Suspension Design

Quote:

---

Originally posted by AxelRipper:
http://img.gawkerassets.com/img/187x...g/original.jpg
... Not sure how exactly you'd package your steering box in the center of the chassis, especially with an engine there.

---

(Note, front-of-vehicle = bottom-of-pic, and driver's-side = right-of-pic.)

Axel and Tony,

Once again...

Almost all US wishbone cars of the same era as the F100 had a "recirculating-ball" steering box located in front of the driver, as does the F100 (at right of pic). At the bottom of the steering box is a rotating "Pitman Arm". On the wishbone cars this PA connects to a "Lateral-Link" that passes under the engine, and in turn connects to a rotating "Idler Arm". The PA and IA have the same geometry, so when switching from LHD to RHD the manufacturer only has to swap steering-box-PA and IA (well, sort of...).

When steering, the LL moves laterally back and forth under the engine. At two positions on this LL are bolted the inner BJs of the two steering tie-rods. Now picture this common steering arrangement sitting slightly behind the I-beams of the F100 (ie. slightly above I-beams, in above pic). Putting the tie-rod inner BJs near the centre of the LL puts them close to the two suspension's pivot axes, giving small bump-steer, or maybe none at all.

Note that when the tie-rod inner BJ is EXACTLY ON the pivot axis, then there is NO bump-steer (well, except for some other effects I'll ignore for now). When the BJ is OFF the axis, but the line through inner and outer tie-rod BJs points directly at the axis, then there is no bump-steer at the instant, although some small bump-steer develops towards full bump and droop (because the tie-rod line no longer intersects the pivot axis).

Anyway, if they used the standard steering linkage of the time, then the F100 would have had minimal bump-steer, with a simpler all-round suspension than double-wishbones. Even if the tie-rod inner BJs were about a foot away from the swing-arm axes (ie. closer to the I-beams and wheels), but pointing at the axes at ride height, then there would only be small bump-toe-out at full bump/droop, which is not too bad (it gives small roll-understeer)

Instead, the F100 has one long tie-rod from the steering-box-PA out to the opposite-side wheel. Some distance along this "major" tie-rod is a BJ attached to the "minor" tie-rod that connects back to the driver's-side wheel. I'll let you figure it out, but bottom line is that bump steer of BOTH wheels is (mostly) towards the driver's side. Worst is when one wheel bounces up and the other droops down.

Z

That is all true Z.

Except the swing axis for each swing arm is located on the opposite side of the vehicle.
The further forward the tie rods are located, the closer to the opposite side they will need to reach.

That still means the tie rods need to cross over.

There quite a few different conversations going on at once here, but I feel the one that should be paid most attention to is the original one about suspension design and not active toe.
I'd like to request that this topic be moved to it's own thread to avoid further clutter.


Z,

I'm still very curious on your input on swing arm suspension.

Tony,

No! You're overcomplicating it. KISS! http://fsae.com/groupee_common/emoticons/icon_smile.gif
~~~~~o0o~~~~~

MCoach,

Sorry for delay. 'Tis the season for family to come a'knockin.

Will post by Friday night...

Z

Z: some pics I took at FSAE-A 2012.

http://www.facebook.com/media/...3272531135703&type=3

Rex,

Thanks for the pics! Lots of interesting stuff there (on UWA's car).

Bevel-gear and Pitman arm (?) steering. Face-spline centrelock wheel attachment. And, of course, the suspension with roll-mode controlled by a single lateral U-bar...

Z

Quote:

---

... the major suspension types are,

1. Beam-axles (eg. on all vehicles ever, can be live or De-Dion at rear),
2. Sliding-pillar (eg. Morgan and Lancia, and most motorbikes and aeroplanes at front),
3. Lateral swing-axles (eg. Tatra (still on their trucks), can be high or low-pivot),
4. Leading and trailing-arms (eg. Citroen 2CV front and rear, and most motorbikes at rear),
5. Semi-leading/trailing-arms (eg. F100 at front, and many mid to late 1900s cars at rear),
6. Strut-wishbone (eg. McPherson at front, strut at rear),
7. Double-wishbone (eg. originally mostly at front to allow shorter wheelbase with front engine),
8. 5-link (eg. recent fashion, mainly for NVH reasons).

Of these, any of the first five are more than adequate for a winning FSAE car.

---

Apologies for this very long post, but I will cover all three of the above bold suspension types under the same general heading.

The SWING-ARM Suspensions.
===========================
These all have a single pivot joint at the chassis (ie. kinematically a single "revolute joint", but physically it may consist of two bushes or BJs). From this joint a single structural "arm" extends out to the wheel. Front wheels have a steer-axis, or "kingpin", at the wheel end of the arm, together with a separate means of controlling this steering DoF (typically via a tie-rod). Rear wheels can have the axle housing fixed rigidly to the end of the arm.

The main difference between the three types is the orientation of their pivot axis wrt the body.

"Pure Lateral Swing Arm (or Axle)" has its pivot axis close to longitudinal wrt the body (ie. parallel to X axis). Historically, Tatra was the first to use these Lateral SAs in the early 1920s (first at rear with their "jointless" diff, then a few years later both front and rear). They still use them on their off-road trucks.

"Pure Leading or Trailing Arm" has its pivot axis close to lateral (ie. parallel to Y axis). The Citroen 2CV, designed in the late 1930s, but only sold post WWII (late 1940s), was the first, and possibly only car to use these at both ends. Pure Trailing Arms at the rear became very common post WWII, mainly on French and some English cars. PTAs are almost universal on motorbikes today.

"Semi-Leading or Trailing-Arm" has its pivot axis somewhere inbetween lateral and longitudinal. The Ford F100 Twin-I-Beams in earlier posts are Semi-Leading-Arms. Semi-Trailing-Arms at the rear were extremely common from about 1960s to 1990s, mainly on German designed cars (VW/Porsche, Audi, BMW, Mercedes, NSU, German Fords and GMs, etc...). Funny how cultural fashions can so influence engineering decisions, eh? http://fsae.com/groupee_common/emoticons/icon_smile.gif
~~~~~o0o~~~~~

