Off Panhard Bar

[Gary Derian]

Hi Joe

The Alfa setup has no bind because the angled arms intersect at the joint.
When two arms are angled to form a virtual apex, they intersect only when
the body is level.  As soon as the body rolls, these two angled arms become
skew and no longer have a virtual intersection.  Now that I am thinking
about it more, if from a side view, the upper and lower arms are parallel,
binding will not occur, but now there is no anti-squat.  Not from torque,
anyway.  Any other orientation will cause bind when the body rolls.

Reversing the arms so the virtual intersection is in front of the axle does
not change the roll center location.  The roll center is always in the
vertical plane that contains the axle shafts.  A line has to be drawn
rearward from the virtual intersection, parallel to the lower control arms
until it intersects that plane.  If both upper and lower control arms form a
virtual intersection, a line is drawn through both intersections.  Where
that line intersects the vertical plane is the roll center.

Where did you post that drawing?

Gary Derian <gderian at oh.verio.com>>
>
>>There was a general discussion of rear suspension design.  My point was
>that
>>a 4 link, where two of the links are angled to provide lateral location
>will
>>bind up when the car leans in a corner.  One angled link will pull
forward,
>>the other will push rearward.  The amount is small but it is there.  With
>>rubber bushings, it works OK.  This type of 4 link is popular on many US
>>cars, notably GM intermediate and full size and Ford Fox chassis Mustangs
>>and their siblings.
>>
>
>Wouldn't the upper bars pull/push the same amount if spaced exactly
>>>>the same distance apart from the center?  I need to draw something to
>>>>understand this better.
>
>
>>No.  The binding is not from the change in angle between the rod and the
>>pivot axis.  Its because one link tries to rotate the axle pinion up and
>the
>>other tries to rotate the axle pinion down.  This occurs only when the
body
>>rolls.
>
>>
>Hello Gary,
>
>I was going to send this off list, but in view of the other interest I
>thought I'd share.
>
>Earlier in this thread Greg wrote:
>"Or go scrounging in a boneyard that has some old Alfas. They used an upper
>triangle, with two pivot points on the chassis, and a BALL joint attached
>just to the left side of the pumpkin, plus two lower trailing arms.
>Particularly if you replaced the chassis pivots of the triangle (rubber
>stock) with bronze on steel, also if you replaced the lower rod ends with
>real rod ends, they were as NICE a live rear axle setup as you could want."
>
>I believe that this worked, and worked well. The ball joint located the
roll
>center. As the body rolled in a turn, the worst action I can see is that
the
>axle rotated slightly. Most certainly there was not binding, even with rod
>ends and bronze bushes, in spite of the off center location. Some might
even
>suggest that the roll center would be better placed to the right, since
this
>is the side which unloads under acceleration. I think this demonstrates how
>much a design can deviate from a theoretical ideal, or how much you
>sacrifice an ideal in one area chasing the overall performance of a system.
>
>When I suggested two angled arms, I saw a virtual triangle with the roll
>center located where lines projected from the arms intersected the
>centerline of the axle. In this case we have two rod ends in shear rather
>than a single ball joint with bending stress. As long as I picture this
>virtual triangle, I can so no reason for bind. The virtual triangle acts
the
>same as the real one.
>
>Also, I don't see any reason not to turn this virtual triangle around so
the
>base is connected to the axle. The virtual apex still determines the roll
>center, the body will roll about this point, and the axle will still rotate
>slightly, but still no bind. Now the roll center has a more constant
>relationship to the center of gravity, and since this relationship
>determines the amount of body roll, perhaps the car will behave more
>consistently.
>
>As I have said before, ASCII is not an easy medium. It's hard to describe a
>concept in a concise manner and show a rapier like wit at the same time.
:-)
>
>Maybe we are not seeing the same picture. I know now that I wasn't when I
>visualized
>your description of the torque arm setup from the Vega. To that end I have
>posted an AutoCad Whip file which contains a drawing of a Satchell Link
>suspension to incomming, Satchell Link.dwf. If you need a viewer it is
>available free at
>http://www.autodesk.com/products/whip/index/htm . If you prefer I can
>convert it to .dxf .wmf .eps or .bmp.
>This suspension has no
>relationship to any car, it is just a "reasonable facsimile" of an
>illustration from Chassis Engineering.
>
>Someone said that the harder you defend a position the more likely you are
>to be wrong. Prolly a corollary to Murphy's Law. So, like Bruce's ECM, I'm
>going to "bench" this thing and build a half-scale model and see what
>happens as soon as I have the chance. Maybe both Andy and I will learn
>something. Maybe that lesson will be to suggest  Shannen switch to Decaf
>:-).
>
>I'll let you know how it works.
>
>Regards,
>
>Joe
>
>
>
>
>
>
>
>
>
>
>
>
>
>