25 June 2003
We've been without a phone line since February, so I've just received your March to June Newsletters. Some comments:
Re: IRS for Dirt (March)
You say "With IRS, we only have thrust anti-squat to work with ...", and later when discussing inboard brakes you imply that only thrust force at the axle can be used for anti-lift. True with a conventional IRS, but it ain't necessarily so!
One way to have the wheel hub moving vertically in side-view and still have plenty of anti-squat is to do what VW did when they adapted the Beetle drivetrain to the early Kombi - fit a set of drop gears at the wheel hubs. This lowers the overall gear ratio (good for the low powered Beetle engine) and raises ground clearance. It also gives these Kombi's about 100% anti-squat under power even though the side-view-swing-arm (SVSA) is close to horizontal. On a low CG buggy with this suspension there is perhaps 200% anti-squat - great for lifting yourself out of the mud with power-on.
One way to estimate the amount of anti-squat with this system is to lock the drive shaft, mark the tyreprint position (ie. at 6 o'clock on the tyre), and then move the wheel up and down. The angle of tyreprint movement relative to vertical, in side-view, gives the amount of anti-squat (as with a live axle). The ratio of the drop gears varies the anti. Using a chain, or three gears, or internal and external gears, can give pro- or anti-squat depending on ratios.
The drop gears are an extra complication, so more weight and more $$. But then so are things like birdcages, etc. With powerful engines the drop gears would be quite heavy. But they could be used in a similar way to a Quick-change for easy ratio changes. With different ratios, left and right, you would vary the wedge under power. With a spool diff. and different ratios you would have "invisible" stagger. You could run a smaller right rear wheel and still turn left better than the opposition (maybe). And when they try to copy your tyre setup ...!!!
There is another way to get rear anti's with IRS, both under power and braking.
Consider power first, with an IRS with standard inboard diff. We mount the diff to the chassis with two pivots close to the axle output shafts. The diff can thus rotate about its crown wheel axis in side-view. We then fit a
linkage to the nose of the diff that resists the diff's torque (ie. Stops the pinion climbing around the crown wheel). The "other end" of this linkage is connected, possibly via the suspension, so it acts vertically on the wheel hubs.
The actual linkage can take many forms - mechanical push/pull-rods and rockers, passive hydraulic rams and hoses, etc. If the linkage is such that a nose up force at the diff acts to push the wheels down, then we have added anti-squat. If diff-nose-up lifts the wheels up, then we have added pro-squat.
The amount of pro or anti depends on the ratio of the linkage. The more the nose of the diff moves for each inch of wheel movement, then the more powerful the pro or anti. The linkage can be symmetrical, acting equally on both wheels, or it can be asymmetrical, perhaps acting only on one wheel. Or it can be really asymmetrical, lifting one wheel and planting the other.
Again, the linkage adds weight and $$. The fact that the diff moves during wheel bounce adds unsprung weight. But the rotational pitch inertia of the diff is much less than the translational inertia of the whole diff plus live axle bouncing up and down. And if you've really got to have the pro/anti-squat, then the complication is no worse than that of say pushrods and rockers.
In fact, if the diff's nose is attached to pushrods that rise upwards to conventional type rockers, with further pushrods down to the wheels, then the setup would resemble a "monoshock" suspension. A single coilover
controlling pitch motion of the diff would control the axle-bounce mode.
The diff's torque would act in parallel with this bounce spring, giving the extra anti-squat. A laterally compliant link between the diff's nose and the two upward pushrods would allow roll motion. Springing the lateral movement of this link would control the axle-roll mode. Easier to explain with a sketch, but not too complicated in practice.
A similar system could be used with inboard brakes. For example, mounting disc brake calipers to the above diff would give the same amount of brake-anti-lift as power-anti-squat. If we mount the inboard brake calipers to a frame that can pivot in pitch, and link that frame to the wheels similarly to above, then we can have vertical wheel hub movement with arbitrary amounts of power and braking antis.
But wait, there's more! With conventional suspensions, beam-axle or independent, the accelerating and braking antis only work at that end of the car. So no matter what anti-squat we have with a rear drive car we
cannot stop the front of the car lifting during acceleration.
BUT! Assume we have an IRS with pivoted diff as described above. We control diff pitch with a single acting hydraulic ram such that oil pressure pushes the nose of the diff down. We connect this ram via hoses to rams which act to push the rear wheels down - so far, a hydraulic anti-squat system similar to the previous pushrod system. We next extend the hoses to rams at the front suspension which act to push the front wheels up. (All hoses are connected without any valves, so equal pressure at each ram.)
