Process of development of electronic shifting
#1
Senior Member
Thread Starter
Process of development of electronic shifting
Was imagining when companies like Shimano, Campagnolo, and SRAM decided to make electronic shifting systems. They could have hired electronics engineers or consulted with companies that already made them, or both.. Possibly the proper consulting companies would be those already making products using stepper motors like used in automotive throttle bodies. I think "stepper motors" is the proper term for electric motors that don't really spin continuously, but stop at a range of levels.
Know anything about how these companies dealt with the electronic aspect when developing these products since their engineers previous only dealt with mechanical devices?
Know anything about how these companies dealt with the electronic aspect when developing these products since their engineers previous only dealt with mechanical devices?
#2
Every day a winding road
What makes you think they don't have electronic engineers?
https://www.modernbike.com/sram-comp...-rate-monitors
https://www.pricefalls.com/product/c...76347360542896
https://www.pricefalls.com/product/s...75728885230148
https://www.modernbike.com/sram-comp...-rate-monitors
https://www.pricefalls.com/product/c...76347360542896
https://www.pricefalls.com/product/s...75728885230148
#3
Dirty Heathen
Join Date: Oct 2015
Location: MC-778, 6250 fsw
Posts: 2,182
Bikes: 1997 Cannondale, 1976 Bridgestone, 1998 SoftRide, 1989 Klein, 1989 Black Lightning #0033
Mentioned: 19 Post(s)
Tagged: 0 Thread(s)
Quoted: 889 Post(s)
Liked 906 Times
in
534 Posts
I can't believe I'm going to be defending engineers, but there's a lot more to making a product than just designing it. In most (large-scale) manufacturing operations, there are far more engineers dedicated to the equipment and processes that make the products than there are design engineers. Some of these guys are pretty smart. Just because you work on bicycle parts doesn't mean you don't know anything about robotics/automation, for example.
Besides, in the Di2 example, Shimano doesn't actually manufacture the stepper motor themselves. It's most likely produced by someone like NGK, Denso or Bosch, who specialize in that sort of thing. Shimano just gave them the specs for what they needed.
Besides, when you're doing proof-of-concept, you usually start with off-the-shelf parts and components, modified to do what you're trying to achieve. This is the Frankenstien workbench stage; It's not a usuable protype yet, but it just proves that you can make it work at all. Then you refine the POC to determine the actual requirments (force, power supply) the prototype will need, and how to make it fit within the constraints of your project.
Besides, in the Di2 example, Shimano doesn't actually manufacture the stepper motor themselves. It's most likely produced by someone like NGK, Denso or Bosch, who specialize in that sort of thing. Shimano just gave them the specs for what they needed.
Besides, when you're doing proof-of-concept, you usually start with off-the-shelf parts and components, modified to do what you're trying to achieve. This is the Frankenstien workbench stage; It's not a usuable protype yet, but it just proves that you can make it work at all. Then you refine the POC to determine the actual requirments (force, power supply) the prototype will need, and how to make it fit within the constraints of your project.
#4
Senior Member
Join Date: Apr 2005
Location: Ann Arbor, MI
Posts: 16,880
Bikes: 1980 Masi, 1984 Mondonico, 1984 Trek 610, 1980 Woodrup Giro, 2005 Mondonico Futura Leggera ELOS, 1967 PX10E, 1971 Peugeot UO-8
Mentioned: 49 Post(s)
Tagged: 0 Thread(s)
Quoted: 1858 Post(s)
Liked 664 Times
in
506 Posts
I can't say what they actually did, but from my 35 years in vehicular electronics I can offer a few thoughts. To actually know the answer to your question, one would have had to work for one of teh companies, and what actually happenned would have been confidential information. The systems perspective had to be established first, and the makers of 8/9/10/11 mechanical indexing systems already knew this part - how much to move the mech for each desired change, given the target positions of the rear and front chainrings, and how to figure in needed over shift and fadeback motions. These spacings and motion increments may have dictated stepper motors or not. Steppers increment a shaft by a certain number of degrees on each step, normally. If the cassette really has even pitch between adjacent then perhaps a stepper is good, but it stepper rotor pole piece systems don't accommodate all spacings and angular increments: it may also have needed a transmission of some sort to exactly match the motor steps to the mech. But then, what about changing the design from 10 to 11? The nominal step width and hence the rotary step must be decreased by exactly 1/11.
