E-Shift

[davamb]

Since the accident, I've not been able to get enough bend in my left leg to change gear on the TRX, so it has sat idle for the last 3 years which is a poor way to treat the old gal. I'm ok on sit-up-and-beg bikes, but just can't accommodate changing gear in the "sporty" riding position. Least, not with anything less than my bum perched two inches or more off the seat.

I looked into power shifters, but these were stupidly expensive, so I decided to see if I could cook up my own. I tried solenoids, but not enough drive range there. Eventually went to a servo motor, so-called "H-bridge" driver and a small microcontroller to process up/down switch requests and supply the relevant up/down and PWM drive signals.

So. After a couple of weeks R&D, I have this:

https://gdurl.com/aiZe

Decided to ditch the gear lever and drive the change shaft directly. It was working ok, but the sloppy worn joints were making the operation less than optimal. Positive results!

The drive control algorithm is a state machine serviced every 10ms. This controls the ramp-up and down of speed and the amount of time it dwells at full speed. I have already included a method to program the dwell, but it is symmetrical. That is, it spends the same time in dwell moving the lever as it spends in returning it to its reference position. It looks like it might need a bit longer dwell on the return stroke. Will add this on the next software session.

I might get Bessie back on the track yet!

[dicky]

Excellent work!

[BigAl-TC]

Hi Dave,

Sorry to hear your injury has impeded your riding. Looks like your on your way to circumventing the problem. Up down paddles on the way ?
Looking forward to seeing your progress.

Cheers, Al.

[HolerTogni]

Hi Dave!

Very glad to read from you again!

Also very glad to read that you obviously are able to ride your TRX again - I'm keeping my fingers crossed that it will work as you want it!

"Keep thinking of you"-greets from Munich!

[davamb]

Thanks chaps,

checked in every now and then, but sorta thought I didn't belong here any more as I wasn't "active" in the TRX department as it were. Stoopid I guess, but the way I look at the world has changed so much since the stack. Good gear meant no damage to pelvis, spine and skull. ATGATT means I'm here today with all my braincells and all my limbs. Don't go out with anything less or I will come around to your place and poke you with a pointed stick and call you "an insipid silly thing!" You have been warned.

Anyways, enough of that, back on topic!

I'm hoping to fire her up today. Trying to fine-tune the shift with the motor not running is not very productive. Them thar doggies don't engage very well with the motor stationary.

I'm also intending to integrate it with the Ignitech module. I've got spare I/O on the microcontroller, so I'll take the "Shift" light signal as an input and use that to automatically shift up. I'll also send out a "Upshift" signal to let the Ignitech know when to skip a few sparx. (Will opto-couple those signals.) Could be a lot of fun on the track!

I've removed the manual shift completely now and placed the shift buttons on a lash-up mount on the LHS handlebar. Will shove up some pix later. Ready to go now I guess.

[BigAl-TC]

Don't be shy mate, we need you, pretty quite on here these days.
Have often wondered how you are coping since your mishap.
All the best.
Al.

[Tarwetijger]

ATGATT means I'm here today with all my braincells and all my limbs. Don't go out with anything less or I will come around to your place and poke you with a pointed stick and call you "an insipid silly thing!" You have been warned.

It's a long journey from you to me, so lucky for you, I'm also an ATGATT type rider.

I've removed the manual shift completely now and placed the shift buttons on a lash-up mount on the LHS handlebar. Will shove up some pix later. Ready to go now I guess.

That sounds like a proper project, I'm looking forward for some pics.

[davamb]

After a bit more work... ok, so after a lot more work, I'm up to this:

https://gdurl.com/6s-7

I've got UP and DOWN shift buttons mounted under the LHS grip and I'm driving the unit using them. I also drive the Neutral light to show the gear I'm in; steady on for Neutral and flash the number of times corresponding to the gear number.
Just ignore the actuator "jitter", I've got the integral dialed-up too high, it's hunting for the reference position, but too much gain means it overshoots and oscillates around the setpoint position. It'll tune out.
I start in Neutral but I haven't "learned" the 1st to N and 2nd to N shifts yet. I'll need a third button for that - a "find Neutral" button and I'll set it up to automatically seek N from any position.
I'll try and take her for a punt around the block tomorrow and see how it actually goes. Could be fun, but this is all running off an external 12V battery perched on the seat, need to connect it to the bike's 'lectrix yet.

Cost of parts so far is under $100, and that's retail pricing, so, looking ok so far. I'll design a custom PCB for it and get the robots at work to build it.

Sorry old Bessie is sooo dirty! That's no way to treat the ol' gal.

