adventures with alignment and tyres...

[johnl]

The problem; my Alfa's steering has been pulling to the left, and noticeably lighter when steering left vs steering right. It's not major, but quite annoying, and destroys the subtle tactility of the steering 'feel' (as even a small steering pull / lead always does in my opinion). This seems a very common issue, a lot of the cars I have driven over the years do this, but none ever seem to suffer a consistent steering pull to the right (...?). Before you say it's a road crown issue, it isn't, trust me.

So, it seems not unreasonable to assume an alignment problem. I do my own alignments (string lines, digital angle gauge, and various bits of steel tubing jigs used to measure in conjunction with the angle gauge). I can measure toe, camber and caster quite accurately, and also place weight in the driver's seat to simulate the cars most common condition (i.e. with 70kg of me and no passengers). Car is levelled with shims placed under the tyres as needed to create a 'flat' platform (car level to 0.1°). Wheels have been lightly machined at the outer rim to eliminate error caused by wheel run-out.

Checking alignment, I find:

Rear toe-in 1mm per side at the rims (17")
Thrust angle = 0°
Camber = 0° LR and + 1° RR (i.e. positive camber...)

I have equalised rear camber by replacing one of the rear 12mm strut to hub bolts with a 10mm bolt, which gives significant camber adjustability. Rear camber is now 0° both sides. Some might be worried that the assembly will move in use with the smaller OD bolt, but it hasn't, and won't (unless I hit something fairly hard, in which case I would expect the flimsy rear control arms to bend first...). All rear bushes seem in good condition. Nothing seems bent.

Front toe-out = 1mm total
LF camber = 0.7° neg
RF camber = 0.3° neg
(cross camber = 0.4°)

My car is pulling left. Accepted wisdom says that this camber difference should cause a pull to the right, i.e. toward the side with the least negative camber. Cross camber values approaching 0.5° are 'supposed' to be when a steering pull may become problematic. I'm not sure it's quite this simple...

LF caster = 3°
RF caster = 3.2°
(cross caster = 0.2°)

Not much difference here, but still. According to accepted wisdom this caster difference should cause a pull to the left (pull toward the side with least caster angle), as with my car. Trouble is I don't really believe it, because this is something I have experimented with in the past (trying to cure another steering pull problem with a Honda Accord). With that car, even very large changes in 'cross caster' made very little if any difference to the steering pull or relative 'weight' left / right. I think it would probably be different if the scrub radius were substantially positive (as with many RWD cars), or maybe substantially negative, but the Honda had near as damn it zero scrub radius (so 'trail' forces follow directly behind the contact patch centre-point, and are not offset to the side of it, creating pull if forces are unequal), and I'd be surprised if the 147 is much different...

Anyway, if nothing else but to eliminate any known asymmetry (even if relatively small) as a possible culprit (while hoping any change would actually fix the problem), I was determined to achieve a properly symmetric alignment.

In an attempt to make a remedial camber change (both sides, equal but opposite) I loosened the sub-frame and levered it laterally toward the right side. Disappointingly the sub-frame could only be moved about one millimetre or so, which would equate to about 0.1° of opposite but equal camber change at each front wheel (not nearly enough).

Depending on direction of ball joint location change, a 1mm location change of an upper or lower BJ will result in approximately 0.1° of either camber or caster angle change, so 10mm of BJ location change equals near enough 1° of camber or caster change (this is due to the distance between the centres of the upper and lower ball joints, about 520mm from memory).

I needed much more BJ relocation movement, so I turned my attention to moving the upper BJs. The upper BJs are attached to the upper 'wishbone', so moving the entire 'wishbone' will result in camber or caster change (depending). To increase RF neg camber (to match LF camber), I removed the aluminium casting to which the upper 'wishbone' is attached. I then used a large rat-tail file to inwardly and laterally elongate the four chassis holes that the four bolts pass through. 4mm of lateral elongation resulted in dead equal camber side to side (and they said it couldn't be done...). This is all with simulated driver weight.
Note also that there is a 'lug' on the aluminium casting (about the size of a 20c piece, you can't miss it) that protrudes through a hole in the top of the suspension tower. This lug must be trimmed to allow the casting to be relocated in the required direction.

More or less the same procedure was done on the left side of the chassis to increase LF caster (to match RF caster), except that the hole elongation is longitudinal toward the rear of the car, not lateral. With slight sub-frame relocation and significant hole elongation, the car now has dead equal front camber and caster angles, with 1.2° neg camber and 3.2° caster.