Here are the major pros and cons, IMO.

ADVANTAGES.
=============
1. Significantly simpler than Double-Wishbones or 5-Links.
The Swing-Arm typically only requires two hard points on the chassis (ie. for the 2 bushes/BJs forming the arm's pivot-axis), plus one more for a direct-acting Spring-Damper. The left and right suspensions can share at least one of these hard points, or even all three (giving only 3 x HPs at each end of car, though benefits here depend on other details).

IMO, one of the biggest costs of DWs is the multiple chassis mounts required for them. For example, each upper wishbone's two chassis mounts are relatively lightly loaded, but on FSAE cars they often have 6, or even 8 tubes converging on each node. This multiplied by the four corners = a lot of work. Furthermore, these nodes usually don't match neatly with the various Rules mandated roll-hoops and side structures. The end result, IMO, is a horrible dog's breakfast of tubes running everywhere. In fact, I can't recall seeing a single FSAE spaceframe that looks at all efficient, let alone elegant.

2. Significantly stiffer than DWs or 5-Ls.
At the rear, the wheel's "upright", namely the axle's bearing housing, can be one continous piece with the Swing-Arm. The ball-joints between DWs and their upright inevitably result in tortuous, high stress load paths through the bolts, brackets, balls, etc. This inevitably results in increased compliance for a given amount of material. And then there is the issue of the BJs developing freeplay! http://fsae.com/groupee_common/emoticons/icon_frown.gif A direct, continuous load path from Swing-Arm to bearing housing is always going to be lighter and stiffer. And also cheaper and quicker to build.

However, at the rear it can be beneficial to provide some camber and toe adjustment (at least during testing). This can be done with different SAs (swap the whole assembly), adjustments at the chassis mounts (up-down, in-out, etc.), or an adjustable joint in the SA near the wheel. This last type of joint requires no movement during suspension travel, so, IMO, it can be made much stiffer for given mass than a typical DW-to-upright joint.

At the front the Swing-Arm's steer-axis can be made with two ball-joints, like a DW. But it can also be made as a simple revolute joint, which can provide significant advantages. Specifically, low friction rolling element bearings can be used, giving lighter steering (important when the aero downforce gets interesting... http://fsae.com/groupee_common/emoticons/icon_smile.gif). Might post a sketch of such a steer-axis one day...

3. Lower CG than DWs or 5-Ls.
Most, or all, of the Swing-Arm's structure can be well below axle height. This can also be done with DWs, but the above complicance issues likely then give too much camber/castor change under load. You can try estimating this compliance with FEA, or better yet, just build it and physically test.

4. Easier to find the "Instantaneous Screw Axis", which is well-behaved.
As discussed at length elsewhere, any suspension's wheel assembly "screws" about an ISA, wrt the car body. In the non-bold suspensions above the ISA usually moves about a lot as the suspension moves through its travel. Knowing the position of the ISA is very useful to the suspension designer, although I accept that this is rarely done.

With Swing-Arms the ISA is easily found. It is the physical pivot-axis! An exception is when the steering linkage gives significant bump-steer, which then requires a knowledge of the "cylindroid" (maybe later...). Nevertheless, even with bump-steer the Swing-Arm ISA is usually very close to the physical pivot axis.

