Steering with Pitman Arm.

[CWA]

Charles, naturally I'm talking about FSAE cars. Two things stand out for me that I believe make the case of your kart different to an FSAE car:

1. You say your kart steering wheel is around 375mm = 15". IIRC our FSAE steering wheel was close to 8" diamater, I assume most other FSAE cars are similar to this. Again, ignore steering effort, but consider the 'response' of the vehicle to be it's Ay as an output caused by steer displacement as an input. Except this time consider tangential displacement of the steering wheel rim at the driver's hands, or driver arm displacement, rather than angular displacement of the HW / column. For any given vehicle, if you fitted it with A) the FSAE hand wheel, then B) the kart hand wheel, I believe the effective response of case A could be considered as double that of case B.

So, perhaps your kart does have favourable 'response' levels with a pitman arm and less than 90 degrees total lock. But if you tried to achieve this with an FSAE hand wheel of half the size, the vehicle may be deemed too responsive by any normal human pilot.

2. In my experience, karts (vs FSAE) do not have to accommodate such a range of vehicle speeds / corner radii (and hence range of steer angles). This is an issue for FSAE cars as per my previous reasoning - you have a limited amount of HWA you can utilise with a pitman arm, the max of which has to deliver quite a large amount of steer angle for a slow hairpin, meaning your steering response for the high speed corners is perhaps too high. The equivalent high-speed corners for a kart may be long sweepers and may be more forgiving than short duration FSAE slaloms / gates for a driver trying to modulate his steering system which is highly sensitive? I do not know if this is true for your kart / race series, perhaps your range of speeds are more equivalent to FSAE events than I realise. I wonder if you have any Ay vs Steer data to give insight into this aspect?

Kevin, that system sounds very interesting. I'd love to see some pictures / CAD if you happen to have any you'd be willing to share?

[mdavis]

The University of Cincinnati for the last 2 years has run a pitman arm steering system, but with no gear reduction of any kind. U-joints were used to get the steering wheel angle at an appropriate position, compared to a straight shaft. The 2014 car did not run, but I had a chance to drive the 2015 car at a test day this past summer. Beyond the spring/damper setup issues that the car had, the steering was incredibly heavy. It had a very fast ratio (slaloms were accomplished with ~+/-10-15 degrees of HWA, if that), but the weight was way too high to be fast over an endurance length. I would have struggled to drive the car flat out for 2 autocross laps at MIS, it was so heavy. More seat time with the car, and training would have helped a lot, but the steering was brutally heavy. More rear weight bias, different steering ratio, and/or some rear steer could have mitigated a lot of the design issues, IMO.

Having driven a car with a pitman arm setup, I think it is very possible to get a lighter than RNP solution in an FSAE car that provides appropriate weight and steering response. I would venture that neither of the UC setups were all that well analyzed or very well executed, and because of this, the team has (sadly) gone back to a RNP for 2016.

Kevin, I would also be interested to see any pictures/CAD you're willing to share of the ECU steering setup. No need to share the tuning details, unless you/the team want to.

[Z]

Ahmad,

Provided is a sectional view from our Titan R&P.
...
Total Cost -If i do remember- was $1100...

Well, at least for $1100 you get a spring-preloaded rack. But ...

If the bushings in the end of the housing are a close fit on the rack, then the spring simply pushes the rack against those bushings, and, when the R&P teeth wear "at centre", ... you still have slop-at-centre!

On the other hand, if the end-bushings are made loose enough to allow the sprung rack to be pushed against the pinion's worn teeth for no slop, and also move away from the unworn teeth for no bind, then ... you always get rack-rattle!