I have also introduced front some toe-in as I have always found that it improves steering feel and directional stability (and the 147 seems no exception). My experience is that toe-out doesn't really improve turn-in all that much, if at all, but does increase tyre wear and make the car unpleasantly 'wandery'.
Side to side wheelbase is within two millimetres, so a non issue. Other than somewhat too soft front dampers (new ones on order) no suspension parts are significantly worn, all bushes and BJs in good condition (previous owner replaced a lot of front suspension parts shortly before selling the car).

So, the alignment is now near enough spot on perfect (I have confidence in my measurements), but how is the steering pull? It was still there...

So next I tried introducing some deliberate asymmetry into the alignment by adjusting the thrust angle to the left and then to the right of the chassis centreline, and even with quite a lot of thrust angle misalignment the steering pull remained (regardless of thrust angle direction, and now reset to 0°).

All this tells me the steering pull problem probably has nothing to do with the alignment, which leaves the tyres. The previous owner had very recently fitted some P1 Pirellis to the front, I'd fitted some Maxxis Victra 511s to the rear, so very near new tyres front and rear. I tried fitting the Maxxis to the front and Pirellis to the rear, no real change. Next I tried a 5psi difference in side to side pressures, no change, regardless of which side had the higher or lower psi. Next I tried side to side rotation, no change.

Next I tried moving the tyres front to rear on one side only, so I had a Pirelli on the RF, a Maxxis on the LF, a Pirelli on the RR, and a Maxxis on the LR.
Interesting result, the steering pull became substantially worse. At last something I had done had made a significant change, albeit in the wrong direction, but at least the magnitude of the problem had changed whereas before nothing had really had much of an affect.

So, not daring to get my hopes up... I changed the front tyres side to side. Bingo, steering pull almost completely gone, hoo bloody ray! The steering now feels more or less equally weighted and responsive turning left or right, has much better feel, and is much nicer in every way.

I'm not yet entirely convinced the pull is absolutely 100% gone. The car now has a Pirelli and a Maxxis on each side of the car, and same at the front and rear ends (i.e. Pirellis are diagonally opposite each other, as are the Maxxis). I'll try swapping the rear tyres side to side so both Pirellis are on the left side and the Maxxis both on the right (maybe tomorrow, or might drive a bit more as it is to get a fuller feel for the change). This may make a slight change (good or bad), or none at all.

I think it's likely that this steering pull might have been (?) to do with 'PRAT' (plysteer residual aligning torque) in the tyres (all radial tyres have this, it's just a question of how much), and that there is (might be) a mismatch between the Pirellis and the Maxxis tyres...? It would be an interesting experiment to fit identical tyres all round to see if this also eliminated the pull, but finances get in the way of science...

Regards,
John.

[colcol]

John, a wheel alignment bloke told me years ago that this can be caused by bad tyres, being heavier on one side of the tread to the other, so one side is always favouring to turn left or right, this was a problem with those new fangled radial tyres of the 70's, but most tyres are the same weight on the left or right side of the tread, with the increase of quality of tyres over the years.
The tyre manufacturers used to check that either side was the same weight, before sending them to the car companies as it would cause trouble like to which you are having.
In the olden days, they would send their dodgy tyres to the discount houses, and send the good ones to the car makers and their named tyre outlets.
Doing your own wheel alignments....gee i thought i was the only one that did that!, go to a race track and see Formula 1 and Supercars being aligned with string lines, so it must be accurate, Colin.

[johnl]

Colcol,
I don't think it's one side of the tyre being heavier than the other (actually, I'm pretty sure your alignment bloke was wrong in his understanding, though not that tyres can cause a steering pull).

I think what you have been told is more to do with 'conicity' in the tread. My understanding is that there are two kinds of tyre conicity, one being a 'true' conicity caused by the tread wearing at an angle across it's face, and the second being a kind of 'effective' conicity caused by the radial belts not being quite centered relative to the tread center-line.

The former acts somewhat like camber, i.e. the tyre tries to roll along a curved path because one side of the tread is of greater diameter than the other. The latter (laterally offset belts) causes one side of the tread to be stiffer than the other, so as it carries load one side of the tyre deforms in a manner that it causes the tread to act like a section of an effective cone. Conicity (externally visible or internally invisible) can be diagnosed if the direction of any steering 'pull' caused by it changes direction when the tyre is swapped to the other side of the car, or 'flipped' on the rim.

