Jerk is it Important?

Interestingly, most designers agree and some are quite emphatic about the the importance of jerk (see Norton for example ). The concerns about jerk came to the fore at an SAE technical conference in 1953.

The following is a quotation from Keith Duckworth about the development of the A2 cam profile for the Ford Kent engine in 1959 (see Cosworth the Search for Power):

My problem was that I had read the books on cam design and I had believed them. That was fatal. Even so, these people seemed to be fairly knowledgeable about the various types of profiles and acceleration diagrams. ..... I reasoned that I didn't know what jerk was all about, and so I was fairly sure that the valve gear didn't know it either. I thought that all the arguments about jerk were grossly over-rated.

The result was that Keith and Ben Rood revised their views on cam design and went for constant acceleration curves. Although not stated, these were presumably accompanied by jumps (infinite jerk) between the constant accelerations, much like triple curve designs or harmonic profiles.

We agree that the importance of jerk is grossly overstated. Valve train dynamics theory shows that acceleration is the highest derivative that needs to be continuous, i.e. jerk remains finite, but can have jump discontinuities. Cubic splines are completely adequate. Most designers are using profiles that are much smoother than necessary. For example, Norton's favorite spline uses quintic functions which have continuous snap (derivative of jerk). The use of overly smooth functions causes a slower opening cam or requires higher acceleration rates to compensate.

We have seen only one source (Hollingsworth and Hodges) that have actually recommended limiting values. They recommend that jerk not exceed 0.0002 to 0.0004 in/deg³ for pushrod engines, whereas direct attack valve trains (OHC, flathead) can use values of 0.0004 to 0.0008 in/deg³. These values are quite large. When used with cubic splines, the profiles are very similar to those with infinite jerk (see example). We have profiled many OHV "performance" cams and typically find maximum jerk of 0.00003 to 0.00004, which is 5 to 10 times smaller. Overly smooth cams are poor performers.

The best way to establish acceptable values of jerk is to perform a valve train dynamics analysis for the cam in question.