Ahh, ... $1100 ... because racecar!
~~~o0o~~~

Doug,

... this could be a lightweight solution that also allows a lot of packaging flexibility?
... see "Fig. 27.49. Split track-rod with relay-rod and idler steering linkage layout".

Yes, the idea of having two "idlers" has significant advantages, though not obvious.

Consider two "idlers" (aka bellcranks/relays/???) mounted at either side of the upper-frame, and pivoting about vertical axes. The lower end of each idler has a Pitman-Arm that drives a short steering-tie-rod going out to the wheel, with this tie-rod at the most suitable height between lower and upper wishbones. The upper end of each idler is at steering-Hand-Wheel-height and is driven from the HW by any suitable means. One way would be via a R&P mounted near the top of the FRH/footbox.

But I would drive the idlers via cables! Fit multi-groove pulleys at the end of the (short) HW-steering-shaft, and also at the top of the idlers, then use multi-strand (7 x 19?) steel cables about 3 to 5 mm diameter to transfer forces back and forth. Of course, you need two sets of cables, one pulling each way. Or use roller-chains and sprockets as on Bill's link. See also speedboat steering systems, or the many (small to very big) aeroplanes that use similar.

One advantage here is that by making the pulleys non-circular it is very easy to get a large range of different Ackermann behaviours and/or "variable ratio steering". In essence, you have two separate variable-ratio steering systems, one for each wheel. So set a little static toe-in for stability on the straights, then get the inner-wheel to turn fastest for early and large "dynamic toe-out" for lightning fast turn-in. All this is very lightweight, easy to make, robust, and slop-free (because you spring-loaded the cables). This is why it is so common on planes.

But the really big advantage is that the chassis floor can now be flat and obstruction free from driver's bum all the way to Front-Bulkhead. No need for the common "stepped-floor" that is necessary for the R&P to be at right height wrt wheels. Lowering the floor also lowers the driver's quite heavy lower legs, and lowers the whole of the footbox structure, including quite heavy pedal-assembly, FB, and IA. (BTW, I believe this "lowering of footbox" is the major rationale for ECU's system.)

So, more complexity by adding the two idlers, but overall a simpler chassis to design and build, and a significantly lower CG, and possibly less total mass.
~~~o0o~~~

Feasibility of PURE PITMAN-ARM Steering.
=================================
The main disadvantages I see with pure PA (as also noted by Mdavis above) are;

* Because of the various practicalities of the full linkage (ie. Ackermann, bump-steer, etc.), only about +/- 60 degrees of steering-HW-Angle is possible (ie. HW turns 60 degrees from centre to full-lock-one-side). Better would be HWA = +/- 45 degrees only. But this could mean too high driver arm-loads, and/or too fast car response to small hand movements.

* For typical FSAE location of driver wrt front-axle-line, the plane of the HW might be too horizontal, or "bus-driver-ish" (ie. near vertical steering-shaft).

But I reckon all these problems can be overcome. Taking them in reverse order;

* A car that has the driver entirely within the wheelbase (ie. feet on or behind the front-axle-line) can have the PA in front of driver's feet and at a suitable height that gives the right ergonomic angle for the HW. And "driver-inside-wheelbase" is generally a good thing for FSAE dynamics, because it gives more R%, lower Yaw-Inertia, simpler, lighter chassis, etc.

* The "too fast car response" is not a problem at all, IMO. In fact, it is what you should be aiming for!

Total car response time = time for driver to initiate the action + time for car's tyres, etc., to respond and make the car move. Reduce the driver-action-time to zero and you still have to wait for the tyres and the rest of the car to do their thing. That can still take a long time for a car with soft tyres, large Yaw-Inertia, etc.

Consider also that modern fighter jets are controlled by a rigid side-stick that barely moves at all. It is strain-gauged and responds to the pilot's hand forces, rather than responding to large motions. But to give the planes really fast response times the designers had to make them dynamically unstable as well (ie. negative "static margin"). Just reducing pilot-action-time to zero is not enough.

* So the problem boils down to finding a way to lower the HW forces required for such quick steering of, say, +/- 45 degrees of HWA. Simply moving the driver, and hence also the car's CG, rearwards, is a big step in that direction. Making the whole car lighter also helps. And centre-plane steering geometry helps a lot, with just enough Trail for the right feel (many FSAE cars already have this, so it is certainly "feasible").

But the "guaranteed to work" solution is found in almost all modern vehicles. I have a book (picked up cheap at school fair) that lists the specs of "All The World's Cars -1964". It lists many big American cars with manual steering that have at least 5 turns lock-to-lock! No modern car has steering that slow, because nowadays they all use POWER-STEERING! Even the very small cars. Even some Quad-bikes these days come fitted with Electric-Power-Steering. And a Quad-bike's handle-bars only turn about +/- 45 degrees. Hmmm...

But all things considered (ie. total mass, electrical load, speed of response, ++), I reckon a small car's hydraulic-PS would be the best choice. Quite easy. Fit an off-the-shelf belt-driven PS-pump+tank on engine, two hoses to an (off-the-shelf) valve-in-steering-shaft, and then an appropriate hydraulic-actuator on PA. The return you get from this investment is lightning quick steering that requires almost no effort from the driver, and is all wrapped up in a simple mechanical linkage.

Worst case, you remove all the hydraulics, reduce the Trail to ~5-10 mm, and slap the Enduro driver on the back and say, "You can do it!!!"

