chassis stiffness / backlash

[Frank]

I've got to thank Kevin Hayward for putting me on track with this!

Have I got it straight?

We "tune"� the suspension by varying the F/R roll resistance, in both steady state (ARB's, roll center heights, and springs) and transience (shocks and roll center heights).

The function of the chassis is to allow us to react the roll moment with roll resistance ratio (RRR) not equal to the roll moment distribution (RMD). The RMD is due to weight split, and F/R tracks. If your chassis is floppy, the RRR will equal the RMD, and the car is impossible to tune.

But unless there's something going wrong, the actual torsion transmitted through the chassis is small.

I'm thinking the largest chassis torsion you're likely to see in a FSAE car is 50Nm?

Therefore we are not so much worried about chassis torsional stiffness alone.

We are worried about chassis torsional stiffness at low torque values, and the slop / take up / backlash? (What is the academic terminology for this? Dead band?)

I'm asking you guys, how do you measure the backlash accurately / repetitively?

Is it possible to measure it accurately / repetitively?

Also, I'm interested, what measures you go to reduce this backlash?

I've noticed spacing your wheel bearings too closely causes quite a bit, I guess flogged out spherical bearings and rods ends will too. Failing to use press fits on deep groove bearing housings gives you more bearing internal clearance (or fails to reduce the bearing internal clearance).

Perhaps try to keep pushrods / pull rods at a decent angle, and using larger rocker levers to increase the actual movement of force multiplication mechanisms?

We're trying to "pre-stress"� a bonded floor pan ATM.

I'm guessing double shear might help a bit. (please, I'm not trying to flare the old single shear / double shear debate).

Frank
Uni QLD (yes, still)

[Frank]

I've got to thank Kevin Hayward for putting me on track with this!

Have I got it straight?

We "tune"� the suspension by varying the F/R roll resistance, in both steady state (ARB's, roll center heights, and springs) and transience (shocks and roll center heights).

The function of the chassis is to allow us to react the roll moment with roll resistance ratio (RRR) not equal to the roll moment distribution (RMD). The RMD is due to weight split, and F/R tracks. If your chassis is floppy, the RRR will equal the RMD, and the car is impossible to tune.

But unless there's something going wrong, the actual torsion transmitted through the chassis is small.

I'm thinking the largest chassis torsion you're likely to see in a FSAE car is 50Nm?

Therefore we are not so much worried about chassis torsional stiffness alone.

We are worried about chassis torsional stiffness at low torque values, and the slop / take up / backlash? (What is the academic terminology for this? Dead band?)

I'm asking you guys, how do you measure the backlash accurately / repetitively?

Is it possible to measure it accurately / repetitively?

Also, I'm interested, what measures you go to reduce this backlash?

I've noticed spacing your wheel bearings too closely causes quite a bit, I guess flogged out spherical bearings and rods ends will too. Failing to use press fits on deep groove bearing housings gives you more bearing internal clearance (or fails to reduce the bearing internal clearance).

Perhaps try to keep pushrods / pull rods at a decent angle, and using larger rocker levers to increase the actual movement of force multiplication mechanisms?

We're trying to "pre-stress"� a bonded floor pan ATM.

I'm guessing double shear might help a bit. (please, I'm not trying to flare the old single shear / double shear debate).

Frank
Uni QLD (yes, still)

[DJHache]

You don't have to guess what the amount of torsion in your chassis is Frank, you can calculate it out pretty well.

If you know how high your cg is above your mean roll centre (the roll centre height at the cg location)and you know what lateral accel you're taking, then you know how much torsion you're putting on the frame/suspension system. Since it's a series spring you know how much torsion is going through the frame.

If you have a particularly sloppy frame, swiss cheese is the preferred image, then your antiroll stiffness will go to 0.

Now, imagine the front of the car as a series front suspension + front frame, and the rear as rear suspension + rear frame. You are twisting the chassis at the cg and it is being reacted by the front series combo and in parallel by the rear series combo. If your frame is infinitely stiff then a 10% increase in front antiroll distribution will equate to a 10% front lateral load transfer increase. On the other hand, the swiss cheese frame will have front and rear stiffness values of 0. Since 10% of 0 is 0, any change you make on the car as far as front/rear ARB for example will have no effect on load transfer.

Now the question is what happens when you put your cg height at your mean roll center height where suspension stiffness is theoretically infinite? infinity times 0 is sticky business. In this case your throw the baby out with the water because there is no adjustability either.

In both these cases you're pushing lateral load transfer towards static F/R weight distribution.

You care about your chassis stiffness in relation to your suspension antiroll stiffness.

[jack]

yeah i allways wondered about chassis slop/backlash too. it seems that in a transient state, it wouldnt even matter what your tosional stiffness was, if you had a "sloppy" frame.

havent thought of a good way to measure it either...have the judges even brought it up?

[Denny Trimble]

Well, we could modify our chassis torsion rig so it's balanced, and so you can put weights on either side of the torque arm. Then you could load it one direction, remove the load, load it the other direction, and see what it looks like when you pass through the zero point.

[Kevin Hayward]

In the last couple of years we changed our testing technique at UWA. We only loaded a torsional load big enough that would lift a front wheel (less weight on the front) with dummy shocks. This alowed us to concentrate on the loading conditions the car would actually see.

Unfortunately the small amounts of twist increase the measurement error. It is this measurement error that is the biggest problem with doing torsion tests at low amounts of twist. The test is done without a bar connecting the two sides of the twisting end. This is typical with some commercial chassis twisters but adds artificial stiffness.

All in all the tests do produce reaptable and surprising results. It certainly prompted us to look at other areas apart from the chassis itself to increase torsional rigidity from one end to the next. The reduction of rigidity around the zero point is quite significant.

Moral is that even if you have tested and found the desired 10 times the roll stiffness difference (as per a couple of papers .. of which I don't completely trust) it is likely that you do not have it where you need it.

Now if someone could post some info on testing chassis damping I would love to have a read.

Cheers,

Kev

Optimum G
(ex-UWA Motorsport)