+ Reply to Thread
Results 1 to 10 of 303

Thread: Suspension Design

Hybrid View

  1. #1
    Hey Z,

    To answer your questions, plus a few extra thoughts. I will try to be brief but it doesn't always work out as I like to be sure I am understood!

    Q1. There are tripod style cv joints at both ends of the drive shaft to (obviously) transmit drive and accommodate the "extreme" drive shaft plunge. The CV at outer end was moved out as far as possible and became housed in the upright. This was to make the shafts as long as possible in order to reduce drive shaft angles.

    Q2. That is tricky one... The car that was pictured on page 9 was the car the team and I built for FS 2009. That was four years ago and my final year at uni. Since then whilst occasionally thinking about such things I have never really sat down to seriously give it thought. Real life and other such things tend to get in the way! But a few thoughts I have had...

    1. I have pondered the idea of how to mount the spring/damper units low down, preferably on the floor. But the obvious problem being that the pivot is effectively at floor level and that leads you to using an upwardly angled push rod.

    The key benefit to this in my eyes (other than the obvious CoG etc.) would be that you could really easily build what would basically be a ladder frame chassis that all the suspension would work off. This would be easy to make with excellent precision and you would have a rolling chassis in no time. The rest of the structure could then be made as light and as minimal as possible just to comply to the regs.

    2. Mono-shocks and air springs. Kind of conflicting again. I have long pondered air springs as a cheaper and lighter option. Also, we used cane creek shocks/springs and there was basically too much adjustment. Most teams I think struggle for testing so a more basic shock system set up well I think would be better than a fancy system set up badly.

    A mono-shock system would also reduce cost. As you are basically de-coupling roll/bump with the LL I think a mono-shock system would work well, although I admit I haven't actually given this more than 2 minutes thought at the moment. If you ran a mono though I doubt an air shock would be beefy enough to cope with the weight of two axles.

    3. The whole thing just needs building with a bit more finesse and precision than we managed! But we did the best with the limited time/resources we had.

    4. More recent Lancaster teams split the top and bottom wishbones. They still pivot along the same axis but they are separate arms. Our systems the top and bottom arms were welded together at the pivot point. I guess this just makes fabrication easier for them but I'm not sure.


  2. #2
    Senior Member
    Join Date
    Mar 2005
    The subject of interconnected springing is currently being discussed on this "Roll Rates in RCVD" thread. I am adding these notes here, mainly to keep as much of my "Suspension Design" ramblings on the one thread.

    The simplest way to "spring" a car is to put one spring in control of each wheel, namely a "spring-at-each-corner" suspension. At the other extreme a single spring can be connected to, and be in control of, ALL the wheels via some sort of linkage. Search with keywords "Balanced Suspension", "Kinetic suspension" +++.

    For now, let's consider the simplest types of interconnection, where a single spring interconnects the chassis and TWO wheels only. The only clear explanation of this particular subject that I have ever seen was an article by Mark Ortiz in RaceCar Engineering magazine in the late 1990s (maybe '97?). There may be other sources for this information, but it is definitely NOT mainstream VD.

    If you students apply enough pressure now, then maybe one day Claude will start teaching it. Claude???

    DEFINITIONS - These covered in more depth elsewhere, but a brief recap here:

    * U-BARS and Z-BARS - These are descriptive names given to torsion-bar versions of interconnecting springs. The names refer to their appearance when seen in plan-view and fitted to a car in the "usual" manner. The difference between these two types of spring is that the lever-arms on U-bars both point in the same direction, while the lever-arms on Z-bars point in opposite directions.

    The functional behaviour of these springs can be implemented in many different ways (eg. with leaf, coil, or other spring types, and via mechanical, hydraulic, or other linkages...). The important functional differences are;

    * U-BARS - RESIST DIFFERENT movements of their ends/wheels (ie. when one wheel moves up and the other wheel down, or vice versa), but ALLOW SAME movements of their controlled wheels (ie. both wheels move up, or both down) with no resistance, other than maybe some friction.

    * Z-BARS - RESIST SAME movements (= both up, or both down), but ALLOW DIFFERENT movements (= one up and the other down).

    So functionally U and Z-bars are complementary to each other.

