Electric clutch system

[Julian Thompson]

I'm playing with putting an F1 style steering wheel electric clutch in my car prompted by the fact that although my blip clutchless downchange system works well it does feel mechanically more sympathetic if you just "dab" the clutch at the same time...

 

...this led me to try and find some kind of electro-hydraulic device that can "pump" a bit of pressure into the clutch on downshifts...

 

...but I could not find anything that was close to this.

 

So, instead I found this...

 

http://www.active-robots.co.uk/ps105-power-servo-p-320.html

 

and this...

 

http://www.active-robots.co.uk/simple-servo-controller-p-851.html

 

which controls it.

 

Now look at the torque on that servo - that's serious power and very quick rotational speed allayed to excellent accuracy - this might be the basis for a superb DIY clutch system?

 

Any thoughts - do you know of an easier way to get what I really wanted (ie a quick automated stab of the clutch on downshifts) rather than going to the trouble of actually doing away with the clutch pedal completely?

[Molecular--Bob]

Saab used "Sensonic" to operate the clutch on their two pedal manuals in the 90s, might be able to adapt it.

[Mr Locust]

at about 1:20 on the video

 

Ian

 

Green and Silver Roadsport 😬

[Julian Thompson]

Amazing that Quattro tribute - worth watching actually in any event (my god they were fabulous, basic, brutal machines those weren't they!)

 

I don't know much about the Saab system except to say that I'm aiming for extreme lightweight so that's maybe why I was keen to do it "from scratch" - and there is another Servo in the same range here...

 

http://www.active-robots.co.uk/ps050-power-servo-p-321.html

 

...which is just 280g...

 

...so I guess that means there exists the possibility to do a "proper" semi-automatic system for actually the same weight as a manual clutch given that the manual system is using a "heavy" steel pedal !

 

 

 

This does feel like a bit of an odd project to start but somehow strangely appealing because it's quite "unusual"

[Colin Mill]

Hi Julian

 

By RC servo standards that is a very expensive unit and not amazingly powerful. Also 0.6s/60 degrees is quite slow (standard RC servos are around 0.2s/60 degrees). It might be worth looking at the Hitec servos. They do some titanium geared servos that are quite impressive. One thing you need to bear in mind is that RC servo manufacturers are rather naughty in specking servos - they quote no-load speed and stall torque so the servo you are looking at will do about 1.2s/60 degrees at 60kg.cm load. My guess is that you will need considerably more grunt from the servo.

 

If you want a really high-spec servo try looking at Maxon. They are a Swiss coreless motor manufacturer that do a very wide range of motors and gearboxes. Certainly they will have something to do what you need.

 

However you want to do it the electronics is now very simple and cheap.

 

Colin

[Julian Thompson]

Thanks Colin - that's really useful. Will have a look at Maxon and report back to base

[Julian Thompson]

Colin

 

Would you mind just revisiting those two links above and double checking the torque figures please?

 

I had a look at Hitec and the biggest they do is this:

 

http://www.hitecrcd.com/servos/show?name=HS-805BB

 

(24.7 KG/CM)

 

wheras "my" two servo's above had 90 odd KG/CM and over 100 KG/CM respectively ?

 

I had a quick look at Maxon but didn't understand the products in there... "Gearheads, brushless DC, compact drive etc..." - any guidance as to what might be a starting point in there?

 

Thanks for your help!

 

JT

[Colin Mill]

Here is one of the titanium geared hitec digitals. 30Kg.cm at 0.14s/60 degrees has the power output of a 0.6s/60 deg speed servo that does 128kg.cm so that one gives you a bit more grunt and under load might be a little nearer the desired speed but my guess is that more grunt still is required - I need to have a play with some spring balance and the clutch pedal to get an idea of what is needed.