(FWIW, I once bought a 30 ton front-end-loader to help me do the gardening. It had a steering-box with about 3+ turns lock-to-lock. This broke one day (it was a conventional 'box like in Doug's link, but the balls managed to escape the worm, and then break other stuff). Of course, given the loader's weight and its "bend-in-the-middle" steering (ie. picture two 15 ton lumps mutually pivoting about a single, central, vertical axis), it had to have power-steering (... via 2 x ~5"-diameter-rams!). So I threw the steering-box away and replaced it with a homemade Pitman-Arm welded to the end of the steering-shaft. Yep, +/- 60 degrees HWA, and I had the biggest go-kart in the neighbourhood! Worked great, much better than the original. )

Z

[apalrd]

But the "guaranteed to work" solution is found in almost all modern vehicles. I have a book (picked up cheap at school fair) that lists the specs of "All The World's Cars -1964". It lists many big American cars with manual steering that have at least 5 turns lock-to-lock! No modern car has steering that slow, because nowadays they all use POWER-STEERING! Even the very small cars. Even some Quad-bikes these days come fitted with Electric-Power-Steering. And a Quad-bike's handle-bars only turn about +/- 45 degrees. Hmmm...

But all things considered (ie. total mass, electrical load, speed of response, ++), I reckon a small car's hydraulic-PS would be the best choice. Quite easy. Fit an off-the-shelf belt-driven PS-pump+tank on engine, two hoses to an (off-the-shelf) valve-in-steering-shaft, and then an appropriate hydraulic-actuator on PA. The return you get from this investment is lightning quick steering that requires almost no effort from the driver, and is all wrapped up in a simple mechanical linkage.

Why not electric power steering?

There are a number of quads and snowmobiles and such that use electric power steering, as well as a large number of production road cars. Less parts, less weight, and less interference with the engine.

[CWA]

mdavis - very interesting to hear of your experience with pitman arm to column in fsae. Perhaps in your case the driver would have actually been fine with the response if the effort wasn't so great after all. What kind of steering axis geometry did your car have, out of interest?

Z, the strain-gauged flight control column is interesting to learn about, I am keen to read into this further. R.e. your comments on response, I had not considered a lack of steer ratio to cause what I’ve heard referred to as a ‘transient response’ issue, as it sounds you are describing. Ignoring transient input / output delays (which I agree, are really not much good at all), I believe the steady-state Ay/Steer gain can often still be too much. This is what I was referring to in my previous posts. Do you not think this could be the case with a lack of steer ratio? Of course this does not tie in with how your aforementioned fighter pilots now manage to control jets with only column force, not displacement. Perhaps the difference is that an fsae car will never be able to actually achieve a linear relationship between input torque and output Ay / yaw rate response (too much hysteresis?), so a driver always has to rely on displacement correlating with response for intuitive control?

Anyway I agree that with cooperation from other designs areas of the car, a steering axis could be designed to reduce steering efforts to acceptable levels with pitman arm to column.

[Kevin Hayward]

I have attached a GA of the first steering version as requested. This years is simpler, stiffer and lighter.

As Z indicated the initial reasoning was to reduce COG height by lowering the nose and driver. These were lowered by around 100mm. Note that this is a larger effect on COG than lowering an engine with a dry sump (while being simpler and lighter).

However through iteration and tuning it looks like what can be done with steering geometry is the larger advantage.

Excuse me not putting too many details of the design, but it is fairly straightforward to work through with some thought. Planetary gearboxes are not too hard to make and is probably not a bad call for someone wanting to go close to a pure Pitman arm while not having to deal with the high steering effort. ECU designed a "pancake" style gearbox instead of the normal output on one side, input on the other. This is possible due to not needing full rotations on the output.

Cheers,

Kev

[mdavis]

But the really big advantage is that the chassis floor can now be flat and obstruction free from driver's bum all the way to Front-Bulkhead. No need for the common "stepped-floor" that is necessary for the R&P to be at right height wrt wheels. Lowering the floor also lowers the driver's quite heavy lower legs, and lowers the whole of the footbox structure, including quite heavy pedal-assembly, FB, and IA. (BTW, I believe this "lowering of footbox" is the major rationale for ECU's system.)