    * ALL-WHEEL-MODES (of a four-wheeled, rectangular pattern, vehicle) - These are different ways of describing the motion of all the wheels as they move wrt the car-body. There are many different ways to define these (in fact, four infinities!), so the following is only a taste.

    HEAVE-MODE (aka Bounce-mode) - All four wheels move in same direction (either up, or down) by the same amount. So +1 cm of Heave-mode might have all wheels moving up, wrt car-body, by 1 cm.

    PITCH-MODE - Both front wheels move up by the same amount, and both rear wheels move down by the same amount. A different, and equally valid, definition for a car with 40F:60R weight split might have "+1 cm Pitch" as front wheels up by 1.5 cm, and rears down by 1 cm. This corresponds to the car pitching about its CG.

    ROLL-MODE - Both right-side wheels move up, and both left-side wheels move down, by the same amount.

    TWIST-MODE (aka Warp-mode) - One diagonal-pair of wheels move up, and the other diagonal-pair move down, by the same amount. Again, a different definition might have the front wheels moving 50% more than the rears, to better describe a Twist motion about a 40F:60R car's CG.

    Note that early work on these concepts (1930s to "active suspension" era in F1) usually had the definitions rather rigidly defined to be "all equal" movements of the wheels, up or down. This can then introduce calculational difficulties when the CG is not at 50% wheelbase (ie. the researchers talk of "mode-coupling", and insist that the Warp-mode must exert a force to help balance the handling).

    IMO these difficulties are most easily overcome by REDEFINING the modes so that they better suit your particular problem (and you are free to define anything in any way you want!). This allows the Warp/Twist mode stiffness to always be zero.

    So far, so simple.

    Now what happens when we start fitting U and Z-bars to cars with four wheels? Here is a summary of all the possible interconnections when using either two U-bars, or two Z-bars, between the various possible pairs of wheels. We are looking for the effect these springs have on the various All-Wheel-Modes described above.

    1. Between End-Pairs (ie. one U-bar connects the front-pair of wheels, another U-bar connects the rear-pair).
    Stiffens Roll and Twist (Heave and Pitch free).

    2. Between Side-Pairs.
    Stiffens Pitch and Twist (Heave and Roll free).

    3. Between Diagonal-Pairs.
    Stiffens Pitch and Roll (Heave and Twist free).

    1. Between End-Pairs.
    Stiffens Heave and Pitch (Roll and Twist free).

    2. Between Side-Pairs.
    Stiffens Heave and Roll (Pitch and Twist free).

    3. Between Diagonal-Pairs.
    Stiffens Heave and Twist (Pitch and Roll free).

    Again, quite simple. These symmetrical two-wheel interconnections always stiffen up two of the all-wheel-modes, and leave the other two free. But what does all this mean in terms of overall car behaviour?

    CONCLUSIONS (briefly).
    * Importantly, the above description is only for "linear" behaviour. That is, when the lengths of the lever-arms at the end of a given (U or Z) torsion-bar always stay in the SAME RATIO throughout the range of wheel travel. Typically (by design, or by accident) the effective lever-arm lengths will change by different amounts, so giving a rising-rate, or falling-rate, behaviour at each end, and a different force ratio between the ends. This non-linear behaviour can then "add stiffness" to the above "free" modes. This can make a real mess of your best laid plans, or it can be used to considerable advantage (see below).

    * In general, anything that stiffens the Twist-mode is, at best, unnecessary, and at worst, VERY BAD (I will explain why in a later post). There are exceptions, but if your suspension layout starts with a zero-rate Twist-mode, then it is usually VERY EASY TO ADD more Twist stiffness. But if your suspension starts with a lot of stiffness in its Twist-mode, then it is all but IMPOSSIBLE TO SUBTRACT that stiffness (ie. it requires a total redesign).

    * So of the U-bars, we can cross out the end-pair (#1) and side-pair (#2), because they add Twist stiffness. Note that end-pair U-bars are the all too common ARBs (which are considered mandatory in FSAE, by some Design Judges!). Likewise, we can cross out the diagonal-pair Z-bars (#3).

    * Diagonal-pair U-bars (#3) are potentially useful, but have disadvantages. The diagonal interconnection can be difficult to package with a mechanical linkage. Also this arrangement does not control Heave, so some other springing must be used to hold the car up (quite important!). Perhaps worst, the Pitch and Roll stiffening are inextricably linked. So, if you want a stiff Roll-mode, say for flat cornering, then you MUST also have a stiff Pitch-mode. This might be acceptable for a racecar, but a soft(ish) Pitch-mode contributes greatly to ride comfort of passenger cars. The Spanish "Crueat" (spelling?, and maybe Portugese?) hydraulically interconnected suspension uses a variation of this type.

