Sounds vaguely familiar...
Oh yes, that's it! Bus pre-selector
Sequential conversion
Ok, I know this has been discussed to death on the forum previously, but I thought I would see if I can get an up to date idea on what is possible.
I know Jack Knight did a conversion (http://www.jackknight.co.uk/press_re...%20release.pdf) but is it;
A) still available
B) any good
C) cost efficient
D) the only one on the market?
I'm looking to play with my spare UN1 shortly.
Jeremy and myself have been looking for the past two years for a sequential system for our projects, nothing seems to hit the mark though. The closest
is this conversion that you may or may not have seen before.
linkage
It seems a lot of money though especially when you still need to use the clutch, though maybe we are expecting too much.
Apologies if this is old news.
Big Money
Not a UN1, but should give you an idea on potential pricing
ATB
Simon
Thanks mr c. I haven't seen that before, but they don't have anything compatible with the UN1. I was hoping for just a stick driven
sequential shift and manual clutch. Nothin special, just simple sequential gears.
Thanks for the post tho. Good luck in your search too.
[Edited on 22-12-11 by MakeEverything]
quote:
Originally posted by Simon
Big Money
Not a UN1, but should give you an idea on potential pricing
ATB
Simon
Hmmm. Interesting idea on the Mastershift, being able the stack the gear changes by multiple pushes on the buttons, then activating the changes by
using the clutch.
Sounds vaguely familiar...
Oh yes, that's it! Bus pre-selector
quote:
Originally posted by Xtreme Kermit
Hmmm. Interesting idea on the Mastershift, being able the stack the gear changes by multiple pushes on the buttons, then activating the changes by using the clutch.
Sounds vaguely familiar...
Oh yes, that's it! Bus pre-selector
Surely a sequential conversion for a standard sychro box is a waste of time, it certainly won't give faster shifts?
You could say the same thing about putting lotus Carlton running gear into an early senator shell.
This does not stop it happening or people enjoying the challenge.
I don't think the control box is the problem. It should be straightforward to use a processor linked to speed and engine revs so that when the
clutch is engaged it will select up or down. Not quicker as sutch but it gives you less to think about in the heat of the moment.
Also it would overcome loads of problems with mechanical lnkages on rear engined cars which tend to be slow.
Looks as if mastershift are using 1/4 up servos the bit I can't work out is how do they select neutral when shifting from 2 to 3?
Is it worth collaborating on the un1 shift to arrive at a locost solution?
quote:
Originally posted by Xtreme Kermit
You could say the same thing about putting lotus Carlton running gear into an early senator shell.
This does not stop it happening or people enjoying the challenge.
quote:
Originally posted by v8kid
I don't think the control box is the problem. It should be straightforward to use a processor linked to speed and engine revs so that when the clutch is engaged it will select up or down. Not quicker as sutch but it gives you less to think about in the heat of the moment.
Also it would overcome loads of problems with mechanical lnkages on rear engined cars which tend to be slow.
Looks as if mastershift are using 1/4 up servos the bit I can't work out is how do they select neutral when shifting from 2 to 3?
Is it worth collaborating on the un1 shift to arrive at a locost solution?
Yes but how do they sense neutral that's the crux
I see where you are coming from.
Do you need to though?
If you can program in three for and aft positions, you don't need to run with feedback from the gearbox, just the positioning servos.
position 1 = (1,3,5)
Position 2 = neutral
Position 3 = (2,4,6)
Of course reverse is in there somewhere too...
And the left/right servo would have known positions for
Reverse
1,2
3,4
5,6
Does that make sense?
Yup but there has to be a reference position where the servo starts from. Say you set it up in first and then move x to neutral and a further y to
second. With all the slop, backlash, vibration the setting can move and second will not be fully engaged. Worse what happens if the variation is so
great that coming back from second to neutral to go to third and it misses neutral and the servo is pulling across the gate in vain?
Having neutral as the default position overcomes this and allows the actuator linkage to be spring loaded ensuring that the gears are rammed home.