So, more complexity by adding the two idlers, but overall a simpler chassis to design and build, and a significantly lower CG, and possibly less total mass.
~~~o0o~~~

Feasibility of PURE PITMAN-ARM Steering.
=================================
The main disadvantages I see with pure PA (as also noted by Mdavis above) are;

* Because of the various practicalities of the full linkage (ie. Ackermann, bump-steer, etc.), only about +/- 60 degrees of steering-HW-Angle is possible (ie. HW turns 60 degrees from centre to full-lock-one-side). Better would be HWA = +/- 45 degrees only. But this could mean too high driver arm-loads, and/or too fast car response to small hand movements.

* For typical FSAE location of driver wrt front-axle-line, the plane of the HW might be too horizontal, or "bus-driver-ish" (ie. near vertical steering-shaft).

But I reckon all these problems can be overcome. Taking them in reverse order;

* A car that has the driver entirely within the wheelbase (ie. feet on or behind the front-axle-line) can have the PA in front of driver's feet and at a suitable height that gives the right ergonomic angle for the HW. And "driver-inside-wheelbase" is generally a good thing for FSAE dynamics, because it gives more R%, lower Yaw-Inertia, simpler, lighter chassis, etc.

* The "too fast car response" is not a problem at all, IMO. In fact, it is what you should be aiming for!

Total car response time = time for driver to initiate the action + time for car's tyres, etc., to respond and make the car move. Reduce the driver-action-time to zero and you still have to wait for the tyres and the rest of the car to do their thing. That can still take a long time for a car with soft tyres, large Yaw-Inertia, etc.

Consider also that modern fighter jets are controlled by a rigid side-stick that barely moves at all. It is strain-gauged and responds to the pilot's hand forces, rather than responding to large motions. But to give the planes really fast response times the designers had to make them dynamically unstable as well (ie. negative "static margin"). Just reducing pilot-action-time to zero is not enough.

* So the problem boils down to finding a way to lower the HW forces required for such quick steering of, say, +/- 45 degrees of HWA. Simply moving the driver, and hence also the car's CG, rearwards, is a big step in that direction. Making the whole car lighter also helps. And centre-plane steering geometry helps a lot, with just enough Trail for the right feel (many FSAE cars already have this, so it is certainly "feasible").

Worst case, you remove all the hydraulics, reduce the Trail to ~5-10 mm, and slap the Enduro driver on the back and say, "You can do it!!!"

Z

On corner entry, the UC 2015 car actually responded very well, almost to the point where I didn't like it. I think that was more of a rear end suspension geometry issue than anything else (RC too high), where I could feel the rear end unloading. More static rear weight would have helped this, and the team missed their weight distribution goal by a decent amount. That car also had very high yaw inertia (I'd have to ask the team what the value was with 160lb driver, then add some for 210lb driver. The front wing (specifically end plates) were very heavy and this caused issues. The rear wing endplates had similar issue, which raised CG, exaggerating the rear weight transfer issue on corner entry. All things that are relatively easy to fix with an iteration of that car. Sadly, the team is going back to a RNP for 2016, so we'll never know if the system could be implemented in an FSAE environment. The pitman arm manufacturing also left something to be desired, but could be easily improved.

I have thought numerous times about building a larger version of an FSAE car for SCCA's A-Modified Solo II class, and the idea of a completely flat bottom of the chassis has some huge advantages for frame construction. You can also run the LCA's of an SLA suspension very far inboard (under the driver's legs) as UC has done in the past (2011) to the point where none of the extra parts would be necessary for reasonable bump steer.

mdavis - very interesting to hear of your experience with pitman arm to column in fsae. Perhaps in your case the driver would have actually been fine with the response if the effort wasn't so great after all. What kind of steering axis geometry did your car have, out of interest?

I think with a few geometry changes in the rear end (or possibly just simple setup changes), the car could be very good. I drove it after I had graduated, and was more on a "consultant" level than anything else. In exchange, I provided a fairly thorough write-up to the team about the whole car, packaging, and behavior from the driver's seat. The response was quite good on turn in, though steady state was pretty terrible. Some changes were made after I drove the car that seemed to trend in the right direction, but tire wear was far worse, especially on the front end. I don't think this was something that couldn't be compensated for after some more setup changes, but I was just suggesting gross changes to make big swings at the balance. If I had to guess, I'd say 2-3 seconds worth of time could be gained on an autocross course with that car from just setup changes.

As for the steering geometry, I do not know the specifics, but I think there was close to 3/4" of scrub radius on that car (7" wide wheels, with a 4" inner half is somewhat limiting here, which is why we ran 6" wheels in 2013), and if I had to ballpark caster and KPI, I'd estimate both were designed to be ~5 degrees each (maybe 1-2 degrees either way for those quantities) and I'm not sure what the mechanical trail was. Regardless of what the numbers were, I think re-designing the outboard suspension could allow for adequate steering weight to not need a bevel box or anything with a pitman arm steering.

If anyone questions whether or not drivers can achieve max lat with very minimal steering inputs or changes, you need to go watch a dirt oval kart race. Watch the hands of the really fast karts during qualifying. If I had to guess, they're moving less than 5-10 degrees HWA from straightaway to corner. This is a racing event where 20/100+ karts make the feature, and it could be a matter of .0015 seconds covering the entire 20 kart feature.

Kevin, thanks for that CAD. I hadn't seen the rocker mechanism in the pictures on the Facebook page last year. Very neat and tidy system, and that brace could provide a nice place to mount pedals.

[Kevin Hayward]

mdavis,

The original system did have the rockers mounted of the pedal box mounts. However it was not stiff enough and the brace you see in the CAD was stand-alone. It could have been made to work, but the team didn't have the time to redo the pedal mounts at that point. The team also started with the gearbox near the driver, however the steering shaft twist was far too high.

The new system eliminates nearly all the mounts and is much better integrated with the existing chassis structure.

Kev

[Z]

Andrew,

Why not electric power steering?
... Less parts, less weight, and less interference with the engine.

I would argue that compared with Hydraulic-Power-Steer, EPS is more complicated, heavier, and messes more with the engine. The main difference between the two is the "source" of the power.

An EPS system "piggy-backs" on the car's existing electrical power source/store, whereas HPS typically requires the addition of an extra "hydraulic-pump + oil-tank" (ie. analogous to "alternator/E-pump + battery" in EPS). Well, unless the car is an older Citroen, which comes standard with the hydraulic-pump + regulator + accumulator to power its self-levelling-suspension, brakes, clutch, gearbox, etc. So on Citroens, HPS is a miniscule addition of hardware (= 1 x very small valve on pinion-shaft + some o-rings around the rack...)

Anyway, add an EPS to a typical FSAE car and you will very likely have to fit a bigger alternator. In recent years Monash have run an extra belt-driven "E-pump", because their standard bike-engine's alternator was not enough for FSAE E-loads. When you add an often demanded extra few hundred watts of EPS to all the other E-stuff on a typical FSAE car, then I reckon you will need a BIG back-up alternator! Any power-steering system, E or H, will have to work quite hard, and almost continuously, in FSAE/Autocross conditions.

But I agree that EPS from a Quad-bike would be very EASY to fit to an FSAE car. And Quad-bike steering is the archetypal Pitman-Arm steering. It could be almost "bolt-on and drive". Just make sure your engine's E-pump is big enough...
~~~o0o~~~