We can now keep the car perfectly level during acceleration. We can even make it dip its nose and raise its tail. Bounce (heave), roll, or twist (warp) modes of the suspension cause no movement of the diff. They just
push oil from some rams to other rams. Only suspension pitch motions act to move the diff, so the torque at the diff acts only on the pitch-mode of the suspension. This system would fit in very easily with my Balanced
Suspension, or other such interconnected systems. The ram at the diff nose would simply be connected to another ram that acts in parallel to the pitch-mode spring.
Not enough? There's more! If we have a mid-engine with transaxle, then the above system is difficult. We might have to allow the whole engine package to move in pitch. But no, we don't have to.
With any of the above anti's, including inclined SVSA, we are simply using engine torque, or a force derived from it, to add a compressing or extending force to the suspension. So imagine we break into the driveline somewhere, say at the gearbox input or output shaft. We cut this shaft in half and mount an oil pump in the middle. The engine side of the shaft drives the oil pump input shaft, and the oil pump housing is attached to
the output (wheel) end of the shaft. We attach the outlet port of the pump to our pitch control ram(s) - one ram for Balanced Suspension, four for a conventional suspension.
Since these rams have only limited movement the oil pump will never pump a continous flow of oil. It will always be "hydraulically locked" (the oil pressure will be a reasonable measure of the engine torque). When the engine torque increases the pump input shaft will rotate a limited amount relative to its housing, thus pumping a small amount of oil and extending the pitch control rams a short distance (assuming rear load transfer doesn't overpower the rams, ie. >100% anti-squat). When engine torque decreases the rams will push oil back into the pump "unwinding" the driveline shaft. This winding-up and unwinding of the driveline is
essentially the same as when we have an inclined SVSA (ie. the wheel rolls a little forward or backward during wheel bounce.)
This oil pump system virtually eliminates the extra unsprung mass of the pivoted diff. It also would be reasonably light and compact, although still an extra expense. It could also be used to damp driveline shocks, etc., and it wouldn't add much frictional drag (there would be some oil leakage but the oil isn't being continuously pumped). Best of all, it is completely "passive", with none of those unreliable electronics!
Likewise, with a suitable interconnecting linkage (mechanical or hydraulic) between pivoted inboard brake calipers and the suspension we can have any braking anti's at any end of the car. Nose-up-tail-down braking if we want - like a horse!
Personally, I'd be happy with slightly inclined SVSA's. But if you've got to have those extra anti's, they're there for the taking.
Re: Acceleration Sensitivity of Dampers (Jan/Feb NL's)
Just a quick thought. For a quick test for acceleration sensitivity, test the damper, then test it again upside down!
If the valve shims or disc are on the stationary part of the damper (in the damper rig) then it probably won't show much acc. sensitivity. When the valves are on the oscillating part then any acc. sensitivity is more likely to show up.
This would also suggest that if you don't want any sensitivity effects, then it is probably better to mount the damper on the car so that the critical valves are on the chassis side (which is subject to smaller accelerations than the wheel).
Re: Tyre Load Sensitivity (April/May/June NL)
You write: "Contrary to the contention of some writers, tire load sensitivity ... does not reverse or work backwards during entry or with cold tires".
I guess this refers to Chuck Hallum's views, as expressed in the recent Racecar Engineering article V13N07?
I can think of several reasons why moving weight forward on a racecar might improve turn in, without having to suppose reversed TLS. What are your views?
Sure enough, you're right that it is possible to get torque anti-squat in an independent rear suspension with drop gears and other devices that make torque react through the linkage or otherwise lift and/or pitch thecar. I also give you credit for being crazier than me for spending the amount of time you must have spent trying to think of multiple ways to do that! With your comment about no phone line since February, I picture you in a tiny cabin in the outback, drinking way too much coffee.
I think a VW Transporter also has some thrust anti-squat. I don't have one around to look at, but I think the torsion bar is above the wheel center a bit. I can't recall noticing a VW Transporter lifting under power, but I can't recall being able to feel much forward acceleration in one either.
I like the idea of trying the shock both right-side-up and upside-down in the dyno. Like trying different strokes and frequencies, it would only provide some indication, but it's certainly something you can do easily, provided of course that you're working with some form of gas shock. I actually managed to think of this one myself a while back, but you have prompted me to think about it some more. I think it would be particularly useful for separating inertial effects in the valving elements from effects due to fluid compressibility and entrapment.