Once the ISA is found, then so also are all of its n-lines (or vice versa). Because the Swing-Arm's ISA and n-lines are so well behaved (ie. relatively immobile), the suspension's behaviour is also very predictable. More below under "Kinematics".
~o0o~

DISADVANTAGES.
===============
1. Packaging.
It is often beneficial for the suspension's ISA to be a long way away from the wheel. For a given amount of suspension travel a distant ISA gives smaller rotations of the wheel-assembly (in camber, castor, or steer) than a nearby ISA. DWs and 5-Ls can give distant ISAs with a compact linkage, which is their main advantage.

However, the open-wheel style of FSAE cars allows relatively long physical Swing-Arms to be easily packaged (easier than under passenger cars). Also the smooth tracks and very short mandated suspension travel of +/- 2.5 cm means that rotations due to nearby ISAs are not too big a problem (eg. negligible gyroscopic effects from too much camber change over bunmps).

2. Highly unusual in FSAE.
Not much to copy, so you have to work it out for yourselves. Worse yet, you can expect lots of irrational bias and prejudice against your ideas from FSAE peers, and, yes, officials too. See last section below.
~~~~~o0o~~~~~

KINEMATICS.
=============
Some brief points here.

If a suspension's ISA has any vertical component (ie. it is NOT horizontal), then any motion about the ISA necessarily results in some steer motion (ie. rotation about the vertical component of the ISA). The amount of this bump-steer depends on the distance of the ISA from the wheel, its steepness, and the travel of the suspension (eg. size of bump, or amount of body-heave/pitch/roll).

With Swing-Arms on FSAE cars you would really have to cock things up big time for this to be a problem. Nevertheless, let's assume the ISA (= pivot axis) is horizontal wrt body.

Pure Lateral Swing Arm.
====================
Historically, to allow easier drive from the diff, the ISA was at axle height. This gives large jacking forces. So only use if some sort of anti-axle-bounce springing is provided (eg. "third spring", lateral-Z-bar, or see Formula Vees). Better to have a low-pivot swing-arm, pivoting from floor level. This gives a small (acceptable) slope to the lateral n-lines (ie. lowish RC).

ISAs close to the centreline of the car give ~100% camber compensation (ie. short Front-View-Virtual-Swing-Arm). So you can run very soft springs and the body-roll doesn't change wheel camber, so good grip is maintained. However, heave and pitch motion will change camber. Again, this best fixed with anti-axle-bounce springing (as above).

Side-View-Virtual-Swing-Arm is infinitely long, giving zero castor change wrt body. Horizontal ISA means horizontal longitudinal (side-view) n-lines, so 0% anti-dive/squat, which is actually quite good.

Pure Leading or Trailing Arms.
=========================
Front-View VSA is now infinitely long, so 0% camber compensation and any body-roll adversely affects wheel camber (but no gyro forces over bumps, so good off-road). Body-roll best fixed with side-pair longitudinal Z-bars, NOT end-pair lateral U-bars like normal ARBs (= bad). Lateral n-lines are always horizontal wrt body, so very predictable handling, even though RCs shoot of to infinity with slightest body-roll!

Side-View VSA now very short, so lots of castor change, but this has surprisingly small adverse effects. Longitudinal n-lines dependent on whether drive or braking is inboard or outboard. Generally, with outboard brakes (eg. at front), the ISA should be at floor level. With inboard drive/braking (eg. at rear), the ISA should be at axle height, or slightly higher. This keeps all side-view n-lines close to horizontal, or sloping slightly up-to-centre-of-car for some front anti-dive and rear anti-squat.

Semi-Leading or Trailing Arms.
=========================
These are somewhere between the above two types. They allow the option of varying front and side-view VSA lengths, although not so that both are very long at the same time.

If the wheel is at a non-zero camber angle, then this suspension will give toe-change (= bump-steer) even if the axis is horizontal (it is a castor/camber coupling effect). This can be a problem for long travel suspensions (such as passenger cars), but should be no problem at all in FSAE.

And a lot of other stuff... But moving on...
~~~~~o0o~~~~~

HOW TO DO IT.
===============
Although Semi-Leading/Trailing-Arms have possibly the worst (most compromised) kinematics of the various suspension types, they probably fit in best with the rest of an FSAE car, so are the simplest overall solution. So these are my "second favourite" solution after Beam-Axles. They are structurally the simplest, so potentially the cheapest, quickest to build, and most rugged. And the dodgy kinematics don't really matter because of the very short suspension travel.