I'm not sure now I'll have to ponder over it but I think Crimbo duties are calling as I hear the kids stirring 
Cheers!
Could it be done with limit switches? Use neutral as centre and then drive until the appropriate limit switch is hit? Could get real messy.
Maybe a PLC with a set of stepper motors is the way. You would probably have to calibrate it every so often , but using neutral as a centre spot, turn
N timess to reach a goal position...
Sorry if I am missing something but even if you could automate the change action with any number of motors etc you would still have to depress the
clutch manually unless you could somehow incororate an hydraulic pump to do that which would surely make the whole thing incredibly complicated. Is
this why all the mainstream manufacturers have not developed a system like this for production cars.
Sounds great and I love these left field ideas that people get working but how much hassle would it be.
With regards to the pre-selector box my dad had such a beastie in a Lancester when I was but a tiny brat. He loved it but a lack of spares for the
diff mean't the car got scrapped, the wire wheels ended up on a cement mixer!!!! one of the solid drieshafts is still in garage having been
turned into a long chisel, the splines are still just visible. The Lancester had what dad calls a "fluid flywheel" guess the ancient
equivalent of a Torque converter, don't know any more than that. Pre war ERA grand prix cars used the pre-selector box, still love watching them
at historic meetings, seeing the drivers at work (especially at Cadwell) and noticing them moving the gear selector as they come onto the strait and
depressing the clutch several seconds later to engage the gear brings back fond memories.
How fast can stepper motors turn? 1 to 2 would take 32 turns assuming 16 tpi rod and 135 oz in stepper to give 135 lbf to match mastershift specs.
Assuming .5 sec shift that's 3840 rpm at rated torque. Can steppers do this?
Hmmm When you put it like that, it's an awfull lot of work to do in 0.5 seconds.
Are there any other alternatives?
quote:
Originally posted by Paul Turner
Is this why all the mainstream manufacturers have not developed a system like this for production cars.
quote:
Originally posted by phelpsa
quote:
Originally posted by Paul Turner
Is this why all the mainstream manufacturers have not developed a system like this for production cars.
Oh but they have! Citroen's sensodrive system works in exactly this way.
quote:
Originally posted by MikeRJ
quote:
Originally posted by phelpsa
quote:
Originally posted by Paul Turner
Is this why all the mainstream manufacturers have not developed a system like this for production cars.
Oh but they have! Citroen's sensodrive system works in exactly this way.
So have Smart and some Fiats also had a robotised manual box. They all share a common theme, they are slow to change gear. They also have a history of being unreliable. These robotised boxes use a proper selector drum like a bike gearbox however, which makes things much simpler - using a mechanism to move an H gate will be more complex and as a result more unreliable and slower. Sorry to be negative, but I simply can't see any benefits for putting work into this, unless you specifically need a robotised manual box (e.g. through disability etc.).
[Edited on 25/12/11 by MikeRJ]
Alfas Selespeed uses a manual box with a hydraulic flibberit mounted on the top to select the gears and operate the clutch. Worth looking into?
Thanks for all the comments and support fellas, some good discussion.
The gears don't necessarily need t be h pattern, s was wondering if the shafts and gear arrange,ents could be modified to make them sequential as
well. Solenoid actuation is relatively fundamental, but the controller / PLC would need some thought.
I believe 3 actuators should be able to do the job
2 for changing the gears and one for the clutch.
Programming all these shouldn't be a big issue. To make things easier, i would have 7 buttons on the steering wheel one for every gear.
Arn't a lot of these boxes on small cars the variomatic type with set belt positions when you go to the paddles and not true manual boxes with an
auto mechanism.
Just spotted the link between Citroen, Smart and Alfa, they are either low powered, unreliable or both. Seems that reliable powerful cars use a
propper auto if you must have one.
Real cars have manual boxes, totally reliable, why mess about.
have a look at this
linky
and no sign of a price anywhere! guess that tells us all we need to know...
Just found the price on their website, looking at 4k dollars..... Cheaper to buy a sequential box!
How is that little lot going to fit in the tunnel on a seven, just about enough room for the gearlever and reverse light switch in mine.
