Cable Steering Design Methodology & DFMEA

[Kealan O Carroll]

Okay so 2 big questions: Design approach and FMEA w.r.t. cable steering.

Design: My approach was as follows:

Assume the car steers around one point, along a line projected from the rear axle, where it intersects with a line perpendicular to each of the front wheels. The inside and outside wheels need different angles to both be co-incident with this point, hence the ackerman effect.

Went into excel, and plotted a table of inside wheel angle, outside wheel angle, and corner radius.

Consult the rules and find the minimum corner radius, then add a kind of F.O.S. so it's designed to turn a little tighter.

Work out a steering wheel angle : Wheel angle ratio. The inside wheel needs to go to 45 degrees, to I kept it simple and used a 180 degree lock-to-lock steering wheel range, for a 90 degree lock-to-lock wheel angle range.

I've then taken the offset from the KingPin axis to the tie rod end, and used Trig to work out the effective rack movement left / right needed per 2 degrees of wheel movement. I've then plotted these increments against the corresponding steering wheel angle.

Using the steering wheel angle and the required steering rack movement, design a cam shaped cable pulley based on arc length segments.

This sound okay to all of yous ?

The second bit is an FMEA required for cable steering. I have a template from work for DFMEA, so i've taken that and filled it all in, but to be honest I dont know if its good enough / what the judges are looking for. Has anyone ever submitted an FMEA and gotten feedback ?
I have the excel file here if anyone would be willing to take a look...

[NickFavazzo]

So how will the front and rear slip angles affect the vehicles turn radius?

What ackermann do you want to run? You described 100% but do you want pro or anti? that will change the difference between the two wheel steered angles.

I would also think about the steering ratio.. we around 300 degrees lock-lock at the steering wheel. I would be careful to not put excessive forces on the steering wheel by running the wrong ratio, you dont want the drivers getting fatigued...

[Kealan O Carroll]

Was planning on running pro.
The conditions for anti ackerman to increase grip is at "high" vertical loading, so I'm making the assumption that a relatively light relatively slow FSAE car on a relatively tight track wouldn't see the vertical loads high enough to justify Anti ackerman ?

How do your drivers find 300 degrees?
That 180 degrees lock-to-lock was playing on my mind, but I was thinking it could be very awkward for a driver to cross his/her arms over eachother ?

Can the slip angle be assumed to increase linearly as the steering angle / wheel turning force on the tire bead increases ?

[Dash]

I'll chip in on the lock-to-lock comment.

Our 2010 car had a very quick ratio and was about 180 lock to lock. Your arms turn to jello after about 6 minutes of driving, and our endurance drivers had to slow down towards the end to make it around corners because they were so tired.

2012 car is about 270 degrees lock-to-lock with a slightly slower ratio and I've driven it for hours on end without ever getting tired.

Picking the right ratio is definitely a must .

[Racer-X]

I would also like to comment on using the minimum corner radius to define your turning circle. Did you account for slip angles and any possible over/under steer?

As a driver I hit steering lock on occasion in our car and I can tell you our turning circle is tighter than that.

[Kealan O Carroll]

As it is, the min corner radius (to the kerb/cones) is 3 metres, disregarding slip angle / understeer

[NickFavazzo]

Originally posted by Kealan O Carroll:
Was planning on running pro.
The conditions for anti ackerman to increase grip is at "high" vertical loading, so I'm making the assumption that a relatively light relatively slow FSAE car on a relatively tight track wouldn't see the vertical loads high enough to justify Anti ackerman ?

How do your drivers find 300 degrees?
That 180 degrees lock-to-lock was playing on my mind, but I was thinking it could be very awkward for a driver to cross his/her arms over eachother ?

Can the slip angle be assumed to increase linearly as the steering angle / wheel turning force on the tire bead increases ?

Pro or anti, what is best depends on plenty of things. If your going for peak lateral grip for example the inside and outside want to be at different slip angles. High loading is relative, the tyres can see huge variations in normal load due to the small wheelbase and high cornering forces as such ackermann can definitely have an effect.

180 lock to lock can work provided your rack ratio and other mechanical advantages in the steering system account for it...
Most of the time the drivers don't need to cross their hands but on a tight corner you can have a double grab, as a driver myself I have no issue with it, it is better to be a little uncomfortable in one corner than exhausted from being uncomfortable in all corners.
You can have systems with a far-from linear steering ratio, benefiting both low steering angles and high steering angles.
Have a look at cornering stiffnesses for how much a steered angle will induce a slip angle maybe.