So I would have the arms attaching to the chassis close to floor level (because already strong structure down there). Each arm would have one BJ on the centreline of the car just under its axle line, giving 100% camber compensation. The front arms would have their second BJs attach to the sides of the chassis close to the front roll hoop. Similarly, the rear arms would attach either side of the seat-back/main-roll-hoop. The kinematics would thus be closer to that of the Lateral-Swing-Arms (ie. pivot axis close to longitudinal).

I would fabricate the Swing-Arms out of sheet-steel (1 to 2 mm thick) as a hollow structure of roughly overall "V" or "Y" shape (picture wheel at bottom of V/Y, chassis at top). The hollow shell would be roughly 80 mm diameter at the wheel, tapering to maybe 20-30 mm at each chassis BJ. See Twin-I-beams in earlier posts as rough idea, but hollow, and attached to centreline of chassis rather than "crossed".

Note that for a given wall thickness and length, the five, say, 16 mm diameter tubes of a DW or 5-Link weigh exactly the same as a single 80 mm diameter tube. And this larger tube doesn't buckle whenever it hits a squishy little rubber cone. A shell-like structure would also suit carbonfibre, but not my choice. I also think all four arms could be identical, with an adjustable, three-bolt connection to the front-kingpin, or rear-bearing-housing (might post pic later...).
~o0o~

Or the Swing-Arm could be an evolutionary simplification of a DW. Just weld the inner ends of the upper-wishbone to the tops of the lower-wishbone. Toss the upper-wishbone-inner-BJs, and all that messy chassis structure they connected to. Also toss all the outer BJs at the rear of the car (the front uprights can stay same). The kinematics are the same as with any short FVVSA double-wishbone suspension.

Would that work?

Well, the HyperProRacer has just this sort of suspension at the front, and is at least as fast as the faster FSAE cars (search their website). And similar Swing-Arms have already been tried in FSAE, as seen on the "Lancaster Link" thread, although I'm not sure of their performance.

Most interesting on the Lancaster thread is the almost universal dismissal of the idea, including by officials who should know better! (The only positive comments come from Pete Marsh, who makes similar suggestions to above.) Much of the negativity is spurious, such as the many comments regarding RCE magazine's claim of "the greatest innovation in 40 years".

Unfortunately, even more of the criticisms are completely wrong (frankly, plain stupid!), such as the many claims that the pushrod will be under very high load. Honestly, the simplest of FBDs, or even just a quick inspection, shows that the pushrods are under very similar loads to typical DW suspensions.

When I see that sort of fallacious codswallop being peddled as the main argument against something new, then I really start to worry about YOUR futures!
~~~~~o0o~~~~~

Much more to say, but maybe later...

As always, comments and (FBD'd!) criticisms welcome... http://fsae.com/groupee_common/emoticons/icon_smile.gif

Z

Z,

Any chance we could get some of your excellent drawings of some of these designs?
I don't know how long they take you, but they are always impressive http://fsae.com/groupee_common/emoticons/icon_smile.gif

-William

Quote:

---

Originally posted by Z:
Ian, MCoach,

Here is an earlier thread about 4 Wheel Steer.

Thinking about it a bit more now, I still reckon that with the current Rules the biggest gains are available from better aero. But if the Rules were changed to greatly lessen aero, then I would probably go for greater "agility" from something like 4WS, before chasing ultimate horsepower, 4WD, etc. The mechanical cost/complexity of 4WS is quite small. It can be done with any of the simple suspension types, probably using very similar suspension components at each corner.

Ergonomically, the easiest for the driver might be two big thumb-actuated buttons on the steering wheel. When neither button is pushed the steering is normal "front-only". Pushing and holding the green button gives stable "same-steer" at the rear wheels, for low-yaw side-stepping through the slaloms. Very briefly pushing and holding the red (danger!) button gives small rear counter-steer, to get around the super-sharp hairpin.

The only way to know for sure is to build it, and then extensively test it. This would make a great "parallel design project" (using last year's car) for a well resourced team, and should be a lot of fun! http://fsae.com/groupee_common/emoticons/icon_smile.gif

Z

---

Instead of buttons, how about a steering wheel that is free to translate in the y direction. push the wheel to left points the rear wheels to the left, pushing it to the right points the rear wheels to the right. This would actually be a pretty simple linkage and gives the driver full independent control of the front and rear wheels for any combination and any type of 4WS or 2 wheel steering

The one thing that still has me hesitant about swing arm suspensions is the high roll center that comes with it and the jacking forces that follow that. Because of the high roll center, and the Corvair comes to mind on, the suspension may have a tendency to tuck under due to the very short swing arm length.

I may be missing a key part here...

Mounting them low as possible, and with the mentioned anti-bounce, then it might be effective.