Ikeya have been doing these for various jap boxes for years
http://www.nengun.com/ikeya-formula/sequential-shifter
Ok done some back of fag packet sums and checked out requirements.
The highest force requires at the gearbox end of my UN1 box was 180oz with a 4" lever. Googling servo motors advice is to double the load when
speccing the rating.
On ebay there are CNC steppers for £19 that give 1.2Nm (170ozin).
Also available are T5 timing pulleys and belts for a few quid.
So with a 4:1 belt drive reduction ratio and if we make the servo lever arm 2" as opposed to the gearbox 4" that will give 1360ozin and we
need 720ozin so it is just shy of double the load.
the gearbox arm moves through 20 degrees when changing from 1 to 2 so the stepper will need to move 160degrees. the stepper has 200 pulses per rev so
to change from 1 to 2 requires 89 pulses which f the change is completed within 0.25sec is 356 pulses per sec - well within the motor specs.
Tons of diy or cheap ready made controllers available based on pic or arduino that can be pc programmed or we could use a raspberry pi!
So with some ingenuity buying all new parts its there for less than £100.
Further linking to road speed and engine speed no gearchange lever is required just actuate the clutch and the processor knows whether to change up or
down. Just like Jenson's gearbox!!!!
Cheers!
FWIW I know someone that made a "twini" Metro Turbo, i.e. fitted another front metro subframe, engine and box to the rear of a Metro, just
like John Coopers famous Twini. To solve the gearbox issue he made a simple robotic change for the rear engine which used windscreen wiper motors as
actuators and micro-switches as limit switches. There were also micro switches installed in the remote gearchange. It sort of worked, sometimes, and
with a bit of development it might have been made better.
One of the problems is the total lack of feedback, both to the controller and the driver. With a basic system the motors can't tell if the
synchromesh is baulking the change (e.g. if the clutch was dragging) and will simply continue to force the box into gear. With no feedback to the
driver, he has no way of knowing if the change was successful, so has to assume it's ok to release the clutch some time after the change. If it
didn't work for any reason, the box may be only partly engaged when the clutch is released, at best causing nasty grinding noises but potentially
a safety problem if you suddenly have no drive right when you need it.
These issues could be worked around by providing positional and force information back to the controller and back to the driver, perhaps by locking
the clutch down until the change has completed, but would make an already complex system more complex. The robotised boxes have control over the
clutch which removes this issue to a large extent, but explains the slowness and reliability aspects as the control system has to constantly judge the
biting point which shifts with temperature and as the clutch ages.
Thanks for the info. We're drifting into complexity now though.
My car isn't a seven, but I also was just looking at a mechanical means or "ratchet system" to see If it's possible to convert a
standard UN1 to a sequential in tune with the locostbuilders ethos. Manual clutch is fine, as is manual gears for feedback and drive ability
purposes. If this means a redesign of the primary shaft and gear arrangement, then that's what I was looking at in the first place.
Agreed trying to do it electronically is going to take a lot of expensive development to get right.
If you take a look at the link from l0rd you'll see the tuners group version seems to be based around a cunningly designed rotating plate
arrangement, so could easily be a purely mechanical system.
But beyond the 'designed it myself' aspect I don't think it adds much to a road car. Sad really as sequential boxes are in principle
simpler in design than the H pattern ones, they just cost more as they are one-offs.
Maybe a DSG gearbox could be used instead of the UN1, Audi?
http://en.wikipedia.org/wiki/Direct-Shift_Gearbox
i have an Ikeya shifter for a nissan gearbox although have never yet used it.
pic of the Mitsi EVO version may give some inspiration. Basically a ratchet system which also moves left to right to give the effect of across DIY i
suspect.
http://www.ikeya-f.co.jp/en/car_type/mitsubishi/e-m-shifter-allparts.html
Most 'bike gear changes were based on a barrel design, as was if the rumours are true, the first sequential gearboxes Ford built for the WRC
Rallycars. Rumour was that various Yamaha bike box bits were cannabilised by M Sport up in Cumbria to produce the first WRC boxes for the focus
(IIRC).
if you zoom in on the pic of this Quaife type 9 sequential, you can see the barrel design
http://www.gearboxman.co.uk/psgearboxes.html
[Edited on 28/12/11 by CNHSS1]
quote:
Originally posted by iank
Agreed trying to do it electronically is going to take a lot of expensive development to get right.
quote:
Originally posted by v8kid
quote:
Originally posted by iank
Agreed trying to do it electronically is going to take a lot of expensive development to get right.