    * This leaves end-pair and side-pair Z-bars. Both these control Heave, which is very useful as it is the mode that you MUST have (lest the car drag its belly along the ground).

    * End-pair (or lateral) Z-bars are common these days in motorsport, and are usually called "third-springs" (more accurate would be "seventh and eighth-springs", given that they came after the 4 x corner-springs + 2 x ARBs). IMO these have evolved almost entirely by random trail-and-error selection, with next to no deep theoretical understanding. Nevertheless, they work well for big-aero cars because they can greatly (and non-linearly) stiffen Heave and Pitch, and thus provide a stable aero platform while leaving Roll and Twist UNAFFECTED. UWA 2013 car's "W-springs" are end-pair Z-bars.

    * Finally, side-pair Z-bars are perhaps the best of all the above options (though the rarest!). Heave and Roll generally have to be the stiffest of the four modes (passenger or racecar), and they carry the largest, and similar, loads (ie. statically each bar carries half the car's H weight, and at high-G cornering the oustide bar carries close to the total car weight = H/2+RollLLT). Also, for given Gs, Pitch-longitudinal-load-transfer during acceleration or braking is less than Roll-lateral-load-transfer during cornering by the ratio of Track/Wheelbase.

    Bottom line, side-pair-(longitudinal)-Z-bars have a lot going for them, and are very easy to implement (see next post). And by arranging the linkages to be rising-rate at each end of the bar, they can also control Pitch. (I call this "pendulum springing" and it comes almost for free, which may be why BL-Austin-Morris used it in the 1960s+). However, this rising-rate method has limitations, and smaller, lighter, dedicated end-pair Z-bars (one or two) can be used for more precise control of Pitch, with no stiffening of the Twist mode.

    More coming...

    Last edited by Z; 03-10-2014 at 06:24 AM.

  3. #3
    Senior Member
    Join Date
    Mar 2005
    (by F. R. McFarland, presented June 15 1955, and in SAE Transactions Vol 64, p284 1956, 560026.)

    I made some comments about this paper on the "Roll rates..." thread (linked on previous post, page 25). Below is Figure 1 from the paper, and my interpretation of the system.

    In the image the "Main Load Torsion Bars" are longitudinal-(side-pair)-Z-bars. The "Z" shape is quite apparent. IMO, all things considered, this particular layout is possibly the best way to do longitudinal-Z-bar suspension, both on production cars and many types of racecars. So also quite suitable for FSAE.

    The entire torsion-bar+lever-arms lies in a horizontal plane at the bottom of the car, so is easy to package and gives a low CG (ie. all the "spring mass" is at the lowest possible position, and big springs, ie. for off-road or luxury cars, can be quite heavy). The front lever-arm cranks outward, allowing easy connection to the wheel, while giving room for the front-wheel to steer. This lever-arm can be in unit with the lower wishbone, or else it can be connected to the suspension by a flexible link (Packard tried both).

    The rear lever-arm must now crank inward (to form the "Z"!). So the main body of the torsion-bar angles out towards the rear of the car. This allows the rear lever-arm to connect to any convenient point on the rear suspension. In this case the connection is by a short "pullrod" to the live-axle control-arm (= "Rear Axle Torque Arm"). Similar pullrod-like connection could be used on any independent suspension.

    Other comments:
    * The "Compensator Bars" at the rear act as simple corner springs (ie. no interconnections), but both can be simultaneously adjusted to reset rear-ride-height (hence "Levelizers"). This is important for a very softly-sprung luxury car, and is described as "an answer to the stylist's prayer"! IMO an adjustable lateral-Z-bar would be much better here.

    * The "Rear Stabilizer Link" is a Watts-linkage for lateral control of the live-axle. This was deliberately made quite soft laterally to reduce "harshness". IMO this is poor design for too many reasons to cover here. Well, just one being rear-axle-(over)steer during cornering! Many ways to fix this, but they didn't...