 

I had a look last night at Maxon's revised website and It's a bit of a nightmare. I may resort to their printed catalogue which I have in the workshop somewhere. Once we have a figure for the speed and torque we need its not too bad to find a motor and gearbox combo that will give it. Compared with using an RC servo there is a bit more work needed as unlike the RC servo you have to engineer the position feedback externally. I have used a hall probe and magnets to do this and it works very well with no danger of a potentiometer failure. I have a prototype servo amplifier design that would do the electronics for you though the output mosfets might do with upgrading if the drive current went above 5 amps. Another possibility would be to monitor the motor current and, with the aid of a correction for the rate of acceleration of the motor measure the clutch pedal pressure as well as the position.

 

Colin

[Molecular--Bob]

How does your gear shift work? just wondering if you could piggyback a small mastercylinder onto the shift actuator. It might have enough spare capacity to dip the clutch as well, and could be plumbed like a hydraulic handbrake. would work on both up and down shifts though unless you got very tricky with the linkage.

 

edit:

 

I am assuming a hydraulic clutch. You could use a cam mechanism so that the cylinder only engages on the down shift

 

Edited by - Molecular--Bob on 6 Jul 2007 09:13:45

[Julian Thompson]

Ah - I think the penny is dropping Colin - so the torque figure AT a given speed is what we're interested in vs this "nonsense" figure they choose to quote "starting torque".....that is naughty - as you say!

 

Because I *think* the plan is to remove and do away with the clutch pedal completely the question is what is the most torque effective way to act on the master cylinder (about 1 inch total stroke I guess) using a rotating motor - and then I can mock that up and measure the load as you suggest.

 

There is such a device as an AutoClutch...

 

(http://www.vehvac.co.uk/brochures/autoclutch-leaflet.pdf)

 

...which seems to use a cable similar to a cruise control system to pull on the clutch pedal itself but this product would not (I guess) be anywhere near fast enough to give the desired "blip" of the clutch during a sub 100ms downshift 😳 and it also weighs in at 2 kg which is waaaay to heavy!

 

I really do prefer the idea of a direct action on the cylinder itself - much more businesslike ! Shame you can't just buy a "fly by wire" master cylinder 😬!

 

Thanks again Colin!

 

 

 

[Julian Thompson]

That's clever thinking Bob.

 

I guess you'd need a non return valve style arrangement to stop the *slave* master cylinder from backfeeding the *master* master cylinder and instead making it act on the clutch but that wouldn't really be a problem I don't think.

 

I think the difficulty there would be a lack of control of the clutch - we don't really know yet what relationship there is between a shift down and what your leg actually does when you get a smooth shift. I think it is just a "stab" but of course because driving is naturally programmed into us we kind of lose the ability to be objective about it. When I get the system set up I think it will be necessary to "fine tune" the clutch actuation and this is not going to be easy if the downshift and the clutch blip are combined? You definately don't want any clutch movement on upshifts as dipping the clutch even for the briefest of moments makes the whole shift less smooth than doing it clutchless and relying on the ignition kill.

 

Great idea though !

 

 

[Mike Molloy]

Hmmm. We do an electric park brake. I'll have to investigate the mechanism (hydraulic or not). Of course ABS uses electrical actuation of hydraulics as well.

[Julian Thompson]

Did think of that Mike but they're big beggers those ABS units?

 

Also I think the electronics might be beyond the scope of the project to integrate them?

 

That park brake sounds interesting though?

 

Do you have any details to hand on that?

[Julian Thompson]

Da Daaaaaaaa!

 

And..

 

Graham Ford comes to my rescue - thanks a million!

 

AP Racing do a part number CP7950 which is a power clutch actuator designed EXACTLY for what I'm trying to do with it.

 

I have a copy of the AP catalogue page here after Graham kindly scanned it in for me but can't find any more info on the web at all....

 

...anyone know any more about usage / cost etc?

[Colin Mill]

Well, I just did some sums based on a Maxon motor gearbox that I have played with recently. The combo weighs in at 700grams and has a stall torque at 14v of 125 kg.cm and a no-load speed of 0.023s/60 degrees. It pulls a starting current of 42 amps!. Its about 25 times more powerful than the biggest of the RC type servos we have so far looked at. Optimum power will be at about 62kg.cm torque and 0.046s/60 degrees (and about 20 amps draw). With a 3cm arm this would give you a 20kg force with a throw of 30mm in under 50ms.