[Z]

Kealan,

"Assume the car steers around one point, along a line projected from the rear axle, where it intersects with a line perpendicular to each of the front wheels. The inside and outside wheels need different angles to both be co-incident with this point, hence the ackerman effect...."

Actually the centre of the turn is a little in front of the rear axle line (because of rear tyre slip-angles), but the above is close enough. Roughly, you want the inside wheel at 45 degrees and outside at 30 degrees at full lock. However, cable steering (see below) can easily get much higher angles, which is always useful (eg. you spin and end up pointing backwards on the narrow track, with cones either side...).

I agree with the others who suggest that ~300 degrees for the steering wheel is better than 180. Light steering with 90 degrees lock-to-lock would be great, but that's power-steering and another story (quite feasible http://fsae.com/groupee_common/emoticons/icon_smile.gif ). For easy steering use "centreplane" steering (zero KPI and Offset) with smallish Trail (~0-2cm).

For ackermann I suggest you start with "100%" (as in your above quote), and then go up from there in testing (ie. more toe-out). Note that conventional R&Ps can only get 100% ackermann at straight-ahead and one other point in the range (either side), and are mostly well away from 100%. The inside wheel also gets close to going "over centre" at full lock, so the linkage loses toe-stiffness (because steer-arm and tie-rod almost straight).
~~~o0o~~~

"Using the steering wheel angle and the required steering rack movement, design a cam shaped cable pulley based on arc length segments."

This sounds like you are using the cables to drive a rack, which is then connected to the wheels via tie-rods?

This leaves you with many of the disadvantages of a rack (like friction, "rack-rattle", and as above). Why not take the cables all the way to the wheels?

You could have a half-pulley fixed to each upright (~10cm radius) with two cables going from the ends of this pulley, via some idler pulleys, to the cam shaped pulleys on the steering shaft. This gives constant toe-stiffness at all steering angles, and allows larger wheel angles than the conventional linkage. Spring preload somewhere in the cable runs gets rid of backlash. Easier explained with some sketches...
~~~o0o~~~

"The second bit is an FMEA required for cable steering..."

I have no idea what sort of paperwork is required here. However, the "cable to the wheels" I have in mind would (IMO) have fewer failure modes than most FSAE steering systems. For instance, the only way to lose control of BOTH wheels would be via a failure between the driver's hands and the pulleys at the end of the short steering shaft (no UJs, etc.).

Z

[Kealan O Carroll]

Originally posted by Z:
Kealan,

"Assume the car steers around one point, along a line projected from the rear axle, where it intersects with a line perpendicular to each of the front wheels. The inside and outside wheels need different angles to both be co-incident with this point, hence the ackerman effect...."

Actually the centre of the turn is a little in front of the rear axle line (because of rear tyre slip-angles), but the above is close enough. Roughly, you want the inside wheel at 45 degrees and outside at 30 degrees at full lock. However, cable steering (see below) can easily get much higher angles, which is always useful (eg. you spin and end up pointing backwards on the narrow track, with cones either side...).

I agree with the others who suggest that ~300 degrees for the steering wheel is better than 180. Light steering with 90 degrees lock-to-lock would be great, but that's power-steering and another story (quite feasible http://fsae.com/groupee_common/emoticons/icon_smile.gif ). For easy steering use "centreplane" steering (zero KPI and Offset) with smallish Trail (~0-2cm).

For ackermann I suggest you start with "100%" (as in your above quote), and then go up from there in testing (ie. more toe-out). Note that conventional R&Ps can only get 100% ackermann at straight-ahead and one other point in the range (either side), and are mostly well away from 100%. The inside wheel also gets close to going "over centre" at full lock, so the linkage loses toe-stiffness (because steer-arm and tie-rod almost straight).
~~~o0o~~~

"Using the steering wheel angle and the required steering rack movement, design a cam shaped cable pulley based on arc length segments."

This sounds like you are using the cables to drive a rack, which is then connected to the wheels via tie-rods?


You could have a half-pulley fixed to each upright (~10cm radius) with two cables going from the ends of this pulley, via some idler pulleys, to the cam shaped pulleys on the steering shaft. This gives constant toe-stiffness at all steering angles, and allows larger wheel angles than the conventional linkage. Spring preload somewhere in the cable runs gets rid of backlash. Easier explained with some sketches...
~~~o0o~~~



Z

Sorry, there is no rack, its "cable to upright", i've just used the steering rack movement or tie rod movement expression so people will get what I mean, its really the change in effective cable length from the column to the upright !
The system I've designed entails 4 cables and 8 half - pulleys mounted just behind the steering wheel.
There's a pulley for the wheel on the inside of the corner, and a different one for the wheel on the outside. These are then arranged to make up 4 asymmetric pulleys, one for each "Corner" of the front wheel system.