Regarding entry, load sensitivity, and Chuck Hallum's theories, yes I was thinking of him when I wrote that comment in my newsletter about load sensitivity not working backwards on entry. I do agree with you that there are ways that more front percentage could loosen entry, without tire load sensitivity having to work backwards. To name just one, in Winston Cup, it is still common to use Ford Galaxy single master cylinders, meaning brake bias is non-adjustable, and roughly the first third of the turn is taken on the brakes. Reducing rear percentage without reducing rear brake puts the rear tires closer to lockup in braking.
In conversations with me, Chuck maintained that more front percentage always improved (meaning loosened) entry, including racing applications, including oval track. I told him it wasn't necessarily so at all. He didn't want to hear it.
In his article, Chuck cited three references from the Milliken book, all from the chapter about setup. The mention of turn-in there specifically refers to response to a step-steer input. As you probably know, that means an abrupt movement of the steering wheel, with roughly constant throttle to maintain as nearly constant speed as possible. So brake bias wouldn't be a factor in that. However, a step-steer maneuver is a passenger car manufacturers' test track technique. It's not the way a competent driver drives a race car. Assuming there is a straight before the turn, turn-in will usually occur with some amount of braking. Steering input is only abrupt if the turn is tight, and even then braking is usually present.
My racing clients never do step-steer tests. Perhaps some race team somewhere has done them. If a team works a lot with a passenger car company's engineers, that would logically increase the chances. If we are thinking about passenger cars, Trans-Am cars, stock cars, or production-based road racing cars, that usually means we have a front engine, usually with a set-back limit, and ballast toward the rear. In that case, moving weight forward means moving ballast or maybe the battery toward the center of the car. That means that we are decreasing the car's yaw inertia as we increase its front percentage, and that will surely make the car turn in more readily.
Now, if we have a case where moving mass forward, with polar moment of inertia in yaw held constant, really does cause increased yaw acceleration for a constant steering input pattern, then we would have a case that would suggest negative or reversed tire load sensitivity. But I question whether that really happens. I would want to see the data.
I hear plenty of complaints about sluggish turn-in from drivers of nose-heavy cars. Of course, other factors enter into any individual case.
The other two references from Milliken both deal with quickening turn-in by increasing roll stiffness -- in one case with springs only, in the other case by whatever means. If the recommendation had been to increase front roll stiffness only, then that would contradict conventional understanding of tire load sensitivity. But the recommendation is to increase overall roll resistance, not alter its distribution. Therefore, Chuck's inference of apparent reversed load sensitivity from these comments is not logical.
I very often work on turn-in properties by varying instantaneous diagonal percentage by various means, including dampers, relative front and rear roll resistance, relative right and left pitch resistance, and static wheel load adjustment. My experience is that the car gets tighter with more diagonal during entry, just like it does elsewhere in the turn. If tire load sensitivity worked backwards on entry, I'd have a lot of very dissatisfied clients, and I'd be very puzzled.
I have had cases where people report looser entry with increased diagonal, but these are invariably cases where the car is being slowed mainly by the rear wheels. What's happening there is that the right/left distribution of rearward force is creating a more significant yaw moment than the front/rear distribution of lateral force.
Chuck was a nice person, and spent a lot of effort trying to get that paper published and otherwise promote the thinking expressed in it. The last time I saw him was when he stopped by UNC Charlotte trying to interest the faculty there. I was given an advance copy of the paper, and informally asked to share whatever opinions I might have, so I got to consider the paper at some length.
Not only do I think alternate explanations exist for looser entry with more front percentage (which only occurs in some cases), I also have problems with his inferring reverse load sensitivity from the shape of the isothermal curves in his Fig. 4. Actually, taking lateral force coefficient values along the isotherm, you see the same direction of lateral force coefficient change with load - and more of it! Reduced loading adds lateral force coefficient, even more at a given equilibrium temperature than it does at peak lateral force values. The slopes of the curves are opposite to that of the line of peaks, but that only tells you that to get a given equilibrium temperature at a higher load, you have to run at a smaller slip angle (implying sub-limit operation), whereas to reach peak, or limit coefficient you need a greater slip angle at greater load. It doesn't imply that load sensitivity is reversed. The conclusion simply doesn't follow from the data.
