2-speed systems (R-D, dog-clutch or derailleur changing)

This topic is for discussing the pros & cons of the different ways of implementing 2-speed transmissions.

Ref:
http://www.endless-sphere.com/forums/viewtopic.php?f=2&t=4930&start=0
http://www.endless-sphere.com/forums/viewtopic.php?f=28&t=7679&start=0
http://www.endless-sphere.com/forums/viewtopic.php?f=28&t=8503&st=0&sk=t&sd=a
http://www.endless-sphere.com/forums/viewtopic.php?f=28&t=8629

You are fast...
So now we have to add primary drive to your gears. I would say that easiest way will be to turn your drive around so exit shaft become enter and also to change one way bearings to that other shaft then, ad shaft with small gear leaning on biggest gear on input shaft. Then connect large pulley to that small gear and connect that large pulley to small pulley direct on the motor. With that we got dual primary reduction and belt drive where it the largest speed to reduce noise a little bit. I think this system can be hidden behind biggest crank sprocket easily. Housing is now must. Also now we have to add freewheel on the crank.

Hal, my head is hurting again, can you do a drawing? I'm really struggling trying to keep up with what you guys are coming up with. Just when I'm starting to get my arms around one notion, however, you guys are on to further refinements/changes. :?

MIles,

I'd like to see this discussion blossom, so I'll give it a shot. It's mandatory for me to have at least a 2 speed due to hills and mountains. It would really be nice to have a 2 speed that worked like a car or motorcycle tranny and you go up and down in gear without issue and use it to take off, but for now I'll live with only very occasional down shifting to climb hills.

I like the simplicity of R-D, though I do worry about changing directions of a brushless motor while moving. ie Do you have to completely stop the motor from spinning first, so BEMF or something doesn't blow your controller? What about the clutch bearings required, are they expensive? Durable? I tried looking for some online once, and quickly realized I didn't have a clue about the specifications needed. You'd definitely want a controller with a reverse to avoid the cost and added complexity and parts of switching motor direction.

Some kind of derailleur rig moving a bike chain back and forth to different sprockets is going to be the easiest and cheapest, so I'm going to try that route first. I'm not delusional and think I can make something that shifts as smoothly as the Phaser does in the video while the guy rides his bike. I expect shifting under load to have issues, and a front type derailleur shifting for cassette size sprockets at the motor may also have issues. A single sprocket front and cassette at the rear is always an option, though I'd like to make it quieter with better chain and sprocket life by maintaining a straight chain using an equal number of sprockets with equal spacing front and rear.

That leaves a 2 speed planetary gearbox using a dog clutch. What kind of efficiency is possible? How easy would it be to make? It seems pretty easy, but in the idea stage they always do. Anyone have any links to drawings of a very simple example?

Too bad the 3 speed hubs don't have sufficient gear range. Nuvinci does, but what's the efficiency hit.

John

I realize the Tomos moped transmission weighs more than many will find acceptable, but I post it here for the sake of discussion. It is a 2 speed automatic that uses centrifugal clutches for both gears.




Power output: 1.8 kW/5200 min-1

Torque: 3.7 Nm/4000 min-1

Gearbox: 2-speed automatic with two centrifugal clutches

Gearbox oil/quantity: ATF Dextron III/220 cc

Transmission ratios:
Gear ratio 1 st speed (66/17) i = 3.8823
Gear ratio 2 nd speed (53/24) i = 2.2083
Gearbox ratio 1 st speed (3.88 * 72/11) i = 25.4117
Gearbox ratio 2 nd speed (2.20 * 72/11) i = 14.4545
Chain transmission ratio (24/26) i = 0.9230
Total transmission ratio 1 st speed (25.41 * 24/26) i = 23.4570
Total transmission ratio 2 nd speed (14.45 * 24/26) i = 13.3426

John in CR said:

MIles,
I like the simplicity of R-D, though I do worry about changing directions of a brushless motor while moving. ie Do you have to completely stop the motor from spinning first, so BEMF or something doesn't blow your controller? What about the clutch bearings required, are they expensive? Durable? I tried looking for some online once, and quickly realized I didn't have a clue about the specifications needed. You'd definitely want a controller with a reverse to avoid the cost and added complexity and parts of switching motor direction.