I think electronically is a lot easier than mechanically! Every time you get a ramp angle wrong or miscalculate a spring k you have to remake the parts. Also all the parts are custom instead of using off the shelf electronics components.
Further it looses the ability to link to the engine revs and speed sensor.
Consider coming up fast to a hairpin in 4th. You are braking like mad to get 90% of the braking done before the entry and then you have your hands full with getting the car to the apex where you need second. Its busy in there and you don't quite get the engine at the right revs as you drop it into second (having had to come through third first!). Bang round comes the back end and you are exiting backwards!.
Now with electronic selection all you do is brake, turn the wheel and dip the clutch when you are ready to accelerate. The wee processor takes care of the gear selection and even knows the right revs for the engine so no more lock ups.
I really think the development of a mechanical system involves far more complex maths and fabrication skills than adjusting stepper motors with a simple processor.
Cheers!
[Edited on 28-12-11 by v8kid]
Ta for input Iank although I don't fully agree with all your points.....
quote:
Originally posted by iank
Designing electronic control systems for a living I'm quite sure it's possible,Ooh! good can you do some of the design stuff on programming them please? but I'll stick with lots of development and road/track testing, and if/when you break a gearbox it's going get expensive - by the time you're done I'll bet it costs more than going out and buying a quaife unless you can sell some to amortise the development.surely you are just as likely to break a box fiddling with the mechanicals?
My reasoning for mechanical over electronic (at least for diy attempts) is that the human brain is an awesome control system which take out a huge amount of the complexity you'll end up with in an electronic solution, but the downside is it will be the same level of involvement as an H pattern manual gearbox.Err what do you mean? Just leave as is?
BUT if you manage to get it right it has the potential to be awesome and I'll be the first to applaud it.I wouldn't mind some help to do it actually
One tip, stepper motors in my experience are either low torque and fast, medium torque and slow or high torque and really slow unless they are physically large. Most high torque high speed applications go to a servo motor approach - which are even easier to drive from a microcontroller PWM if you manage to find an integrated one you can use.Yup I agree but see my earlier post only 360 pulses per secons assuming that the shift is completed in 1/4 sec. This is well within the stepper specs that I highlighted for £19 on ebay Even with a stepper you'll need to have position feedback (or constantly recalibrate on limit switches) as you'll lose steps now and again and hit problems when the accumulate.again agreed as i highlighted in first post how do they sense neutral as reference position?
quote:
Originally posted by v8kid
quote:
Originally posted by iank
Agreed trying to do it electronically is going to take a lot of expensive development to get right.
I think electronically is a lot easier than mechanically! Every time you get a ramp angle wrong or miscalculate a spring k you have to remake the parts. Also all the parts are custom instead of using off the shelf electronics components.
Further it looses the ability to link to the engine revs and speed sensor.
Consider coming up fast to a hairpin in 4th. You are braking like mad to get 90% of the braking done before the entry and then you have your hands full with getting the car to the apex where you need second. Its busy in there and you don't quite get the engine at the right revs as you drop it into second (having had to come through third first!). Bang round comes the back end and you are exiting backwards!.
Now with electronic selection all you do is brake, turn the wheel and dip the clutch when you are ready to accelerate. The wee processor takes care of the gear selection and even knows the right revs for the engine so no more lock ups.
I really think the development of a mechanical system involves far more complex maths and fabrication skills than adjusting stepper motors with a simple processor.
Cheers!
