01/07/12 The recent addition of a CNC mill to our shop has greatly improved our ability to do custom work, so I thought I'd illustrate what is involved in a typical project. In this case it's a set of bellcranks.

The design process starts with fundamental geometry analysis based on dimensions and other info supplied by the customer.

The goal is to make sure that throughout the full desired suspension travel the bellcranks don't overcenter and that the linkage is either linear or progressive, depending on design goals.

One common mistake on roadgoing cars is to undersize the bellcrank, resulting in way too much angularity for street-oriented wheel travel and leading to binding up, overcentering and other issues. Also, on a streetable car the bellcranks should be oriented in a vertical plane to avoid large off-axis forces due to the changing angle of the pushrod (race cars have a lot more options because suspension travel is a lot smaller). A regressive linkage is a bad thing that can lead to excessive roll, unpredictable behavior with weight transfer and outright suspension failure. Linear linkage is the most predictable and easiest to tune. It is recommended if antiroll ('sway') bars are planned for tuning the roll stiffness balance at either or both ends.

Progressive linkage requires a fairly indepth understanding of the car's dynamics but when done right it can have many benefits. It acts essentially like an antiroll bar by increasing stiffness under load and decreasing it in unloaded conditions. It does so not only in roll but also in pitch, potentially allowing elimination not only of a/r bars but also antidive/antisquat geometry and therefore allowing the shocks/springs to better do their job. Unlike when using an a/r bar, damping remains matched to spring rate and roll stiffness at each end is proportionally increased/decreased with weight transfer. The downside is that it can be a pretty complicated system to design and to tune so it's not automatically the best choice.

In this particular case I designed a linear linkage with 1:1 motion ratio for simple, predictable behavior. The next step is the design of the bellcrank itself based on the geometry.

A quick FEA check confirms stresses are in the ballpark (I usually use 4g load case). To get fully accurate results that can be used for things like weight optimisation is a lot of work so for 'reality check' simulations I use constraints that result in higher loads than the part would actually see. If everything has good margins in this overloaded simulation I typically call it good. If excessive stresses are found then either a more accurate simulation or a more conservative design (or both) is needed. In this case the overloaded safety factor was over 1.5.

After the basic design is done the details like bearings and hardware are added in and checked for fit/clearances. If a full CAD model of the car were available this would be the point at which the assembly is plugged into the overall design and checked again.

Now the parts are ready to machine. Even having 3D CAD this isn't as straightforward a process as one might think. There are lots of different ways to accomplish the task which actually makes it harder - how do you choose the 'best'? The exact machining steps chosen affect many things. Even the size of raw material can change depending on what the plan is. Once the plan is set and the machining operations are defined, the specific tools and fixtures are chosen and a program is created for each step. There is a lot of art to this and I was fortunate to get some training from an experienced machinist to get me started. The process can be frustrating at first but gets to be quite fun after getting some experience - thinking in 3D and visualizing objects in multiple coordinate systems is something I do well and actually enjoy.

In this case I ended up with four separate operations. First, a clean flat face is machined on the raw aluminum blocks. This will be used as a precise reference for subsequent steps.

Next, the part is stood up on end and the 'pockets' and clevis openings are machined. In order to precisely align these features with everything else, a reference edge is also machined that will be used in the next setup. You can see it on the left where some extra material was taken off.

Once all the parts are done in op2, the next step is machining the first side. This can still be done in a vise using the reference edge for alignment.

Now comes the tricky part. Since we no longer have two parallel straight faces to clamp in a vise, a fixture has to be made to properly hold and locate the part. In this case I made a 'soft jaw', basically a piece of aluminum that is bolted to the vise with a pocket machined into it to match the contour of the part. Notice that the tool's part number is machined into it in case we ever need to use it again.

Now the final operation can be performed...

The resulting parts are then deburred, anodized, assembled with all the bearings and hardware, and then shipped off to the customer.

While all this requires quite a bit of effort, having all the resources (and experience!) inhouse allows us to be quite efficient and therefore cost-effective at doing this kind of work. In reality this design is drawing on all the testing and experimentation I've done to date which is quite a bit. By using known and tested parts such as bearings, inserts and hardware we plan on making semi-custom projects like this one an ongoing part of the busines. These can range from individual parts to complete vehicles designed from the ground up and fully documented for production, or anything in-between. Contact me by email if I can be of help.