In any case, I would probably have begin to recruit for design capability in motor drives, power electronics, and system control algorithms perhaps 2 years before the product is planned for road testing. I'm certain consultants in those skill areas could have been found, but they may not have had bicycle component engineering experience.
As I alluded, I think these systems could have been designed based on any type of electric motor: DC, DC Brushless, stepper, or even variable reluctance. Some type of position sensor would have been needed for each of them but it might also have been needed for a stepper-based solution. The cost and complexity trade-offs should have revealed why the stepper (or other) was the preferred design solution.
In any case, I would probably have begin to recruit for design capability in motor drives, power electronics, and system control algorithms perhaps 2 years before the product is planned for road testing. I'm certain consultants in those skill areas could have been found, but they may not have had bicycle component engineering experience.
As I alluded, I think these systems could have been designed based on any type of electric motor: DC, DC Brushless, stepper, or even variable reluctance. Some type of position sensor would have been needed for each of them but it might also have been needed for a stepper-based solution. The cost and complexity trade-offs should have revealed why the stepper (or other) was the preferred design solution.
Last edited by Road Fan; 10-17-18 at 09:19 AM.
#5
Senior Member
Join Date: Apr 2005
Location: Ann Arbor, MI
Posts: 16,880
Bikes: 1980 Masi, 1984 Mondonico, 1984 Trek 610, 1980 Woodrup Giro, 2005 Mondonico Futura Leggera ELOS, 1967 PX10E, 1971 Peugeot UO-8
Mentioned: 49 Post(s)
Tagged: 0 Thread(s)
Quoted: 1858 Post(s)
Liked 664 Times
in
506 Posts
I can't believe I'm going to be defending engineers, but there's a lot more to making a product than just designing it. In most (large-scale) manufacturing operations, there are far more engineers dedicated to the equipment and processes that make the products than there are design engineers. Some of these guys are pretty smart. Just because you work on bicycle parts doesn't mean you don't know anything about robotics/automation, for example.
Besides, in the Di2 example, Shimano doesn't actually manufacture the stepper motor themselves. It's most likely produced by someone like NGK, Denso or Bosch, who specialize in that sort of thing. Shimano just gave them the specs for what they needed.
Besides, when you're doing proof-of-concept, you usually start with off-the-shelf parts and components, modified to do what you're trying to achieve. This is the Frankenstien workbench stage; It's not a usuable protype yet, but it just proves that you can make it work at all. Then you refine the POC to determine the actual requirments (force, power supply) the prototype will need, and how to make it fit within the constraints of your project.
Besides, in the Di2 example, Shimano doesn't actually manufacture the stepper motor themselves. It's most likely produced by someone like NGK, Denso or Bosch, who specialize in that sort of thing. Shimano just gave them the specs for what they needed.
Besides, when you're doing proof-of-concept, you usually start with off-the-shelf parts and components, modified to do what you're trying to achieve. This is the Frankenstien workbench stage; It's not a usuable protype yet, but it just proves that you can make it work at all. Then you refine the POC to determine the actual requirments (force, power supply) the prototype will need, and how to make it fit within the constraints of your project.
I think you're right on with the proof of concept discussion. One starting point might have been to simply add motors and geartrains to existing Dura-Ace or Super Record groups as sold. If I was on the project I would have started with as much as possible that is already known. But of course that would not be what you want to manufacture.
I do think it is not always the case that the design comes from refining the POC. Certainly one learns a lot from the POC, but it could well be that the next version will be a fresh start based on the total of what has been learned to date.
#8
Super Moderator
Join Date: Jul 2004
Location: Ffld Cnty Connecticut
Posts: 21,843
Bikes: Old Steelies I made, Old Cannondales
Mentioned: 12 Post(s)
Tagged: 0 Thread(s)
Quoted: 1173 Post(s)
Liked 927 Times
in
612 Posts
Mavic ZAP was the first electronic group, afaik.
__________________
Bikes: Old steel race bikes, old Cannondale race bikes, less old Cannondale race bike, crappy old mtn bike.
FYI: https://www.bikeforums.net/forum-sugg...ad-please.html
Bikes: Old steel race bikes, old Cannondale race bikes, less old Cannondale race bike, crappy old mtn bike.
FYI: https://www.bikeforums.net/forum-sugg...ad-please.html
#9
Senior Member
Thread Starter
"...and how to figure in needed over shift and fadeback motions"
Road Fan- Do you know for sure that electronic mechs are programmed to do these complex shifting motions? They must be or you wouldn't have suggested it. That's awesome.
I'm glad that you discussed the types of motors possibly used and that maybe the motors are not steppers with some type of transmission or gearing. Would stepper motors be powerful enough? I know that some drive by wire throttle bodies are powerful enough to injure fingers if somebody presses the accelerator pedal when fingers are in there. They are apparently quite fast acting and powerful.
Road Fan- Do you know for sure that electronic mechs are programmed to do these complex shifting motions? They must be or you wouldn't have suggested it. That's awesome.
I'm glad that you discussed the types of motors possibly used and that maybe the motors are not steppers with some type of transmission or gearing. Would stepper motors be powerful enough? I know that some drive by wire throttle bodies are powerful enough to injure fingers if somebody presses the accelerator pedal when fingers are in there. They are apparently quite fast acting and powerful.
#10
Senior Member
Join Date: Apr 2005
Location: Ann Arbor, MI
Posts: 16,880
Bikes: 1980 Masi, 1984 Mondonico, 1984 Trek 610, 1980 Woodrup Giro, 2005 Mondonico Futura Leggera ELOS, 1967 PX10E, 1971 Peugeot UO-8
Mentioned: 49 Post(s)
Tagged: 0 Thread(s)
Quoted: 1858 Post(s)
Liked 664 Times
in
506 Posts
On electronic throttle bodies you are correct. I first came to Michigan to design one at Ford, and they were frighteningly powerful. Stepper motors can certainly be built with enough torque for a bike derailleur. But torque and power-wise, all the major motor/electrical machine technologies were suitable for a throttle control design. In other words any motor technology could be had in any power output required.
To your question, no I do not know directly that overshift profiles are followed by eshift systems. I assume that they are, based on the following. I know that in some of Frank Berto's old analyses of how derailleurs need to work, when the tooth design was not very "grabby" (my term, not a technical term), the chain needs to be shifted past the point where it runs smoothly on the new sprocket and then the cage pulled back to the smooth-running point after the shift has been completed. This manual overshift was accomplished by the human hand by riders like me who are old enough to have grown up with pre-indexing derailleur shifting. We knew we had to do what was needed to shift those gears smartly on the road, social ride, tour, or race, so overshift was a fact of life. Somewhere, perhaps BF, I also read that overshift is sometimes built in to existing indexed shifters. From this point I conclude, if some kind of controlled overshift is still necessary for mechanical indexed shifting it is most likely still necessary for electronic indexed shifting. And I have high confidence that designers of the control algorithms can handle it, if it is really needed. After that, it's just code, and the software engineers live to make code work, among other things.
So the fact I mentioned it means that I think some method to provide overshift is necessary based on my experience, not that I am certain they are actually made that way. Engineers can get pretty tricky, and somebody may have thought of a way to make a full-range derailleur system that does not need to control overshift. Maybe there is not a necessity, at least not one using overshift.
To your question, no I do not know directly that overshift profiles are followed by eshift systems. I assume that they are, based on the following. I know that in some of Frank Berto's old analyses of how derailleurs need to work, when the tooth design was not very "grabby" (my term, not a technical term), the chain needs to be shifted past the point where it runs smoothly on the new sprocket and then the cage pulled back to the smooth-running point after the shift has been completed. This manual overshift was accomplished by the human hand by riders like me who are old enough to have grown up with pre-indexing derailleur shifting. We knew we had to do what was needed to shift those gears smartly on the road, social ride, tour, or race, so overshift was a fact of life. Somewhere, perhaps BF, I also read that overshift is sometimes built in to existing indexed shifters. From this point I conclude, if some kind of controlled overshift is still necessary for mechanical indexed shifting it is most likely still necessary for electronic indexed shifting. And I have high confidence that designers of the control algorithms can handle it, if it is really needed. After that, it's just code, and the software engineers live to make code work, among other things.
So the fact I mentioned it means that I think some method to provide overshift is necessary based on my experience, not that I am certain they are actually made that way. Engineers can get pretty tricky, and somebody may have thought of a way to make a full-range derailleur system that does not need to control overshift. Maybe there is not a necessity, at least not one using overshift.
Last edited by Road Fan; 10-17-18 at 08:46 PM.
#12
Senior Member
Join Date: Oct 2013
Location: Oahu, HI
Posts: 1,396
Bikes: 89 Paramount OS 84 Fuji Touring Series III New! 2013 Focus Izalco Ergoride
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 285 Post(s)
Likes: 0
Liked 74 Times
in
54 Posts
I would think a good bit of effort went into the power budget. Both the instantaneous and overall requirement and the battery storage to supply that power. In the case of SRAM I wonder how much existing wireless comm technology was used or adapted.
scott s.
.
scott s.
.
#14
FLIR Kitten to 0.05C
Join Date: Sep 2014
Location: Lincoln, Nebraska
Posts: 5,331
Bikes: Roadie: Seven Axiom Race Ti w/Chorus 11s. CX/Adventure: Carver Gravel Grinder w/ Di2
Mentioned: 30 Post(s)
Tagged: 0 Thread(s)
Quoted: 2349 Post(s)
Liked 406 Times
in
254 Posts
"...and how to figure in needed over shift and fadeback motions"
Road Fan- Do you know for sure that electronic mechs are programmed to do these complex shifting motions? They must be or you wouldn't have suggested it. That's awesome.
I'm glad that you discussed the types of motors possibly used and that maybe the motors are not steppers with some type of transmission or gearing. Would stepper motors be powerful enough? I know that some drive by wire throttle bodies are powerful enough to injure fingers if somebody presses the accelerator pedal when fingers are in there. They are apparently quite fast acting and powerful.
Road Fan- Do you know for sure that electronic mechs are programmed to do these complex shifting motions? They must be or you wouldn't have suggested it. That's awesome.
I'm glad that you discussed the types of motors possibly used and that maybe the motors are not steppers with some type of transmission or gearing. Would stepper motors be powerful enough? I know that some drive by wire throttle bodies are powerful enough to injure fingers if somebody presses the accelerator pedal when fingers are in there. They are apparently quite fast acting and powerful.
https://carltonbale.com/shimano-di2-...-need-to-know/
In general...all these companies do a fairly bad job at making documentation about all this stuff. For eyars Carlton Bale's page has been the consumer bible about all things Di2--due to how bad Shimano was at consumer documentation. Di2, for its part (and the on e-shifting system I'm fluent in) is very cleverly engineered in all it does, not only in precision but in power management.
#15
Banned.
Join Date: Oct 2018
Posts: 209
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 140 Post(s)
Likes: 0
Liked 2 Times
in
1 Post
"The final position of the eTap Yaw front derailleur will allow access to all of the rear cogs with no chain rub; it therefore doesn’t require a trim function, meaning that once a front shift is fully completed, the front derailleur will stay in exactly the same spot regardless of the rear cog you select following the completion of your front shift."
#16
Disco Infiltrator
Join Date: May 2013
Location: Folsom CA
Posts: 13,446
Bikes: Stormchaser, Paramount, Tilt, Samba tandem
Mentioned: 72 Post(s)
Tagged: 0 Thread(s)
Quoted: 3126 Post(s)
Liked 2,105 Times
in
1,369 Posts
Any reasonably clever person with a micro processor and a servo could make a bike shifter. The difficult part is turning it into a product.
#17
Every day a winding road
#18
Disco Infiltrator
Join Date: May 2013
Location: Folsom CA
Posts: 13,446
Bikes: Stormchaser, Paramount, Tilt, Samba tandem
Mentioned: 72 Post(s)
Tagged: 0 Thread(s)
Quoted: 3126 Post(s)
Liked 2,105 Times
in
1,369 Posts
Well you need a Shimergo article and a Basic Stamp or whatever the modern equivalent is. Then you wind up with a project box attached to your bike, which is sub-optimal. You need something that can stand 20 years of use and keep out the weather and look good and be made for a profit. That's hard.
__________________
Genesis 49:16-17
Genesis 49:16-17
#19
Every day a winding road
Well you need a Shimergo article and a Basic Stamp or whatever the modern equivalent is. Then you wind up with a project box attached to your bike, which is sub-optimal. You need something that can stand 20 years of use and keep out the weather and look good and be made for a profit. That's hard.
Could a clever person envision how sucha a product would be designed? Of course. I am not that clever and I have some ideas of how this would be built. Could a clever person actauylly build the thing and have it work reliably? Likely not. At least not without vast resources.
#20
Senior Member
Join Date: Aug 2017
Location: Central Io-way
Posts: 2,673
Bikes: LeMond Zurich, Giant Talon 29er
Mentioned: 17 Post(s)
Tagged: 0 Thread(s)
Quoted: 1221 Post(s)
Liked 629 Times
in
472 Posts
That's what I've read as well, surprised no one else has mentioned it?
It's unique in that it didn't use a servo motor, instead relying on a solenoid and the cassette/chain to physically move the derailleur.
I suspect the seemingly slow advent of e shifting is from need, and not technology limitations. A bike transmission is dead simple, and cable actuation is dirt cheap. Of course as technology progresses prices come down.
What I think would be cool? Electronic shifting paired with a power meter and cadence meter, so that shifts could always be at the perfect point in your pedal stroke, and the derailleur could adjust the force required for silent shifts under full power. Maybe this is a bad idea, as we all have a certain amount of pride in a well executed shift while under power as we pass someone
It's unique in that it didn't use a servo motor, instead relying on a solenoid and the cassette/chain to physically move the derailleur.
I suspect the seemingly slow advent of e shifting is from need, and not technology limitations. A bike transmission is dead simple, and cable actuation is dirt cheap. Of course as technology progresses prices come down.
What I think would be cool? Electronic shifting paired with a power meter and cadence meter, so that shifts could always be at the perfect point in your pedal stroke, and the derailleur could adjust the force required for silent shifts under full power. Maybe this is a bad idea, as we all have a certain amount of pride in a well executed shift while under power as we pass someone
#21
Every day a winding road
It is a solution looking for a problem. Who wants to go for a ride and risk losing charge on a battery?
My Aurora has both index shifting and friction shift. The reason? In case index shifting fails. You arre getting that in electronic shifting.
The only advantage that I see to electronic shifting is that you could have individual settings between gears. But with ta properly designed/installed/maintained shifter/cassette, I don't see where it would really be needed.
My Aurora has both index shifting and friction shift. The reason? In case index shifting fails. You arre getting that in electronic shifting.
The only advantage that I see to electronic shifting is that you could have individual settings between gears. But with ta properly designed/installed/maintained shifter/cassette, I don't see where it would really be needed.
#22
Super Moderator
Join Date: Jul 2004
Location: Ffld Cnty Connecticut
Posts: 21,843
Bikes: Old Steelies I made, Old Cannondales
Mentioned: 12 Post(s)
Tagged: 0 Thread(s)
Quoted: 1173 Post(s)
Liked 927 Times
in
612 Posts
That's what I've read as well, surprised no one else has mentioned it?
It's unique in that it didn't use a servo motor, instead relying on a solenoid and the cassette/chain to physically move the derailleur.