[Edited on 28-12-11 by v8kid]
I was thinking of a conversion for mine. It helps that its a cable operated system on the mondeo box I have, as all you are looking for is to move 2
cables backwards and forwards. I was thinking of a cam type system pivoting round the gear lever pivot (one per cable obviously) to move the cable
back and forth ( a bit like the photo where its been converted to a horizontal plate that rotates), but since mines a midi, the 'tunnel' is
pretty empty and the rotating plates can be in the vertical axis which makes it simpler. You could have a third cam and some cheap switches to
indicate which gear you are in (I thought I'd at least have neutral and reverse).
Don't forget that in the end the gear lever just moves the dogs back and forth in the gearbox.
I have not got as far as deciding whether a ratchet or lock type mechanism would be required to stop the rotating plate rotating when you dont want it
to - I imagine something could be arranged like a spring loaded tapered pin that is released by the weight of your hand when you place it on the gear
lever.
As to the comments about 'imagine if you're approaching a corner.....' - well basically if you select the wrong gear you have a problem
- motorcyclist, including myself, have managed to cope with this for quite a while on motorbike sequential boxes where you may knock it down 2 or 3
gears at a time while braking.
Regards
Hugh
What happens if the system somehow fails and gets confused. If it downshifts instead of upshifting you could end up with a buzzed engine and a big bill or if it downshifts the wrong number of gear into a corner you could end up exiting this life backwards. What happens if it cannot find a gear when you are overtaking with a 42 tonner on your left and another coming strait for you.
@paul turner- One of my friends had something similar to that when the gear lever came off in his hand during a slightly dodgy overtaking manouvre in
an audi quattro (standard H box). Having said that, I'd trust a mechanical system over electronics every day, and I've worked as an
automation engineer in chemical plants for the best part of 20 years! (Note criticals safety system will always have mechanical interlocking like the
castel keylock systems)
Regards
Hugh
quote:
Originally posted by Paul Turner
What happens if it cannot find a gear when you are overtaking with a 42 tonner on your left and another coming strait for you.
quote:
Originally posted by MakeEverything
Thanks for all the comments and support fellas, some good discussion.
The gears don't necessarily need t be h pattern, s was wondering if the shafts and gear arrange,ents could be modified to make them sequential as well. Solenoid actuation is relatively fundamental, but the controller / PLC would need some thought.
Well I know that my original suggestion of doing this electronically did not go down well but......It need not be that bad.
More back of fag packet calcs and I think that a hobby servo motor would do it and still have nearly 100% power in reserve.
The key points are the motors cost under £20 each and they have all of the power electronics already inside them - and they can be easily modified to
give position feedback.
I found some programmes already written for Arduino controllers that are just about OK as they stand - the programme remembers positions and
regurgitates them on demand. See http://forums.trossenrobotics.com/tutorials/how-to-diy-128/get-position-feedback-from-a-standard-hobby-servo-3279/
It would still need two servos with a toothed belt drive and the servo time for each gearchange would be 0.12 sec per servo movement. 1 to 2 is .12
down to neutral plus another .12 down =2.4sec 2 to 3 is .12 up plus .12 across plus .12 up =.36sec etc...
Is that too slow do you think?
Can't see an easy way past reverse other than retaining the manual indent though.
So no collaboration then?
Cheers!
I've had this in the back of my mind for about 2 years now having developed a couple of similar systems (I develop prototype mechatronic
engineering solutions to order regularly). The problem I see is doing it using electromech actuators really doesn't seem elegant and while I LOVE
this sort of project I just don't see the point. You're just going to create a slow auto box because....it's not a dog box. Half a
second shift time is a lot slower than a human can do it with mechanical sympathy, without mech sympathy you're going to wear the box FAST. Your
typical bike box clutchless shift is coming in at ~100ms ish as far as I can see, that's slow. You could add in automatic mech sympathy but then
you're going to slow it back down again, so all you're really going to gain is the loss of a few push-pull cables and about a £500 bill for
electronics and actuators.
That's not to say don't do it. But I wouldn't do it how you're suggesting, if I did it at all.
I would tell you my plans but it's a potential earner so I'll not
quote:
I would tell you my plans but it's a potential earner so I'll not