    * A "Front Stabilizer" (ie. lateral-U-bar) is also fitted. This, and the "Rear Axle Torque Arms", both act as ARBs, thus stiffening both the Roll AND Twist modes. IMO this shows a lack of deep understanding of the whole system...

    * IMO the Packard engineers didn't seem to grasp this whole concept nearly as well as the French, who were doing this sort of thing twenty years earlier. The whole paper seems to be focussed on the side-view, 2-D behaviour of the car in "bounce and pitch". Admittedly, they had better understanding of these motions than is currently the case with the modern concepts of "front and rear ride frequencies" (see extensive ranting elsewhere! ). But nowhere in the paper is any Twist or Warp-mode behaviour mentioned. Well, except, and only (!), the last Summary point "8. Reduced torsional stresses in frame.").

    * Note that to understand Twist-mode motions, you have to think in 3-D, with the four wheelprints starting in a horizontal plane, and their vertical motions taking them out of this plane. By comparison, the "bounce and pitch" motions discussed in the paper are contained entirely in a 2-D, side-view plane. This limited 2-D thinking seems to have prevented the Packard engineers from fully appreciating the advantages of a soft Twist-mode.

    * Finally, the paper starts by noting that "automobile developments seem to appear in cycles", such as manual to automatic gearboxes, straight-8 to V-8 engines, and so on. It ends with "The time is ripe for a cycle of development in automotive suspensions ... it would seem that within the next 2 to 5 years, we should see some radical changes in suspension design.". (My emphasis.)

    Well, that didn't happen! And a good indicator of why not is in the subsequent discussion to the paper. The other manufacturer's engineers quite clearly did NOT understand the Packard system (not even its dumbed-down, 2-D, side-view version), as they gave some completely false criticisms of it. Having led with this bulldust, they then launched into a marketing spiel about how great their suspensions were!

    Ahh, nothing changes...

    More in a few days...

    Last edited by Z; 03-20-2014 at 09:09 PM.

  4. #4
    Senior Member
    Join Date
    Mar 2005

    It has been suggested (on another thread) that while a soft Twist-mode might be advantageous on bumpy roads, it might offer NO such advantages on smooth, sealed roads, such as those typical in circuit racing.

    Those of you who prefer to "follow the numbers", rather than the unquantified opinions of experts, please read on...

    Milliken's RCVD, Chapter 18 "Wheel Loads" starts with,
    "The [vertical] loads at each wheel are extremely important in determining a car's maximum steady-state cornering capability." (my emphasis).

    The chapter goes on to give examples of how to calculate these vertical wheel loads. Quite reasonably, these calculations are simplified by assumptions such as,
    "...steady-state operating conditions - that is, smooth roadway, constant speed cornering, constant longitudinal acceleration, constant grade, etc.
    ... roll rates, spring rates ... are linear,
    ... chassis of the car ... is [torsionally] rigid."
    , and so on.
    In fact, there are about 5 pages discussing the importance of a torsionally stiff chassis, because,
    "... if the chassis torsional spring is weak, attempts to control the lateral load transfer distribution (and "balance" the car's handling by resisting more of the rolling moment on one track than the other) will be confusing at best and impossible at worst." (my emphasis again).

    Equations for calculating the variations to wheel loads from a large number of different factors are then given, including,
    * CG position,
    * lateral and longitudinal load transfer from horizontal Gs,
    * banking,
    * crests and dips in the road (albeit in a 2-D vertical-longitudinal plane only),
    * aero loads,
    * engine torque reaction (for front-engine -> live-rear-axle drivetrain).

    It is quite clear, however, from the seven-plus pages devoted to it, that the Millikens believe that Lateral Load Transfer Distribution is the most important factor to be considered when "adjusting handling balance" (which, in this particular area, I agree with). I repeat this for emphasis, if the wheel loads do NOT change as per your intended LLTD (or Claude's "Magic Number"), then the car will not handle the way you expect it to.

    At the end of the chapter is "18.11 Summary Example". This works through some of the above calculations for what might be a "sportscar", or perhaps a fairly softly sprung racecar (the corner-spring and ARB rates are a lot less than the tyre rates, so the car is not a very stiffly-sprung aero-car). Right at the very end of the chapter, on page 708 (my older version), is Table 18.1 summarising the changes in wheel loads due to the various factors. For this particular example the "Banking" effect is quite large (ie. oval track racing), the "Aero" effect quite small (ie. no big wings), and, quite clearly, the LLTD is by far the most important effect.