 

I guess that to the critical part of the disengagement of a clutch is over a stroke of about 50mm and a force of about 20kg. So this motor/gearbox would be in the right parish.

 

Colin

[Julian Thompson]

Wow!

 

What sort of £££ is the motor Colin and how easy can we control it?

 

 

[Colin Mill]

Hi Julian

 

I'll have to give the Maxon rep. a bell on that one - I was not the one that had to pay for the one I was tinkering with - which is nice when you are wringing the nuts off something 😬. I must confess it did ooze quality! (we need a sado geek smiley)

 

I suppose I'm sticking my neck out here but controlling it is fairly straightforward. I have produced a couple of circuits that would both form the basis of what is required. In both cases they have an "H" bridge of mosfets controlled by a single PIC microcontroller. With the aid of a pot or hall probe for position sensing we would be in business. I guess the most difficult bit would be to code the clutch let-in for starting from rest. It would be useful to know what it would have to interface with in terms of the gearbox control and engine control.

 

Colin

 

Edited by - Colin Mill on 6 Jul 2007 22:50:29

[Julian Thompson]

RIght I seeeee.

 

OK.

 

The brief, then.

 

Basically I think that there are 2 requirements here and maybe an additional one which would, possibly, overcomplicate things 😬

 

1) The blip on downchange. Firstly, the ideal circuit would accept a negative pulse in from the "down" paddle and send an adjustable length of signal to the gearbox actuator system in the form of a "down wire" to "common wire" link.

 

At the same time, the clutch servo would receive an adjustable length of pulse and maybe amplitude? to create our smooth shift.

 

The upchange would of course contain the "up" to "common" to the shifter system as an adjustable pulse from the negative input on the paddle BUT it would not contain a clutch component as I have proved this unnecessary for a perfect shift

 

2) Taking the system to the next level the steering wheel would also be fitted with a paddle connected to a potentiometer which *simply* controls the clutch between the two extremes of travel as per normal. ie the driver still drives the car and deals with biting point, clutch wear, slip etc etc etc... but the clutch pedal is removed completly.

 

3) You could take this the whole way and make it a proper "ferrari" style system that detects clutch rub and does everything for you. I think the logic for this would be more of a job and to be honest is not necessarily desirable.

 

I am not even 100% sure that the 2nd level of control (ie the hand clutch) is the way forward except that it would give exceptional control of a BEC clutch which we all know can be tricky in the heat of the moment.

 

The motor sounds good though Colin.

 

If you send me your email address I'll send the datasheet that Graham sent me for the AP system since this also looks wonderful but would likely still require some control circuitry.

[Colin Mill]

OK - Thinking about that requirement I could fairly easily arrange the PIC code to accept an input from a pot to give the manual clutch control so that the servo position tracked the pot position with some adjustable parameters to get the feel the way you want it (almost all my PIC devices have a PC interface to allow internal parameters to be adjusted). So this would allow you to manually depress the clutch, select first and launch as usual. For the down shifts the PIC could have a separate logic input that when set true would cause the servo to drive to the disengage point regardless of the input from the manual clutch position input. This could be made adjustable so that the travel was the minimum required to just disengage the clutch and make the shift time as short as possible. If needed for synchronising the system the PIC could have a logic output that went true when the clutch reached the disengagement point. Would the system require the clutch re-engagement at this point to be profiled or would you want it to re-engage as fast as possible?

 

I guess an auto launch control would be reasonably easy depending on just how much of the launch it had to control (i.e. responsible or not for monitoring wheel-spin). We do an engine speed governor for model helicopters that has to hold head-speeds constant through violent aerobatics and that can detect speed losses that are equal to the blades lagging behind their ideal position by a fraction of a millimetre while travelling at 100m/s so keeping an eye on the engine rpm and feeding the clutch in as needed would be a similar sort of problem.