Then there is PRAT (ply-steer residual aligning torque), caused by the 'diagonality' of the 'grain' of the radial belts in the tread (keep in mind that the belts are steel wire woven 'textile'). There are at least two radial belts laid over each other with more or less opposite 'grain diagonality', but the outer belt has the greatest effect, so one belts PRAT doesn't fully counteract the other belts opposite PRAT (at least that is my possibly faulty understanding...). PRAT is also affected by tread pattern, to a lesser degree than the the internal belts. A steering pull (or tendency to drift or run at some slight angle other than that in which the wheel is nominally pointing) caused by PRAT cannot be changed by changing the orientation of the tyre (unlike conicity).

If you have a tyre with a PRAT that pulls strongly one way, then it is possible to counteract this by fitting another tyre on the same axle line with an opposite PRAT tendency. It's my understanding that the Honda NSX specified original tyres just like this, for this very reason, i.e. tyres on the left side (and marked as 'left') had an opposite PRAT tendency than those on the right side (marked as 'right'), these tyres having the left and right side versions made with oppositely laid up belts. I suspect this might be more or less what has occurred when I fitted different brand tyres on the same axle line...?

Please understand that my understanding of this topic is very limited...

Regards,
John.

[cc]

Hooray! a new 932 discussion topic!  a good one too

[poohbah]

Please understand that my understanding of this topic is very limited...

If you don't mind me saying John, I think you might by underselling your grasp of the dark arts of wheel alignment!

I bet you drive your missus nuts...

[bazzbazz]

As Manuel would say . . ."Que?"   

Baz

[johnl]

Just to update,
As per the first post, I had the tyres fitted so they were mismatched side to side on each axle line, which did improve the steering pull but not eliminate it. I didn't like the idea of the tyres not matching across each axle line, so reverted to fitting the two Pirellis on the front and the two Maxxis on the rear. Still not sure the pull is being caused by the tyres themselves.

I do have an uneven ride height issue, where the right front is 15mm lower than the left front (measured to wheel arch lip). The heights are:
RF = 670mm
LF = 685mm
RR = 700
LR = 705

Something is not right here, maybe it's contributing to the steering pull? I have a plan to make front ride height adjustable...

Regards,
John.

[colcol]

Sometimes car manufacturers put a slightly longer spring on the drivers side, to compensate for the weight of the driver, so that when the driver is seated, the car sits at the same height.
The same way some car makers put the exhaust system on the opposite side of the driver, to balance out the weight difference.
As a Bargearse, this was pointed out to me by a wheel alignment place years ago, as sometimes the wheel alignment will be spot on when aligned, but when Tubby sits in the drivers seat the alignment goes out a bit, what were they implying?, Colin.

[bazzbazz]

Just something one should remember, the rear suspension of your car contributes just as much to the steering, handling, ride height ect as the front does. If you have a faulty/worn/sagging corner on the rear of the car it will affect the front dynamics.

Just a thought.

Baz

[johnl]

Colin and Baz,
I pulled the front 'coilover' units out and stripped them down so that I could measure the free length of each front spring. They are identical in length, no sign of either having 'sagged' to any degree (unless both have 'sagged' very slightly to an identical degree). If both springs have the same free length then they will also be the same length when both are loaded by X weight.

Checked the rear springs too, which are very slightly different length, but it's only about 2 or 3 mm (at most). I am scratching my head a bit as to just where the uneven ride heights are coming from.

Regards,
John.

[bazzbazz]

May I suggest this:

1/ Get a pair of jack stands, set them to the closest you can of vehicles normal ride height, making sure each jack is
     identical in height
2/ Jack up the rear end of the car
3/ Place Jack stands under rear of car at the exact same points on either side, in the same orientation, lower car onto jack
    stands.
4/ Measure your front ride heights. (Note, use the top of rim as measuring point, not top of tyre)
5/ Repeat above steps with front end.

The results may surprise you.

Measuring springs unloaded is not very useful, they have to be under equal loading to give any indication to their condition. Also by doing this method, it removes the influence of a faulty component at the OTHER end of the car on the end you are measuring. Also a faulty component on one corner can make one or more on the others look suspect.

Good Luck

Baz

[bazzbazz]

If both springs have the same free length then they will also be the same length when both are loaded by X weight.

Sorry, that is only true if the springs are new/are identical in condition. I have seen springs that are of same length & gauge, same materiel, collapse at different rates under the same loading. This can be caused by many things over time, heat, stress ect, ect.

Also if you have coil overs on the front with the rear on jacks you should be able to adjust the coil lengths so that you have equal ride heights from left to right.

One point I forget to mention in the above post, the jack stands must NOT be mounted on the usual "Sill Jack Points", they must me sitting on a something at the rear like the forward mount of the trailing arm and at the front on the inboard jack/hoist mounts. This allows the weight to "pivot" slightly and not to be influenced by a flat support surface. (I hope that is clearer than mud)

And just be careful.

Baz

[johnl]

Baz,
It's my understanding that: the stiffness of a spring (coil, torsion bar, whatever) doesn't change over time / cycles unless the metal is actually starting to fracture (micro cracks, or worse). If any cracks are appearing, then that spring will fail (break) fairly quickly in further usage (a crack being a stress riser that will quickly propagate with continuing cycles of deflection). If no cracks exist, then the spring will retain it's 'rate' (i.e. stiffness) even if that spring has lost free length (i.e. sagged). A spring that has started to crack is also very likely to have lost at least some free length, which is not to say that a sagged spring will have cracks, most often it won't. I think it very unlikely for two dimensionally identical springs to have different spring rates, especially if one has not 'sagged' more than the other...

Also, since all steels that are remotely likely to be used in a spring (of any likely steel alloy) have the same modulus of elasticity (within about 1%), then it doesn't even matter what steel is used in any given spring, nor what state of heat or work hardening the steel may have been subjected to (as long as the steel isn't stressed to or beyond it's elastic limit, which will cause permanent deformation). So long as two springs are identically dimensioned (wire diameter, wire length, coil diameter etc), then those two springs will for all practical purposes have an identical spring rate. So, if you had two similar springs, one of which had sagged and the other not, then both springs would still have the same spring rate (unless one were cracking up, which is relatively rare). You can shim the shorter spring and restore equal ride heights, with both springs still supporting X weight to the same degree. The difference between a good spring steel and a 'bad' steel is that the good steel will last longer than the bad, without fracture or sagging.

As I said, the above is my understanding, but I think my sources are sound...

Regards,
John.

[105gta]

I'm assuming you have thoroughly checked for wear (very common in 147) in all you suspension bushes?
Have you considered the sway/anti roll bar, if you disconnect one drop link does the threaded rod (for instance ) that goes through the strut mounting sit central in its hole or do you have to apply force to get it to line up? Could be transferring more vehicle weight to one side. Tyres can be very sensitive at times.
Also all your measurements are taken at rest position have you fitted a load bar between the front wheels and the rear for the matter. 
Brake drag?
What condition are you struts in? Both damping and SAI
Wheel bearing?
Do you have a q2 Diff? 
Many variables

[johnl]

105gta,
you assume correctly. I've checked all moving joints in the front and rear suspension / steering and found nothing wrong. The previous owner had the upper 'wishbones' and lower control arm bushes replaced not long prior to my buying the car. No brake drag. No pre-load in the ARBs. I don't know what you mean by "load bar".

I experimented with preloading the 'struts' by modifying the left side 'Y' shaped yoke that attaches the damper body to the lower control arm. This modification allowed me to lower or raise the yoke on the damper body by about 10mm either way, which alters ride height and changes the load on the diagonals. Keeping in mind that the steering pulls to the left, lowering the damper on the left side (thus unloading the left front and right rear, while increasing load on the right front and left rear) caused the steering pull to become worse. Raising it on the left side (thus increasing load on the left front and decreasing it on the right front) caused the steering pull to disappear. However, the handling and degree of body roll became quite asymmetric in left vs right hand corners. It seemed I'd created a pull to the right that was more or less counteracting the pull to the left, but not actually curing the root cause of the problem, while creating others.

Anyway, widely accepted wisdom tells us that if the caster angle is unequal side to side, then the steering will tend to pull toward the side with least caster. Earlier I'd increased the left side caster in order to cause the steering to pull less in that direction, as 'wisdom' tells us it should do. It didn't help, in fact I had the impression that the left pull had increased slightly, which I put down to some sort of anti-placebo affect (since this didn't fit with anything I have ever read related to caster and steering pull). Recalling this, I set left side caster back to where it had been previously, i.e. decreased left side caster which in theory ought to increase any pull to the left. Except the steering pull decreased quite a lot, which doesn't make sense according to accepted wisdom...

It still pulls left a bit (still mildly irritating, as it was when I bought the car). It seems that this car for whatever reason (related to interaction with other geometries I assume) reacts to unequal side to side caster angles in an opposite manner than most / many other cars do. I now plan to increase right side caster a little bit and see if this gets rid of the remnant steering pull. There is about 2mm of longitudinal movement in the subframe holes, which might be enough. If not then I can elongate the four holes in the 'strut tower' (that attach the wishbone to the chassis).

We'll see what happens...

Regards,
John.