SleeperGSR

i would be interested in hearing why it depends on if the sway bars are disconected. I had my corner weight done at a shop and they left everything connected. I didn't really want to argue with the guy about it, price was good and everything, but i read that they should be disconnected. Thoughts or opinions?

beanbag

If your swaybar has adjustable endlinks, you should disconnect it. The reason is that even though the swaybar is supposed to act like a linear spring and thus preload doesn't matter, in actuality due to bushing slop, the swaybar has a little bit of a floppy region when it is unloaded. For symmetry in making right and left turns, it is best to adjust the endlinks so that the swaybar is un-tensioned at static ride height.

If you have a swaybar with non-adjustable endlinks, you should leave it on during corner balancing. This is so that you don't throw off your corner heights and weights if you re-attach it at the end, and it has a little bit of preload. The equations get a little more complicated because now the right hand knows what the left hand is doing.

Ekasey

Is there a problem here?

solo-x

Not surprising at all. The engine is on the right side in that situation, balancing out the driver's weight. Stock, the engine and driver are on the same side of the car.

Stinkycheezmonky

In theory then, a RHD Honda with a stock motor would most likely be 50/50, or close to it then, yes? Along that line, my brother's 240SX with an SR20 was like 54/46 L/R, with him on the left side, as well as all the steering, brake, and turbo garbage. We figured the RHD chassis would be the only way to get proper L/R balance.

descartesfool

50% cross-weight is not the correct way to corner balance a car that does not have 50% left side %. If you work out the left and right turn cornering forces on all four tires assuming a simple model of tire grip vs vertical load, you will get uneven total max grip for left and right turns if you adjust for equal cross-weights. Now that is for figure 8 track. On a real track, which goes mostly clockwise, you may want to optimize right turns over left turns for more overall speed. Then comes braking and accelerating. If you put equal weights on front tires, FWD car will accelerate better in a straight line, and have more even grip under braking for the front tires which are doing the bulk of the braking. If you put equal weights on rear tires, car will be more stable under braking, and less likely to lock up one rear tire, and accelerate better if RWD.

The axle that gets the most uneven tire load under cornering loses the most % grip, so if the LF weight is higher than the RF at rest, when cornering right, the load will be more un-even between the front tires and thus provide lower front axle grip than when cornering left for the same axle. Say LF is 850 lbs and RF is 790 lbs, and you transfer 500 lbs under cornering. On a right turn, LF goes to 1350 and RF goes to 290 (diff between tires of 1060 lbs). On a left turn, LF goes to 350 and RF goes to 1290 (diff between tires of 940 lbs) . The difference across the axles between a left turn and a right turn is (1060 - 940) or 120 lbs, which is a ton in terms of grip. The difference in cornering loads just turns out to be twice the difference between the static weights. And then you have to consider effect on the other axle. If you can move weight around to get even left to right weights, all the better.

So you have to test which gives better lap times, because there are many, many factors involved. You can start at 50% cross-weight, and move either way and see what the effect on lap times is, and if you can't feel the difference, don't worry about corner balancing.

Corner balancing in real life is very easy if you've done it a few times. First you have to start with an extremely level set of scales. If you don't think that is important, just weigh the corners, drive the car off the scales and drive it back on to see if things are consistent, then drive it off again, and change height of just one scale by 1/4". re-weigh and see how far off all four corner weights are. I've heard of people tweaking one quarter turn on a spring perch (one quarter of 1/8" or 1/32" height change) and they don't bother leveling the scales properly.

Corner balancing is not so easy if car has all rubber bushings and lots of friction and stiction in suspension components. That is because you move the spring perch, but stiction prevents suspension motion, and you can never seem to get things quite right. With spherical bearings, things are much more consistent. I find that you have to take car off scales, drive it around and settle suspension over some bumps and then see where you are really at by driving back onto scales. I have seen cars with hundreds of pounds out of proper corner balancing done by others.

Another good test to see if your scales are level is to drive the car back-wards onto scales and see if you get the all the same corner weights, after settling suspension each time by driving around.

Once I have my car set up the way I like it, aligned and corner balanced, I like to adjust one spring perch up and down and take pictures of the scale readings to see what the effect is. Then go back to original setting and hope you get back original readings. Now you can go to the track and play with corner balancing very easily by just turning one spring perch up or down and see how lap times are affected. I don't worry about ride heights anywhere near as much as I do about corner balancing, as long as you have set up car for approximately equal corner heights for your base setup. But the worst thing you can do is set up the car to equal ride heights and not put it on scales and think your corner weights are correct. They will not be. You need scales. Beg, borrow or steal. don't forget to adjust tire pressures to hot pressures, put driver dummy weights in car, fuel, and align it. As for disconnecting sway bars, that is not important as said if you can't adjust them for free play with adjustable end links, since when the car is at rest with sway bars connected, that is when the cross-weight should be set, as otherwise you'll upset the balance when you hook them back up. Use adjustable end links if you can, and disconnect sway bars before balancing. And don't play with all perches at the same time, or you may be at it all day. Start with just one perch or perhaps two, and once you see where you are in terms of desired % wedge, you can adjust all 4 up or all 4 down to raise or lower the car. You cannot have all the right corner and cross-weights and ride heights even unless you can move real weight around, which is not usually possible. If you are driving a formula car with spherical bearings all around, then things can be perfect. Otherwise you have to live with compromises.

solo-x

And who said symmetric spring rates were the right way to go either? Each corner has a different static load, so each corner should have a different wheel rate and tire pressure dependent on it's loading....

