Laymans Jerk question

Hi all,

I'd love to know what Jerk means in a Rockwell Kinetix motion system.

I have done some light googling and I hope what I have summed it up to be is right???
The jerk value determines the shape of the Velocity S curve in the I, III, V and VII segments in a point to point move (stop go stop)? I.e less jerk = more S, more jerk = less S and further... max jerk = trapeziod profile?
Is what I wrote correct?

Cheers all, Nathan

Nearly, min Jerk = trapezoid profile.
So more Jerk = More S curve
You can have a good view on this in Motion Analyser.

Try thinking about it this way.

Velocity tells you how fast position is changing.
Acceleration tells you how fast velocity is changing.
Jerk tells you how fast acceleration is changing.

Steve has it right. Nightline has got it backwards.
This file was generated using wxMaxima which is free.
http://deltamotion.com/peter/Maxima/Seg1234567.html
If you download the wxMaxima file:
http://deltamotion.com/peter/Maxima/Seg1234567.wxm
You can change the jerk values and see the affect on the motion profile.

And here I thought a Rockwell Automation jerk was the service tech guy on the phone telling you your PLC can't possibly be doing what it is, as it was not designed for that.

Peter Nachtwey said:

Steve has it right. Nightline has got it backwards.
This file was generated using wxMaxima which is free.
http://deltamotion.com/peter/Maxima/Seg1234567.html
If you download the wxMaxima file:
http://deltamotion.com/peter/Maxima/Seg1234567.wxm
You can change the jerk values and see the affect on the motion profile.

Click to expand...

Peter, If I'm wrong on this, Motion Analyser from Allen Bradley is also.
If I type in a jerk of 0% I got a trapezoid motion, if I type in 100% for jerk I got the S-curve acceleration and deceleration....

Thanks Peter, I had a play with wxMaxima and your file. It appears the Jerk influence can be best observed on the acceleration plot. Just for fun I mucked around with alot of the "user" parameters and I generated some very ugly looking plots!

nightline said:

Peter, If I'm wrong on this, Motion Analyser from Allen Bradley is also.
If I type in a jerk of 0% I got a trapezoid motion, if I type in 100% for jerk I got the S-curve acceleration and deceleration....

Click to expand...

Jerk does not have units of %. Jerk has units of distance/time^3. Rockwell specifies jerk in a non standard way. Rockwell specifies jerk in % of the ramp time so a jerk % of 25% would mean that 25% of the total ramp time the jerk is increasing to the maximum acceleration. The jerk is then 0 while at the maximum acceleration which lasts 50% of the ramp time. The last 25% of the ramp time the jerk is negative to bring the acceleration rate back to 0 for the constant velocity phase. Therefore a 0% jerk time means the acceleration instantly changes from 0 to maximum acceleration so the ramp is linear or second order.

As I said above, jerk should be expressed in distance/time^3 but most people have no feel or clue what is a good value for the jerk and can't think it through. Rockwell's non standard solution keep people from really screwing things up but it is obvious it doesn't help with understanding what is really happening.

Using a jerk time percent of the acceleration time will mean the maximum jerk will change if the the maximum velocity or acceleration changes.


In many applications the peak acceleration is in the range of 0.25g to 0.5g. A good way to estimate a jerk value is to think the acceleration should change from 0 to maximum acceleration in 0.01 to 0.1 seconds and sometime longer. Divide the maximum acceleration by the time you think it should take to get to the maximum acceleration and you get a good value for the jerk. For instance if you want to accelerate to 0.25g in 0.1s then the jerk should be 2.5g/s. If a g is about 10m/s^2 then the jerk should be 25m/s^3.

This makes sense Peter.
Thanks for a very clear explanation!

In my mind, I tend to think of jerk as (m/s^2)/s or sometimes g/s. The numbers are the same as m/s^3 but it seems to be easier for me to get a "feel" for the quantity when I think of it that way.

Likewise, I have worked on some rotary systems that program in rpm and for angular acceleration on those, the unit that seems to give me the best feel for the motion is rpm/s. It is sort of a kludge mixing minutes and seconds but I can picture it in my mind much better than rev/s^2 or radians/s^2.