    Please go through the RCVD example in more detail, but for now take it that the car is slightly front heavy, but roughly with about 900 lbs weight per wheel. There is a Total LLT of about 800 lbs (from the two inside wheels, to the two outside wheels). This is distributed by the "springs, bars, and RC heights" as +/-430 lbs front, and +/-370 lbs rear, giving LLTD = 54%F, 46%R.

    Now the twist in the story. Nowhere in this 40 page chapter is any mention made of any TWIST in the road! All four wheelprints are ALWAYS considered to be lying in a PERFECTLY FLAT plane!

    Fortunately, there was a large blank space at the end of the chapter, so I added some more calculations. I imagined that the road is very "smooth", but it is also cambered in the usual manner so that the road surface has a cylindrical shape, which in "end-view" has a radius of about 40 metres. So, if the two edges of the road are 10 metres (30 ft) apart, then the centreline of the road is 0.3 metres (1 ft) higher than the edges (quite typical of real roads).

    Driving parallel to the centreline of this road introduces no Twist into the suspension, even if the road curves around a bend. But a car with ~3 metre wheelbase and ~1.5 metre track, driving diagonally across this road at an angle of about 15 degrees to the centreline, has about 7 mm (1/4") of Twist-mode between its four wheelprints (ie. one diagonal pair of wheelprints are up 7 mm, and the other diagonal pair down 7 mm, wrt car-body). Please do the calcs to assure yourselves of this.

    Furthermore, if the car is doing 100 mph (~45 m/s) while following this diagonal line from the outside of the road towards the inside "apex", then it will spend almost a full second with its suspension constantly "Twisted" by 7 mm. So the Twist is effectively "steady-state". But when exiting the corner, from inner apex to outside of road, the Twist will be in the opposite direction!

    And even furthermore, if the road surface is smoothly cambered "concave up", as is common with banked corners, then the Twist introduced by a diagonal driving line is of the same magnitude as above, but of opposite sign.

    So, the big question:
    What does this twist-in-the-road do to your precisely calculated wheel loads?

    Based on the (quite soft) corner-spring and ARB rates in the Milliken example, the 1/4" Twist changes the wheel loads by about +/-160 lbs! And depending on which way the Twist is, the LLTD ends up being either 74%F, 26%R (for corner entry of convex-up road), or 34%F, 66%R (corner exit, convex-up road). Put simply, the handling balance changes from massive understeer on corner entry, to massive oversteer on corner exit. Yippeeee!!!

    Anyway, there are a whole lot of other effects which should also be considered, some of which lessen the above changes, others which exacerbate them. But the bottom line is that with conventional suspensions, all your precise "handling balance" calculations get tossed out the window as soon as you put the car on a real road. And THE STIFFER THE SPRINGS, especially the Roll and Twist-mode stiffening Lateral-U-Bars (= ARBs), THE WORSE! Please do the calcs.

    Finally, it is worth noting that FSAE's short-wheelbase-small-track cars don't feel the above sort of twist-in-the-road as much as larger cars (because the further the wheelprints are apart, the further the road surface moves out of a flat-plane). But any "twist-in-the-road" will still change the wheel loads of your FSAE car.

    How much? Easy to measure! Put your car on its four corner scales, on FLAT ground. Adjust your spring-mounts so that the corner-weights are symmetrical side-to-side. Now slip a 6 mm thick piece of plywood under two diagonally opposite wheels (or a single 12 mm piece under one wheel). This represents a Twist-mode of 3 mm (1/8"), which might represent some parts of some of the "wilder" FSAE tracks. Write down the changes in the four wheel loads.

    Now ask yourselves why you bothered doing all those precise "handling balance" calculations in the first place. Because, with conventional suspensions, the road decides what the LLTD is, not you!

    Last edited by Z; 03-20-2014 at 09:05 PM.

  5. #5
    Quote Originally Posted by Z View Post
    the handling balance changes from massive understeer on corner entry, to massive oversteer on corner exit. Yippeeee!!!
    Funny, thats what just about EVERY NASCAR driver complains about ALL THE TIME. Especially when comparing the "Old Cars" to the "COT" (with it's life savingly strong center section)

+ Reply to Thread

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts