isn't any mechanism to prevent a sudden up-shift. A sudden up-shift won't injure a rider who is pedaling hard like a sudden downshift can do but under motor power it could "spill your coffee" as you say (not a big problem under pedal power only but with motor power could be an issue). Torque locking won't work because that only works in one direction. It took a while of thought but I finally came up with the idea of using a inertia disk to help with the sudden up-shift problem. Inertia is effected by acceleration so you can "damp" the up-shift using rotational inertia. It can still "hard shift" once you top out the motor and you are no longer accelerating but it does delay the up-shift during hard acceleration which is where you can "spill the coffee" so to speak:



"How it works" is a whole lot harder to quickly see and understand with this one but everything from the previous designs is there just in a different form. Instead of a connecting link between the two pawls to ensure they open and close together they are both connected to the inertia disk and in order for the pawls to open or close the disk must rotate very slightly (about 1.6 degrees). The disk rotates clockwise for the pawls to open and counter-clockwise for them to close. The disk itself serves to connect the two pawls to ensure they both open and close together and the spring is also connected to the disk rather then directly to the pawls so the force of the spring applies force to rotate the disk counter-clockwise which pulls the pawls closed. Under positive acceleration in the clockwise direction the inertia of the disk provides additional force in the same direction as the spring which serves to delay the pawls from opening under heavy acceleration and thus delay an up-shift. The reason the disk is not directly connected to the pawls pivot point but rather a set of two links is used is to provide the same "let off cam effect" as before so that the pawls "Pop!" open or closed and you get a clean shift without any grinding or chatter on the pawls.

. . . I don't fully understand the design but I want to suggest an improvement anyways What if there is a disc, next to the disc with the pawls, and it is coupled directly to the shaft. The disc with the pawls is on a carrier bearing on the shaft. There are strong springs connecting the pawls to the disc that is coupled to the shaft, so that when the shaft is experiencing high torque, the pawls are pulled inwards to prevent engagement. But to keep the springs from stretching too far, the discs would be in some way loosely coupled, like a slot in one and a protrusion in the other, so that the shaft-coupled disc could not rotate more than a few degrees past the pawl disc when the motor was pulling hard.

That's not a bad though, using the disk the pawls are mounted on as the inertia disk itself rather then having a second disk. Not sure if it could be made to work that way but I'll ponder a bit on it. Thanks.

 

[fizzit]

Ah, I see that the spring connects the two discs. Never mind, I think your original plan is better than my suggestion. Although I am still a little confused. The light gray disk is the inertia disk, so when the pawl disk speeds up counterclockwise, wouldn't that push on the connecting rods and push the pawls outward?

 

[turbo1889]

I believe the source of your confusion is in that I drew it as the outer ring being the driver (sprocket attached to outside of outer ring not shown) and the inner shaft is the driven (connects to rear wheel via fixed drive arrangement such that RPM of rear wheel = RPM of keyed center shaft). Rotation is clockwise (as if you were looking at the bike from the right side with standard chain on right arrangement).

Realize this sort of controlled engagement freewheel is for the 2nd and higher gears. First gear (stacked side by side in the tranny with common input and output shafts with paired sprockets with chain loops - see the pictures in the OP of this thread) uses just a standard freewheel such as those used on single speed bikes (outer sprocket is the driver and the inner shaft is the driven). Once 1st gear gets the center driven shaft spinning fast enough it engages the 2nd centrifugal freewheel which drives the output center shaft at a faster gear ratio then 1st gear and the slower 1st gear just freewheels. With more then two gears the process continues cascading through the gears as speed increases and the output center shaft speed gets higher and higher it engages centrifugal freewheels on higher and higher gears that are equipped with stronger and stronger springs holding their pawls in and as each higher gear engages the lower gears continue to move but due to their lower gear ratio they just freewheel with the center shaft running faster then the outer sprocket ring and the pawls on those lower gears just going - "click, click, click . . ." just like on a regular bicycle freewheel.

 

[speedmd]

Hi Turbo

Looks like you worked out some challenging design elements. Good work. Look forward to seeing how the prototypes come out.

 

[Thud]

I've been looking for an exploded diagram of Husqvarna's old 3 speed auto transmissions just for a reference. It had a nice centrifugal mechanism that selected a gear based on rear wheel/out put shaft speed. It was driven through a centrifugal off the crank. I've ridden a couple of the old 430cc auto's. kinda weird without any engine braking...very bicycle like :lol:

Turbo, are you planning to prototype this for kicks? Or just enjoying the design aspect?

 

[fizzit]

Oh wow. I completely forgot that the outer ring would drive the pawls.
2 things: It seems to me that since the inertia disc relies on acceleration to impede the engagement of the pawls, you could still have hard up-shifts while going up a hill and accelerating slowly. Also, wouldn't quick deceleration of the jackshaft cause the inertia disc to push the pawls out, potentially engaging the pawls if the outer ring was moving slowly?

 

[turbo1889]

. . . Turbo, are you planning to prototype this for kicks? Or just enjoying the design aspect?