I just have one possible problem to check now, as a wheel turns, the effective cable length increase at the front of the upright must be the same as the effective decrease on the back side of the same upright, to keep the tension on the cable and the wheel angle constrained...since the cables are running over different asymmetric pulleys I'll need to double check this one!

A mild steel steering column, aluminium pulleys, and a big beefy aluminium support brace is weighing in at just over 700g on solid works, its taking up about 120mm wide x 100mm long x 80mm deep, so it should slot nicely in behind the dash

[Z]

Originally posted by Kealan O Carroll:
there is no rack, its "cable to upright"
...
The system I've designed entails 4 cables and ...
4 asymmetric pulleys, one for each "Corner" of the front wheel system.
Kealan,

That sounds pretty much like my thinking.

But... at the upright, are you going to attach the two cables via two BJs on the ends of two "steer-arms"? This (like a conventional linkage) gives reduced toe-stiffness at large steer-angles, because the offset of cable-from-steer-axis becomes less.

My thinking is to have a half-pulley at the upright, which gives constant delta-steer-angle per unit linear-cable-movement, and thus constant toe-stiffness. This is made possible by an idler pulley just inboard of the upright/half-pulley...
~~~o0o~~~

PRACTICALITIES of CABLE STEERING...
================================

When you are finished with your digital tinkering, I strongly suggest you build a real mock-up of the system. You only need to do this from the steering wheel out to one upright, preferably with some simulated suspension movement. Fit a heavy mass as the "road wheel" so you can "shake" it with the steering wheel and note any flex or slop in the system. Photos of this mock-up might (?) also help in the early days of dealing with officials.

For the cables and their terminations you might try yacht suppliers, or these days architectural suppliers. Both do stainless steel cables and swaged fittings (although you don't need SS). Make up a few dozen short cables+swaged fittings, and test them all to destruction (good fun http://fsae.com/groupee_common/emoticons/icon_smile.gif). Incorrect/inconsistent swaging technique can be a problem...

You might also get off-the-shelf idler pulleys from the yacht suppliers. Otherwise these can easily be made from aluminium, or even nylon, running on sealed DG ball-bearings. Make the pulley grooves quite deep and use close fitting sideplates to prevent the cable jumping out of the groove.

One cable run per wheel should be spring-loaded to eliminate backlash. This can be done with a compression spring used in a similar manner to tent guy-ropes. I would try a spring preload of 50-100kg (???), with maybe 1-2mm movement before coil-bind.

Again, IMO, all the details are best resolved with lots of iterations on a physical mock-up. http://fsae.com/groupee_common/emoticons/icon_smile.gif

Z

PS. Does anyone have toe-stiffness values for their (conventional) steering linkage? Ie. with steering wheel fixed, vary steer torque on road wheel (Mz) and measure delta-steer-angle. Also, any estimates for how much this drops off at large steer-angles? My experience is that it drops off dramatically, especially with worn BJs... http://fsae.com/groupee_common/emoticons/icon_frown.gif

[Kealan O Carroll]

Originally posted by Z:

My thinking is to have a half-pulley at the upright, which gives constant delta-steer-angle per unit linear-cable-movement, and thus constant toe-stiffness. This is made possible by an idler pulley just inboard of the upright/half-pulley...



When you are finished with your digital tinkering, I strongly suggest you build a real mock-up of the system. You only need to do this from the steering wheel out to one upright, preferably with some simulated suspension movement. Fit a heavy mass as the "road wheel" so you can "shake" it with the steering wheel and note any flex or slop in the system. Photos of this mock-up might (?) also help in the early days of dealing with officials.

For the cables and their terminations you might try yacht suppliers, or these days architectural suppliers. Both do stainless steel cables and swaged fittings (although you don't need SS). Make up a few dozen short cables+swaged fittings, and test them all to destruction (good fun http://fsae.com/groupee_common/emoticons/icon_smile.gif). Incorrect/inconsistent swaging technique can be a problem...

You might also get off-the-shelf idler pulleys from the yacht suppliers. Otherwise these can easily be made from aluminium, or even nylon, running on sealed DG ball-bearings. Make the pulley grooves quite deep and use close fitting sideplates to prevent the cable jumping out of the groove.

One cable run per wheel should be spring-loaded to eliminate backlash. This can be done with a compression spring used in a similar manner to tent guy-ropes. I would try a spring preload of 50-100kg (???), with maybe 1-2mm movement before coil-bind.



Thanks for the input Z !
Finding it tough enough with nobody to benchmark !

I know what you mean by keeping a constant delta wheel angle per unit length of cable movement, but I don't get how you'd do it with the pulleys at the uprights ?
The way it is in my head at the moment is the cable end slotted into a machined holder on the wishbone, much like a bicycle cable boss sticking out of the frame. With the cable mounted into the upright w/o any pulleys etc, this would decrease the effective offset as the wheel angle increases, so the steering would effectively become faster / more twitchy and at the same time require more force / become "heavier" yeah ?

Re the spring. It wouldd (Should :P ) take up any backlash in the system but i was worried about it making the steering feel vague...is this why your 50 - 100 kg preload figure is so high ? If the spring constant was significantly bigger than the force on the cables then the wheels should move long before the spring starts to move, so it shouldn't take from the feel of the system ?

[Z]

Kealan,

"... but I don't get how you'd do it with the pulleys at the uprights ?"

Oh... for a simple sketching facility!!! (Maybe later....)
~~~o0o~~~