Click to expand...

As long as your controller can handle the surge then it should be fine. Devices at my work will use reverse polarity as a braking mechanism for brushed motors. I'd probably check out controllers designed for robot combat as they are designed to take high loads, be abused, and whatnot and would have good specifications.

Keep in mind that on a R-D type system the motor will be unloaded as soon as you let off the power or reverse it.. it won't hook back up with the complete drive until the motor is up to speed and going in the reverse direction.

Some kind of derailleur rig moving a bike chain back and forth to different sprockets is going to be the easiest and cheapest, so I'm going to try that route first.

Click to expand...

That's very effective. The only gotchas is that you can't shift when your not moving and you have to do regular maintenance. But if you don't mind that and keep them in shape then they can't be beat in terms of efficiency.

Too bad the 3 speed hubs don't have sufficient gear range. Nuvinci does, but what's the efficiency hit.

Click to expand...

Well there are hubs with more gears. A modern road bike with derailers front and rear would have a gear range of about 520%. A Alfine/Nexus 8-speed hub has about 8 speeds and a range of 350%. (something like that) I wouldn't think they would be much, if any, weaker then the 3-speed versions. For transmission life expectency the thing you have to worry about is shocks to the system... like sudden jerks coming from the motor or binding and stuff like that. If your smooth with the throttle these sorts of things will be able to handle a lot more power then if you were very jerky with it.

wow man, I grew up on Tomos automatic... I might still have one around ...

A bit off-topic, but I didn't know where else to ask this. For the more mechanically-challanged amongst us, can someone briefly explain the differences between a "dog" clutch and a "sprague" clutch, and how they differ from a one-way bearing or freewheel? :?

-- Gary

Not that good with english names but I think that one-way bearing and sprag clutch is basicly the same and they both work as freewheel. Dog clutch freewheels only if one side is pulled in axial direction.

Wikipedia to the rescue:
http://en.wikipedia.org/wiki/Sprag

Or maybe not.

The sprag patent from 1950's...
http://www.google.com/patents?id=PKRIAAAAEBAJ&dq=2777551

Basically you have a bunch of little bars in the clutch that if you rotate it one direction they slide freely, but if you rotate in the other direction they jam up due to the friction. Then they have springs in them that hold them in place and such. A sprag bearing seems to be a combination sprag clutch and roller bearing. They also call them 'backstops'.

There are other types of freewheel mechanisms with different plusses and minuses, but apparently the sprag will give the best performance for a given size/weight. Of course like any other freewheel mechanism or clutch system they will wear out after a while so you'll have to design your system with that in mind.

http://www.renold.com/Products/SpragClutchFreewheels/Sprag_Clutch_Index.asp

this has nice diagrams showing how they work:
http://www.koyousa.com/brochures/pdfs/cat286ex%20-%20Sprag%20Clutches.pdf

Here are some combination roller bearing sprag devices:
http://www.peakbearings.com/one-way-bearing.htm

All this stuff is new to me also, but it's a interesting concept.

Gary,

A dog clutch locks gear to the shaft. The easiest example is the spinning platter in a microwave. The platter is like a gear, and the dog clutch part is the typically 3 sided thing turned by the motor. When the "dog" is properly engaged in the indentations in the platter, the platter turns too.

John

Provisional comparison:

Retro-Direct:

+ Easy change system (motor reversal)
+ Change can be easily automated.
+ Wide ratios possible.

- Difficult to implement as a common system with the pedals.
- Difficult to back-drive for regen.
- Can't go backwards.

Dog-clutch:

+ Quick change.
+ Wide ratios possible.

- Needs custom design.

Derailleur:

+ Lightweight
+ Relatively simple to implement.

- Narrow ratios only.

Comments? Disagreements? Additions?

Perhaps give each scheme a list and a rating (1-5)?

Ease of gear change
Ease of automation
Span of ratio
Integration with pedals
Integration of regen
OTC parts
Strength
Weight
Cost
Size
Other
Other
Other
...

(re clarification of 'dog-clutch': we are referring to schemes that feature constantly engaged gears or pulleys that can be locked or unlocked to a shaft, similar to "synchromesh" transmissions Y/N?)

Good idea.

TylerDurden said:

(re clarification of 'dog-clutch': we are referring to schemes that feature constantly engaged gears or pulleys that can be locked or unlocked to a shaft, similar to "synchromesh" transmissions Y/N?)

Click to expand...

Affirmative, captain....

Could also be chain/sprocket drive system.

Miles,

Do you know of any simple examples of dog clutch rigs that we can use as a starting point? I tried some searching, but got lost. My local machine shop buddies are reasonable enough, that if we come up with something, I'll get it built, but I can't even find detail of how the Sturmey Archer 3 sp works, which I understand has a dog clutch. Before proceeding further in this direction, how big of an efficiency hit are we looking at with a well designed planetary gear transmission? If it gets up into the 5% range, then maybe it's better to just forget it and go for a Nuvinci. Plus an infinitely variable gearbox can probably pick up 1-2% overall on any 2 or 3 speed just by being able to dial in those in between ratios, that is as long as you have real time feedback like with a CA to adjust accordingly.

Also, as far a 2 speeds go, let me throw one more in the mix, that is an automatic transmission. I saw one for scooters, but forgot to bookmark and the name. It's belt driven, and basically the way it works is the low speed belt has a one-way bearing on the wheelside pulley, eg allowing freewheeling. The high speed belt freewheels at low speed by means of a centrifugal clutch. That clutch then engages at a certain speed, and once engaged the low speed side turns faster too just without any torque, because the shaft spins freely in that direction due to the clutch bearing. This seems to me like the idea modification to Matt's RC drive. I know the RC guys think you can just throw power at it and 1 speed is sufficient, but if you combine hills and the desire for speed on the flats, you're either going to need a big heavy motor with lots of batteries to power it, or you're going to stress the motor, which is neither efficient nor durable.

This belted 2sp automatic begs my question, at how low an RPM do centrifugal clutches function? With the Comet Torq-A-Verter I have, the standard springs results in engagement at 2krpm, and the factory rep told me they have alternate springs that engage at 1400rpm, so my guess is that's getting pretty close to the limit, so a significant stage of reduction would be required after such a transmission.

John

John,

I'll try and work out a design, sometime next week.

Not sure about the Nu Vinci..... It would be less efficient than a hub gear and weighs as much as a hub motor.......

John in CR said:

I can't even find detail of how the Sturmey Archer 3 sp works, which I understand has a dog clutch.

Click to expand...

See attached...

I've got a burned up Bafang with pristine gears including one that's metal. Any chance of putting that to good use in a relatively easy manner to do a 2 speed? If not, then I can definitely visualize a hollow axle on what was the wire side, where I attach one of the 90° angle grinder gearboxes I have. That would give me an pretty simple and bulletproof shaft drive including about a 12:1 reduction. I think everything even spins the proper direction to use the existing Bafang internal freewheel. That might be perfect for an over-volted Kollmorgen or maybe even a low Kv RC motor. It sure would be nice to make it a 2 speed as the icing on the cake.

John

Miles said:

John,

I'll try and work out a design, sometime next week.

Not sure about the Nu Vinci.....? It would be less efficient than a hub gear and weigh as much as a hub motor.......

Click to expand...

I would not get to hung up about efficiency at the rear wheel like that. Efficiency is very nice at the motor because any wasted energy ends up as heat, which is bad, and puts heavier loads on the batteries.

On a upright bicycle once you get over a certain speed inefficiencies in aerodynamics far far far far outweigh any efficiency lost through drive train. Hell weight doesn't matter that much either. People who spend hundreds of dollars to loose a few ounces on their bikes are completely delusional.

At low speeds weight and efficiency in the drive train matter becuase they are significant and bicyclists operate under their own power... at higher speeds people are aiming at with electric bikes they become such a small matter that it's laughable to worry about it to much. The Nuvinci's drag is noticable on road bikes and on Velomobiles.. but for this sort of application, upright electric bike, it's not really going to be a huge deal. It may be a big deal if your comparing two things with similar performance and price characteristics.. but I wouldn't worry about it at this point.

Here you can see the difference illustrated in a artificial manner:

Well... the espeed page is down, getting a 404 error, but here is the calculator in the Internet Archive...
http://web.archive.org/web/20080213094200/http://www.kreuzotter.de/english/espeed.htm

It's kinda interesting to play around with, but here is something telling....

So say you have a 750 motor.... Put that into the wattage category. Now compare the MTB to the Racing upright bike...
MTB top speed: 29.1 mph
Racing Bike (upright position, hands on tops) top speed: 30.7 mph

So at these speeds the differences in the weight of the bikes (about 9 pounds) and the much lower rolling resistance offered by the high pressure tires amounts to less then 2mph difference.

Now look at the difference in riding position offers:
Racing Bike (lower position, hands on bottoms) top speed: 34.8mph
Racing Bike (superman style): 40 mph

So a simple change in riding position is worth 2-5 times more then high efficiency tires and much lighter weight.

Now with a Velomobile you have very good aerodynamics... Their top speed with 750 watts is: 69.7
I think at that range the calculator is probably loosing a lot of it's accurancy, but hopefully it's good enough to drive the point home.

With pedal power at low speeds on a Road bike being compared back to back with derailers, or with endurance racing were your trying to get as much energy efficiency to the ground as much as possible then the Nuvinci is going to come up short. But otherwise it's not so much.

-------------------------------------------

I think the priorities here are going to be:

- Reliability (complexity is the enemy of reliability)
- Ease of Manufacture
- Cost
- Ease of Maintenance

Then things like Weight and Efficiency are also very important, but I wouldn't sweat it if it's not too bad. The more power your planning on driving through the system the less relevant... a 1Kw motor it's not so important... a 300 watt motor then it's probably very important.

I'm not hung up on efficiency, I just don't see that the Nu Vinci has enough going for it.

Weight is certainly important to anyone that has to carry a bike up steps or stairs.
Minimising unsuspended weight is certainly important if you want decent suspension.

I posted a useful ebike power calculator here:
http://www.endless-sphere.com/forums/viewtopic.php?f=11&t=538&start=2

I think we understand the relative power needs, by now..... Personally, I have no interest in using power-assist to go at a higher top speed than I'd otherwise go without it - I realise I may be in the minority, though

The evil of unsprung weight in the wheel is thrown around here with the same certainty as the 1st Law of Thermodynamics, but is that really the case? Don't many lightweight motorcycles have arguably good suspension with more unsprung weight than all but the biggest hub motors? It seems that total sprung weight, it's distribution, length of wheelbase, and the planned use of the bike all play a tremendous role in the impact of unsprung weight in the wheel.

I grasp why lighter is better, but without diving all the way into the math involved, are there any rules of thumb or useful basic ratios that can help determine how much weight in the wheel is tolerable?

For example, wouldn't the added weight of the Nuvinci have minimal impact with 25lbs of batteries mounted over the rear wheel, when compared to the effect of that weight with batteries in the triangle?

John

John in CR said:

For example, wouldn't the added weight of the Nuvinci have minimal impact with 25lbs of batteries mounted over the rear wheel, when compared to the effect of that weight with batteries in the triangle?

Click to expand...

There wouldn't be a lot of difference, unless you mean that the batteries are unsprung too......

To take us back on track...

Starting from the spur gear schematic (below):

- Belt drive from motor to green input shaft.

- Substitute synchro pulleys and belt for the yellow gears.

- Substitute sprockets and chain for the bright red gears.

- Dark red gears remain as spur gears (PTFE impregnated, hard-anodised Alu-alloy, for dry running?).

- Grey shaft pivots around axis of green shaft to get independent spacing adjustment.

NB These changes reverse the blue output shaft rotation direction.

Nice workup Miles. I didn't realize such an alloy exists, but now that I know the ideal I'll start checking around. For someone without the knowledge base, such as myself, how do we spec out the minimum sized gears we can get away with?...or do we have to start with the one way bearings and that will determine the gears? Also, what type of thing would use gears of that PTFE impregnated alloy, so I could maybe find it as an off the shelf spare part or a used item that could be cannibalized for its gears?

John

Hi John,

It's only the hard-anodised layer that is impregnated - it's really just a PTFE surface treatment, but quite durable.....

See also: http://www.twi.co.uk/content/faqmdfh004.html