VTECIntegra9

knowledge is power!

RR98ITR

Sometimes I push the steering wheel up with my outside hand, but other times I pull the steering wheel down with my inside hand...

Am I hearing you right Nate? Right like Riiiiiiggggghhhhht?

Scott, who has abandoned unquestioning loyalty to the ideal of symmetry...that said I continue to hold proportionality in high regard...

solo-x

I've done some testing with it, and the results were promising... In a sport where 65mph slaloms aren't uncommon, getting the car to handle the same way turning both directions is even more important.

beanbag

Since I didn't see any corner balancing programs that actually told you how much you are supposed to move the adjusters, I made my own in excel. The left side of the chart is the program telling you what you should adjust, and the right side is if you feel like you are so dam smart and can figure out yourself how much each adjuster needs to be moved. The calculations to create the right half of the chart were a PITA to make, which is why I went on a long rant initially about the utility of eigen operations and all that other crap. I'm not good at matrix transformations in my head, nor on paper.
(see pic on page 3)

beanbag

Even though I wrote this program, I still think Descartes' method of using only one adjuster to tweak the corner balance is a pretty good idea. It works well in the case of a stiff chassis with stiff spring rates, and if the car has already been previously corner balanced and leveled.

Tyson

its not your fault.

essex

Can i download it beanbag

descartesfool

You should modify it so you can input a perch adjuster height change on any corner to get the final cross-weights and corner heights.

beanbag

Not yet, I'm checking for errors and stuff

It's already like that. The gray boxes on the right are for user defined adjuster heights

descartesfool

I had not considered the stupid way!

You need to consider unsprung weight, as it does not contribute to compressing the springs.

Also consider that the ride heights as they are related to the spring lengths by the motion ratios are only valid at the tire. The tops of the spring mounts form a plane, which if tilted changes the fender heights by an amount related to how far they are away from the center of the car. Bumper moves up and down more than the jack point for any given corner.

I usually use the fender heights above the scales as it is very easy to measure for ride heights, and is on the tire axis.

beanbag

Thanks for the suggestions.

Unsprung weight doesn't come into the equations since everything is done as delta's, i.e. changes.

Your point about choosing a measurement point for the corner height is a valid one. I considered adding that in, but then you'd also have to input track, wheelbase, x and y offset, and another pair of matrices to transform from shock tower coordinates to corner measurement point coordinates. Ahhh...that's so troublesome.

Anyway, you'd only be off by a small amount if you neglected this effect, and you would know the direction in which you were off.

Just being lazy.

Thanks for adding some practical considerations about corner weighting.

descartesfool

Unsprung weight does matter. If corner weight is 850 lbs, of which unsprung is 120 lbs, then only 730 lbs actually compresses the spring. Then that determines the deltas, not the total corner weight. The wheel, tire and brake and suspension weights are not sitting on top of the spring.

beanbag

I'd have to respectfully disagree that unsprung weight matters in a corner balancing operation. Yes, it affects how much the spring is compressed, but IF you believe my original idea that corner balancing is split up into two groups of operations:
a) raising and tilting the car without affecting corner weights
b) shifting the corner weights without raising or tilting the car

Then in a), unsprung weight doesn't matter because raising all the spring perches by 1/MR will raise the car by 1".

In b), shifting the diagonal weights only has to do with spring rate and mr, not sprung weight, which compresses the spring by a fixed amount. to shift weight, you need to compress the spring by an additional amount. It does not matter how heavy the initial weight on the spring is.

I tried to visualize the difference between corner balancing a piece of plywood on top of lead wheels, vs a thick granite slab with no wheels at all, and I couldn't see where they differ. (except in the total weight transfer allowed)

In any case, I am open to corrections, so maybe you can point out where in my original calculations I should split the weight into sprung and unsprung?

superpilun

I apologize in advance if I'm chiming in with seemingly obvious information. My understanding is that the primary, if not only, goal of corner balancing is to ensure the understeer/oversteer characteristic is the same turning left or right. Equalizing left/right balance seems secondary, as you have to physically move the cg in relation to the wheels by tilting the car. You'd have to adjust the spring collars much greater than you would if adjusting cross-weight. If more weight is on the left-front and right-rear pair, the car will tend to understeer right turns, and oversteer left turns.