In fact, looking at a tire's lateral force coefficients for thermal equilibrium at a given temperature, in given ambient conditions -- or for a given heating factor -- at various loadings, tells us nothing at all about its transient limit lateral force coefficients at that momentary temperature, when generating maximum available force and therefore undergoing rapid temperature rise. When we look at a set of thermal equilibrium situations for a single temperature, with varying loads, we will be looking at sub-limit operating conditions for all points except possibly one.
I think the way to actually find out what temperature does to limit lateral force coefficient would be to vary ambient temperature when testing on a TIRF machine, measure tire tread temperature while running, and take readings at conditions stable enough to allow good limit data, with different tire temps. This would be a somewhat complex process, because you have to try a tire at various slip angles to find its limit, and the temperature is changing all the while.
In the course of exploring Chuck's theories, I e-mailed Doug Milliken, knowing he would have experience with laboratory tire testing. I asked if he had ever seen data that would lead one to suppose that load sensitivity reverses at low temperatures or in transient conditions. He said he had not. I asked him how long a tire is generally run before taking readings. He said it varies, and this is usually determined by the individual company testing standards that prevail at the client tire or vehicle company. I'm going to bcc Doug here. Hopefully if I've misunderstood anything he'll straighten me out.
To me it seems obvious that in most real-world situations, and in laboratory testing of limit lateral force coefficient, tires never do reach thermal equilibrium. In freeway driving, yes. But not in limit cornering. They may reach a steady enough lateral force to allow an accurate reading, but that's steady state in terms of force and slip angle, not steady state in terms of temperature -- that is, it's not the same as a condition where heat input equals heat dissipation. As Chuck points out, tires just keep getting hotter all the way through a turn, and don't start cooling off again until they get to a straightaway. Thermal equilibrium for a given load will surely depend enormously on track and air temperatures, and airflow to the tire -- plus vehicle speed, as Chuck also notes. I am inclined to suppose that in many cases it is actually impossible to reach thermal equilibrium at the limit of lateral force, at high road speed; the tire will get hotter until it comes apart, or until the tread wears off, if you drive in a circle long enough. Or if it does eventually reach thermal equilibrium, this will be a condition little resembling real-world cornering, and the tire will be very hot indeed.
I couldn't find it in the Racecar Engineering version of the paper, but the preliminary draft I had mentioned Chuck's belief that load sensitivity works backwards in road car limit cornering, because this is mainly a short-lived situation and therefore occurs on cold tires. If that were so, a lot of people would surely have found out by now, because everything we do to tune road cars wouldn't work. I have plenty of direct personal experience with road cars, and I guarantee that load sensitivity in street tires works the way conventional theory would predict, in all weather, and in the first of a series of turns or the last.
Race tires do have an optimal temperature range, above or below which they don't work as well. It is true that if a pair of tires are below optimum temperature, loading them more unequally will move the loaded tire more toward optimum temperature and the unloaded one away from optimum, and loading them both more will move them both toward optimum temperature. If both tires are above optimum temperature, loading them more unequally will move the loaded one away from optimum and the unloaded one toward optimum, and loading both more will move both away from optimum. This does imply that grip will improve as we go through a turn when the tires are cold, and will deteriorate as we go through a turn when the tires are hot. It implies that making the car (or one end of the car) heavier, or the tires narrower, hurts the car (or that end of the car) less at low temperature than at high temperature. It may also suggest that load sensitivity should be less at low temperatures. But it doesn't imply that unequal loading helps grip at below-optimum temperature, or that we get a higher lateral force coefficient with higher load at low temperature. Low temperature may reduce load sensitivity, but it doesn't reverse it or make it negative.
The only magazine I know of that still tests tires is Grassroots Motorsports. Their tests include sustained lapping on road courses, with a view to seeing if the car gets faster or slower as the tires get hot. Mostly, they test performance radials for sports cars, both true street tires and "R compound" ones intended for competition. What they find is that, with some exceptions, real street tires work best when cold, and deteriorate with increasing temperature, at least if "cold" means a comfortable summer temperature. With racing compounds, it is more common for heat to help up to a point, although this is not universal either.
I might mention that there are parts of Chuck's paper that do look okay to me. I find his theory of a localized grip-slip cycle within the sliding portion of the contact patch entirely plausible, and his method for calculating heating factor makes sense.
The faculty at UNCC told me they shared reservations about the paper. Nevertheless, they did schedule Chuck for a presentation at the University. Regrettably, he never got to deliver it.