I suspect the seemingly slow advent of e shifting is from need, and not technology limitations. A bike transmission is dead simple, and cable actuation is dirt cheap. Of course as technology progresses prices come down.
What I think would be cool? Electronic shifting paired with a power meter and cadence meter, so that shifts could always be at the perfect point in your pedal stroke, and the derailleur could adjust the force required for silent shifts under full power. Maybe this is a bad idea, as we all have a certain amount of pride in a well executed shift while under power as we pass someone
It's unique in that it didn't use a servo motor, instead relying on a solenoid and the cassette/chain to physically move the derailleur.
I suspect the seemingly slow advent of e shifting is from need, and not technology limitations. A bike transmission is dead simple, and cable actuation is dirt cheap. Of course as technology progresses prices come down.
What I think would be cool? Electronic shifting paired with a power meter and cadence meter, so that shifts could always be at the perfect point in your pedal stroke, and the derailleur could adjust the force required for silent shifts under full power. Maybe this is a bad idea, as we all have a certain amount of pride in a well executed shift while under power as we pass someone
Wouldn't interest me. I like deciding when to shift, and choosing my cadence based on how I feel that day.
__________________
Bikes: Old steel race bikes, old Cannondale race bikes, less old Cannondale race bike, crappy old mtn bike.
FYI: https://www.bikeforums.net/forum-sugg...ad-please.html
Bikes: Old steel race bikes, old Cannondale race bikes, less old Cannondale race bike, crappy old mtn bike.
FYI: https://www.bikeforums.net/forum-sugg...ad-please.html
#23
FLIR Kitten to 0.05C
Join Date: Sep 2014
Location: Lincoln, Nebraska
Posts: 5,331
Bikes: Roadie: Seven Axiom Race Ti w/Chorus 11s. CX/Adventure: Carver Gravel Grinder w/ Di2
Mentioned: 30 Post(s)
Tagged: 0 Thread(s)
Quoted: 2349 Post(s)
Liked 406 Times
in
254 Posts
That's what I've read as well, surprised no one else has mentioned it?
It's unique in that it didn't use a servo motor, instead relying on a solenoid and the cassette/chain to physically move the derailleur.
I suspect the seemingly slow advent of e shifting is from need, and not technology limitations. A bike transmission is dead simple, and cable actuation is dirt cheap. Of course as technology progresses prices come down.
What I think would be cool? Electronic shifting paired with a power meter and cadence meter, so that shifts could always be at the perfect point in your pedal stroke, and the derailleur could adjust the force required for silent shifts under full power. Maybe this is a bad idea, as we all have a certain amount of pride in a well executed shift while under power as we pass someone
It's unique in that it didn't use a servo motor, instead relying on a solenoid and the cassette/chain to physically move the derailleur.
I suspect the seemingly slow advent of e shifting is from need, and not technology limitations. A bike transmission is dead simple, and cable actuation is dirt cheap. Of course as technology progresses prices come down.
What I think would be cool? Electronic shifting paired with a power meter and cadence meter, so that shifts could always be at the perfect point in your pedal stroke, and the derailleur could adjust the force required for silent shifts under full power. Maybe this is a bad idea, as we all have a certain amount of pride in a well executed shift while under power as we pass someone
We take for granted how well things can work... When the reality is that they are frightfully complex and can come apart.
#25
Dirty Heathen
Join Date: Oct 2015
Location: MC-778, 6250 fsw
Posts: 2,182
Bikes: 1997 Cannondale, 1976 Bridgestone, 1998 SoftRide, 1989 Klein, 1989 Black Lightning #0033
Mentioned: 19 Post(s)
Tagged: 0 Thread(s)
Quoted: 889 Post(s)
Liked 906 Times
in
534 Posts
Could a clever person envision how sucha a product would be designed? Of course. I am not that clever and I have some ideas of how this would be built. Could a clever person actauylly build the thing and have it work reliably? Likely not. At least not without vast resources.
Could I build a fully-integrated Di2? Heck no, getting everything shrunk down and integrated is what the OEM's do.
Last edited by Ironfish653; 10-21-18 at 06:26 PM.