The cross-weight will also be moving all over the place when the car is actually moving. If 1/4" unleveled plates makes such a huge difference, what happens with the tires are all oscillating over 1/8" surface uneveness on the track?

Guess-and-check method corner balancing does not seem like giving up. If you try to calculate all adjustments, you still have to check how much error you have between actual and theoretical values after your "one-shot" adjustment. After you're done with your equations, you're going to ditch the calculator and do it like everyone else.

By the way, unsprung weight matters. An uneven table with no springs still rocks back and forth. Put some wheels on it, and the uneven cross-weight will yield the same result.

descartesfool

You are right. While the unsprung mass is what affects the total spring deflection, when dealing with only the difference in spring deflection due to perch height changes for corner balancing, you do not need to consider the unsprung vs sprung masses. I had not realized that in the way you did your calculations, the initial weight on the spring is effectively factored out and you never actually have to calculate the initial and final deflections, but just the differences. I has assumed you were calculating initial and final spring deflections and subtracting one from the other. Without seeing your spreadsheet formulas, it was not obvious. The corner weight change for your case b) is the change in unsprung mass at that corner due to diagonal changes, and then as you said you just need the spring rate and MR.

With regards to precision, a 0.125" error (typically equivalent to one turn on a spring perch) in the height of only one front scale for a car with 850 front and 1150 rear springs, MR of 0.64 F and 0.71 R and corners weighing 850, 850, 490, 490 will cause each wheel to have a change of 20 lbs. So that is 40 lbs more on one diagonal and 40 lbs less on the other, for a total cross-weight error of 80 lbs. Thus instead of the real 50% cross-weight of this example, you would think you have 870, 830, 470, 510 lb corner weights and a 48.4% cross-weight, which is a large error, and people who do cross-weighting would think they have to adjust that out, while in reality, it is an error in the pad height, and the car is perfect. The softer the springs, the less important is the error in scale or perch height. For 400 front and 250 rear springs, the cross weight would read 49.5% for a scale off by .125", which is 7 lbs per wheel, or 28 lbs total wedge. Like I said, if you want to know if your scales are level, drive the car forward and backward on the scales. You should get the same corner weights. And if you have the LF scale up by .125" and the LR rear scale down by .125", you get a 160 lb cross-weight error giving an incorrect reading of 47% cross-weight for the car with the stiff springs. I would suspect most cars are very poorly cross-weighted from what I have seen.

beanbag

I agree with everything you wrote in the second paragraph.

I'm curious where you got the values for 20 lbs and 7 lbs weight transfer. I plugged those values into the program, and in the case of the softly sprung car, got the same thing. In the case of the stiffly sprung car, there is one extra parameter, namely the chassis twisting stiffness, that affects how much weight is transferred. For infinitely stiff chassis, I got 21 lbs, and as the chassis stiffness goes down, so does the weight transfer. To get 20 lbs, I needed to use a stiffness value of 6000 lbs/in (IOW, if you push on one diagonal with 6000 lbs, and pull on the other by 6000 lbs, then the chassis will deflect by 1" at each corner).

In the examples you gave, I'd rather move only one adjuster 1/8", and accept the change in corner heights, than all four approx 1/32".

redboost10

wow, im glad i just drag race, haha. just joking. good job beanbag. looks good.

descartesfool

I used infinite stiffness for the chassis, assuming it does not twist, as I have never measured chassis stiffness. And I got 21 lbs, but just rounded down to 20 to discuss the need for precision. I think you would benefit from a calculation of cross-weight, since the "weight transferred" box only gives the weight change per corner, which is somewhat deceiving. What you care about is the total cross-weight, which is 4 times the individual corner change, and more easily shown as a %, from weights of (RF + LR)/Total. It is also useful to show to show left and rear %'s, to remind yourself of just how challenged the front tires are in a FWD car, and why you should try to move weight around. And change your heights from numbers in the 5" range to the heights at the fenders like 23" due to the fact that that is the only place your calculations are exactly correct unless you calculate the equation of the plane. It is a whole lot easier to measure at the fenders in any case to check your work, since if you are going to level things, you will level at the tires below the fenders, and you just measure up from your leveled points. The need for accuracy is further shown in your example above with the very small changes from 5" for the final chassis heights. And as you have found, it is far easier to just move one perch for fine tuning rather than play around with all 4, unless you really need to change your ride height by 0.06", which is rather hard to measure. And it also shows why it is futile to try to corner balance a car using only heights at the corners. If the actual corner height measuring points are off by only those few hundredths of an inch, you can get very large errors in cross-weight %'s, particularly on a stiffly sprung car. And there is no guarantee that your corner measuring points form a plane, and of course you have to consider chassis twist, and tire deflection, etc. You need to use scales, and you need them leveled.