CWA,

... I believe the steady-state Ay/Steer gain can often still be too much.
... Do you not think this could be the case with a lack of steer ratio?

No. Consider a car that is set-up correctly as a whole (see below), and has a "rigidly mounted steering-HW" fitted with strain-gauges that control lightning fast actuators that steer the road-wheels, such that the front-road-wheel-angles change (almost) "instantaneously and effortlessly" in direct proportion to the driver's hand forces (and, yes, I know all this is probably illegal in FSAE, but...).

I reckon good drivers would find this car great fun to drive, but, after a while, they would wish that the whole car would react MORE QUICKLY to their steering demands! Furthermore, if this type of PS was fitted to a variety of FSAE cars with different overall-masses, tyre-size/stiffnesses, yaw-inertias, etc., then the drivers would very easily differentiate the soft-tyre/high-inertia cars from the faster reacting cars, and would much prefer the faster ones.

The key here is to get the "whole car set-up" right. Briefly, the car should be stable with "hands off the wheel" (in simple terms this means toe-in front and rear, stiffer cornering-compliance rear-tyres, positive static-margin (*), etc.). Then, as soon as the driver signals "go left", the car becomes "unstable" to the left (front-wheels turn left and develop big toe-out, etc.). When driver lets go of HW the car proceeds stably in its current direction (well, after a minimum amount of yaw overshoot).

(* I haven't checked the sign convention for SM. Here and in previous post I assume "positive = good/stable", "negative = bad/unstable", ... like a bank balance. Corrections welcome. )
~~~o0o~~~

FWIW, the Ford Model-T (a reasonably successful design!) has a planetary-reduction-box just under the steering-HW, and a Pitman-Arm at the end of the output-shaft from that planetary-box. The planetary-reduction was ~5:1, so the T's 1.25 turns lock-to-lock of HW gave +/- 45 degrees of PA motion.

Z

[CWA]

Right then. Let's not worry about steer ratio chaps. Get some wheels with a nice big positive offset, chuck in a vertical steering axis, no worries over steering effort, slap on your pitman arm with no gearbox, job's a good'un.