Edit: Here's a wmv version for anyone without VLC:

https://gdurl.com/qlp8

Edit 2: and for Appleites:

https://gdurl.com/Hwq0

(mov) and (mp4):

https://gdurl.com/fnz72

[HolerTogni]

Hi Dave!

Wow, sounds really great what you have achieved so far!
(besides, not sure about the "automatically find neutral from all positions" function - but you will know or at least find out!)

I wish you all (!) the best for your test ride and I'm sure ol' Bessie well understands and knows that this solution is far better than any of the thinkable alternatives under these circumstances.

Keeping my fingers crossed greets from Munich!

[davamb]

Thanks Jochen,

any and all feedback appreciated. Yeah, maybe the "neutral from all positions" idea isn't so great, but I need a way of selecting Neutral from 1st or 2nd and I think a third button is the way to go:
If pressed in 1st it'll do a "half-shift" up and if pressed in 2nd it'll do a "half-shift" down into Neutral.
Press-and-hold might do the trick, but I don't want to slow down the decision process for normal 1st to 2nd upshifts. Ditto downshifts.

If I get the opportunity for a test ride today, I'll give the poor old thing a decent wash!

[HolerTogni]

Hi Dave!

I think the third button as you described it is a good idea - as also is the TLC for ol' Bessie!

Still curious about the results of your test ride greets from Munich!

[davamb]

Alas, other matters took me away from the project yesterday, but I'll see if I can get out on her today. Still got to get the shifter working off bike power, but that shouldn't be an issue.

I'm thinking of something like this for controls:
https://www.thirdgear.com.au/Black-Alum ... itch-Combo
Up = upshift, Down = downshift and Middle = neutral (only active is 1st and 2nd).
I've ordered one to assess. Might be a bit too clumsy to operate while using the clutch lever, but I guess I'll find out.

Gotta be better 'n this!

[davamb]

I managed a short run around the block before another actuator gearbox shat itself. Thought the damper / cush drive thingy would cure the problem, but alas no. I've got to get to the bottom of this, it should handle the steady-state torque, but I think the hard stops are what's doing the damage. I've got some new parts to increase the current-sense gain so need to revisit this area. Will look at running it in constant-current (software current-limit) mode.

I've ordered a new motor sample that supposedly will have a stronger gearbox, but I think I'll end up with a worm drive solution. Don't really want to go down this track as you can't reverse drive them. Maybe just have to live with that. Hate that sort of design compromise.

In the meantime, I've started on Mk III. This is an in-line (with the frame) version with a set of bevel gears to bring the drive out to the side. Just a lash-up here with a cable tie to assess size and position.



It also gives a better angle on the drive push/pull rod. No problem with any drive reversals, I'll just have a "flip" option in software. Or just reverse the drive to the motor. Simples.

I've also sourced a handlebar switch:



Only one is momentary, the other two are push-on, push-off. No problem, I can detect both make and break edges to make them appear as momentary. Even has blue LED illumination on each switch! Golly! If I had spare I/O lines, I'd drive the LEDs, but I don't so I'll leave them just connected to the lighting circuit. Might be better to drive these LEDs instead of the Neutral lamp - I could do this with zero extra cost of I/O lines.

Been thinking about pre-selectors. These gearboxes from days of yore allowed the driver to pre-select the next gear they wanted and it would then be engaged when the clutch was activated. I could do this by sensing the clutch switch. Would mean you don't have to stuff around trying to thumb the gear switch while activating the clutch at the same time. Could be interesting! Better make it self cancel after a couple of secs though, just in case...

[dicky]

Can you pull/push on the actuator arm to manually change gears when it all goes titsup*?

*total inability to support usual performance

[davamb]

Yep, I can Dicky,

that's why I don't want to go down the worm drive path. You can't apply a reverse torque to turn a worm drive and have it rotate the motor.

Worms are useful in
(a) "self-parking" situations,
(b) where you want a huge gear ratio without a stack of gears and
(c) able to handle good torque
but not my preference here. (Really only for (a) - (b) and (c) are just fine.)

There are really nice robust hi-torque DC servo motors for industrial control applications, but these cost a shitload. If I can come up with an engineering work-around for these "hobbby" type motor/gearbox combos that I've been trialling then I can make a very cheap solution. Well, much cheaper than what's out there at the moment at least! Think current-limiting is the go, it translates to torque-limiting. That'll be my next round of experiments. I have the ADC reading the current, and tried a few variations, but I need to revisit this with more detail. The damper/cush drive will then take care of the "shock" and leave the microcontroller to deal with the torque.

Any thoughts from the brains trust on pre-selector operation? Guess I could make it optional. And let the Shift Light input over-ride it when going WOT clutchless up-changes...