That mainly depends upon whether the project becomes important enough to me for me to actually build it. I found a long time ago that it is a whole lot easier to change things on the drawing board then after things have already been cut and welded together. About a third of my "mental projects" actually get put down on paper or more often now into CAD, about a third of those I find interesting enough to post on forums like this one in order to bounce the idea of more minds then my own and get changes that need to be made pointed out by others that I hadn't seen myself. And then about a third of those actually get built in real life. At every bottle neck culling point my interest level, need level (as something I actually need for my life not just cool factor), and "cool factor" determine what gets dropped and what continues forward. This idea made it through the first two culling down points, we shall see if it makes it through the next one.

Yes, I enjoy the design aspect but I also am practical and no when to drop an idea and when to continue forward to actually building it, usually stuff that fills a need application is the stuff that actually gets built, especially if it is to be built sooner rather then later. This particular idea isn't entirely "for kicks", one of the next builds I have in mind to fill a "need" category for me is an urban run-about light enough, nimble enough, and fast enough to play well in in-town stop and go traffic fully taking the lane like you would ride a motorcycle with light cargo capabilities to haul a little bit of stuff, about a back-pack worth without having to put the pack on my back. Shifting in stop and go traffic is a PITA especially if you have a lot of gears and I've got to use at least one jack-shaft anyway to combine the motor and pedal power the way I want so two jack-shafts forming an automatic transmission between them with just a few wide spaced gears sounds about right especially if it allows me to eliminate the rear derailer hanging down where it can get damaged and leave a guy stranded and tuck everything up inside the frame where I can slab it over on both sides with aluminum sheet to keep the elements, vandals, "funny guys", and sticky fingers out. Might end up using manual shift gears but looking into my options about the same time I saw this thread and came up with an idea about how to build an auto-tranny.

. . .

. . . I don't fully understand the design but I want to suggest an improvement anyways What if there is a disc, next to the disc with the pawls, and it is coupled directly to the shaft. The disc with the pawls is on a carrier bearing on the shaft. There are strong springs connecting the pawls to the disc that is coupled to the shaft, so that when the shaft is experiencing high torque, the pawls are pulled inwards to prevent engagement. But to keep the springs from stretching too far, the discs would be in some way loosely coupled, like a slot in one and a protrusion in the other, so that the shaft-coupled disc could not rotate more than a few degrees past the pawl disc when the motor was pulling hard.

That's not a bad though, using the disk the pawls are mounted on as the inertia disk itself rather then having a second disk. Not sure if it could be made to work that way but I'll ponder a bit on it. Thanks.

Okay, this be the results of me pondering upon your suggestion:



The disk that the pawls themselves are mounted on and the pawls themselves together serve as the inertia disk. In addition it is nearly foolproof that both pawls work together in-sinc with each other. Also the strongest design I think so far since all the force on the pawls is directly braced against the driven shaft on the back side of the pawls instead of on their pivot pins so it takes the stress of their pivot pins. Main thing I don't like is that the position of the springs subjects them to centrifugal force themselves so the their centers can be bowed out towards the outside, not sure of how big of a problem that could be.

. . . 2 things: It seems to me that since the inertia disc relies on acceleration to impede the engagement of the pawls, you could still have hard up-shifts while going up a hill and accelerating slowly. Also, wouldn't quick deceleration of the jackshaft cause the inertia disc to push the pawls out, potentially engaging the pawls if the outer ring was moving slowly?

Correct, and sort of correct. Using an inertia disk only "calms things down a little" for the potential of hard up-shifts it doesn't eliminate the possibly. On a pedal only bicycle I wouldn't even be worried about hard up-shifts. With the SRAM 2-speed auto hub on the market right now every automatic up-shift is a hard shift and only the down-shift is a soft shift that will only happen when you back off the pedals for a second (I've ridden a bike equipped with one, not my bike but had the opportunity to try it out to see how that hub works out in practice). Reason for this is because a sudden unexpected up-shift is not a problem for the human engine, a sudden unexpected down-shift, however, can damage the human engine. With a hybrid powered bike that combines both human and another power source things are a little more complicated. You still must ensure that there are no unexpected down-shifts in order to not damage the human engine but for the sake of the non-human power plant and a quality riding experience something probably needs to be done to calm down the up-shifts as well. So far "damping" them a little using rotational inertia is the best I've come up with.

Up-shifts at the wrong time during quick deceleration really isn't a problem because quick deceleration means braking and since the tranny re-sets itself and "floats into the correct gear based on rear wheel RPM" every time you let off on the power (pedaling or otherwise) that incorrect upshift should be canceled out during that re-set as soon as you let off the brake and go back to putting power back in.

I've been looking for an exploded diagram of Husqvarna's old 3 speed auto transmissions just for a reference. It had a nice centrifugal mechanism that selected a gear based on rear wheel/out put shaft speed. It was driven through a centrifugal off the crank. I've ridden a couple of the old 430cc auto's. kinda weird without any engine braking...very bicycle like :lol: . . .

If you ever get a good diagram or some good photos of how that tranny works pass them along to me as well please!

 

[turbo1889]

Oh, yah, forgot to mention. As far as the RPMs in question with the gearing combinations I've got in the back of my mind about 180-RPM for the up-shift from 1st to 2nd and then about 270-RPM for the up-shift from 2nd to 3rd.