 

Colin

 

Edited by - Colin Mill on 7 Jul 2007 08:38:12

[Julian Thompson]

I did a lot of model helicopters in the early 90's when 3D aerobatics was just becoming possible with the advent of U shaped throttle curves programmable on the radio - that head speed problem as you transitioned from one maneuver to the next was a nightmare and always looked messy! I never dreamed that there would be a head speed RPM control ! Amazing!

 

 

 

 

😬

 

Anyway:

 

"Would the system require the clutch re-engagement at this point to be profiled or would you want it to re-engage as fast as possible?"

 

- This is the key point, actually.

 

A quick drive out this morning reveals that at low and medium speed the answer is "no" - and at high speed it is "not quite"....

 

... *idea* *idea* *idea* *idea*

 

BECAUSE THAT MUST BE THE REASON WE'RE HAVING THIS CONVERSATION!!

 

If both clutch components are spinning at the same speed there is no need to "rub" the clutch at all and the downshift would be beautifully smooth ! Still, with no clutch it must be more difficult for the blip unit to alter the engine speed obviously.

 

If the system could watch engine rpm in the expected engaged gear it could wait until the engine speed and wheel speed was the same and then the answer would ALWAYS then be "yes" - you want the clutch to re engage at any speed....this WILL happen because we have a throttle blip - it's just a question of timing!

 

goodness me.

 

The missing link I guess is engaged gear - although we can derive that from the wheel speed/rpm and if it knows you were in third and you've just pressed "down" shift then it is logic that you should now be comparing propshaft rpm to engine speed using gear 2's ratio....

 

....and furthermore, we already HAVE a throttle blip so the key is just to monitor the engine rpm in the EXPECTED engaged gear and then dump the clutch when the two rpm's match!

 

Am I making any sense?

[Julian Thompson]

Colin - just reread your post and I think you had worked all that out, hadn't you -

 

"keeping an eye on the engine rpm and feeding the clutch in as needed would be a similar sort of problem."

 

Still - helps for me to understand the issue as well 😬

 

[Colin Mill]

Hi Julian - Do you already have the fly-by-wire throttle sorted?

 

Colin

[Colin Mill]

Just been kicking the synchronisation thing about a bit and wondering how best the sensing could be done. It would be possible to put a hall sensor to monitor the rotation speed of the input shaft of the gearbox which on the face of it is the logical place to take the 'target' rpm from However, it could be quicker (depending on the time taken for the box to change ratio) to use the signal from the speedo sensor together with the system having knowledge of the ratio of the gear you are heading into to provide a target rpm value before the box changes gear. In some ways I like the second method - less messing about with fitting additional sensors and a small technical advantage balanced against a slightly greater complexity to the control code and a need to 'educate' the system about ratios for each gear (which it could self-learn anyway)

 

Colin

[metal mickey]

Hi guys,

been watching this thread with interest after talking to Julian last week. I have a bec with a 'blip' downshift system. As I understand it the blip, which is undetectable at

higher rpm is to unload the pressure off the gearbox dogs as you are on a trailing throttle and allow the easy and seamless downchange to take place. I'm wondering if you introduce a clutch into this system will the blip actually unload the dogs or merely spin the clutch and result in crunched gears 🤔 Or could the clutch stab you are thinking of be timed to occur milliseconds after the actual gerachange 🤔

 

[Julian Thompson]

"Hi Julian - Do you already have the fly-by-wire throttle sorted" -

 

Not exactly Colin - Mike (metal mickey) and I both have "trickshifter" blip modules that blindly squirt the throttle when they receive a power source.

 

Mike - the idea IS to time it, yes. Just been out in the car and downshifts at road speeds are definately smoother in mine with a very quick jab of the clutch - and I dont mean a proper Toad Of Toad Hall style push on the clutch - I mean the lightest possible brush - and that isn't easy to get right every time. You have experience though of the car on track which I dont have yet so it'd be interesting to see how your car likes a very quick jab of clutch on the way down ??