"The way it is in my head at the moment is the cable end slotted into a machined holder on the wishbone, much like a bicycle cable boss sticking out of the frame.'

This sounds as if you are running the cable through an outer (flexy tube) housing, like a "Bowden" cable? My thinking was to have the tension cables running over, and guided by, various pulleys (so with no outer housing), like used on countless yachts, and quite a few aeroplane control systems. The Bowden cable style can work but will have more friction, so minimise the number of bends and the preload (because F1/F2=e**(Theta*Mu)).

The wheel-end of this cable can be fed to a half-pulley on the upright, a lot like the throttle control cable on most modern cars.
~~~o0o~~~

"Re the spring. It would (Should :P ) take up any backlash in the system but i was worried about it making the steering feel vague...is this why your 50 - 100 kg preload figure is so high ? If the spring constant was significantly bigger than the force on the cables then the wheels should move long before the spring starts to move, so it shouldn't take from the feel of the system ?

Yes. Below the spring preload the system moves with no backlash. Above the spring preload, say hitting a curb or pothole, the "cable" stretches a bit, but only by a predetermined amount set by the "distance to coilbind" (spring is working in compression). The spring is only there to prevent big variations in load (from too slack to too tight) that might result from inaccurately machined pulleys, etc.

Z

[Kealan O Carroll]

Originally posted by Z:
Kealan,

"... but I don't get how you'd do it with the pulleys at the uprights ?"

Oh... for a simple sketching facility!!! (Maybe later....)
~~~o0o~~~

"The way it is in my head at the moment is the cable end slotted into a machined holder on the wishbone, much like a bicycle cable boss sticking out of the frame.'

This sounds as if you are running the cable through an outer (flexy tube) housing, like a "Bowden" cable? My thinking was to have the tension cables running over, and guided by, various pulleys (so with no outer housing), like used on countless yachts, and quite a few aeroplane control systems. The Bowden cable style can work but will have more friction, so minimise the number of bends and the preload (because F1/F2=e**(Theta*Mu)).

The wheel-end of this cable can be fed to a half-pulley on the upright, a lot like the throttle control cable on most modern cars.
~~~o0o~~~

"Re the spring. It would (Should :P ) take up any backlash in the system but i was worried about it making the steering feel vague...is this why your 50 - 100 kg preload figure is so high ? If the spring constant was significantly bigger than the force on the cables then the wheels should move long before the spring starts to move, so it shouldn't take from the feel of the system ?

Yes. Below the spring preload the system moves with no backlash. Above the spring preload, say hitting a curb or pothole, the "cable" stretches a bit, but only by a predetermined amount set by the "distance to coilbind" (spring is working in compression). The spring is only there to prevent big variations in load (from too slack to too tight) that might result from inaccurately machined pulleys, etc.

Z

Yeah Bowden cables were what I had in mind !

The friction in bowden cables is pretty minimal when they're new / properly maintained...some of this higher end ones are teflon coated, so along with some teflon lube they're pretty free-moving.
The concern I'd have with running a cable inner only is that I'd need 3 or 4 idler pulleys per cable, all of which would need a stiff mounting braket, bushing, etc, and it's a lot harder to avoid bump steer without an outer cable