Understanding Design of Motion Graphs Like Cams

[SchmidtMotorWorks]

Design of motion is a fascinating subject,

I wrote a program to help people learn how motion can be designed with polynomials.
Polynomials were first used in cam design in the early 1950s and are still used by some people today.
Polynomials can be used to design any kind of motion by connecting short polynomial curves together.
There is nothing wrong with the capability of polynomials, but there are easier ways now.
Still, you really need to understand the polynomials before the easier spline-based methods will make sense.
The key to learning to design motion with polynomials is to learn where to break up a desired motion into pieces and make short polynomial segments to fit.

The attached application was written in Visual Basic, it will only run on Windows and will require that your computer has the .net framework installed. You can get that from Microsoft.

A brief explanation of what you can see in the app:

The orange curve is measured from an existing cam.
The table below is a series rows and columns:
The columns are for the individual segments of polynomials.
The columns contain cells to adjust the duration, velocity, acceleration at the END, of each span.
The next span begins with the values of the end of the previous span so that the curves are continuous.
The range fields just change the display scale height of the curves in the graph.
PQRST inputs change the shape of the curve, the values should begin with 5 and increase form P to T.

Some things are easiest to learn by trying.
Download and install the app and start adjusting fields to see if you can make the polynomials fit the shape of the orange curve.
If you find it to be a difficult challenge, keep in mind that the fist people to do this did the computations on paper or a slide rule.
UD Harold did it on an early computer, it might have used punch cards.
He didn't even have graphics to see the shape of the curves, he only had text files to look at.

I don't have time today for much more explanation, so try to make some progress by trying and learning through experimentation, I can explain some things or answer questions latter.

If this is new to you, once you get this, cams will have a different meaning to you.

motion_design.jpg

smw_release_motion_design.zip

[F-BIRD'88]

Interesting

[SchmidtMotorWorks]

Interesting

The deeper you understand it the more interesting it gets too.

I hope it gives people an appreciation of the work that cam designers go through.

Who will be the first to post an image of a design that matches the orange curve?

[Sparksalot]

That's an interesting challenge. I learned to do it differently as a young ME but it was way over 50 years ago, then I left the automotive field to join NASA. Met my first computer in 1963 and that took the drudgery out of such work if one could write the code to do it. All the good early code was simply putting the math we did laboriously into means for the machine to do it. Many heavy IBM card boxes at a time.

I still scream in my sleep with "syntax error" failures.

Hope some young folks take up your challenge, it will definitely be a learning experience for them.

[Rick Finsta]

Punch cards? You mean the big floppy disks?

Just kidding. My grandfather was the first grocer in the state of IL to have a computerized scanning system. My ol' man grew up punching cards to program pricing. My first computer was an Apple IIe, followed closely by an IBM 8086.

I'll give it a try!

[Kevin Johnson]

Staedtler 971 60-24.gif

https://www.google.com/?gws_rd=ssl#q=st ... +971+60-24

Highest quality German instrument made in China.

[RevTheory]

Winner, winner, chicken dinner!

Joking aside, I've been cursing my own name for over a decade now for choosing lame electives instead of higher math classes in high school just so I could check out of class and go build engines in the auto shop. Stupid, stupid, stupid!

After high school, I turned down an auto/diesel scholarship to go to work in the underground gold mine making more money than an 18 year old knew how to spend, even with a PAW catalog, lol, and never went back. Now it's caught up with me. Again, stupid, stupid, stupid!

[pdq67]

IBM 360A1... Or whatever UMC's big engineering computer was?

Anyway, stacks of cards, one card per line!

I was taught to use SIMULQ and FPLOT to calculate/graph N+1 data points into an Nth order polynomial back in the early '70's in ME School.

At least I think I was?? Damn, haven't thought about this in quite a while...

And I can remember seeing a Marchant mechanical calculator in my Nuke class and shortly thereafter, the H/P-35 came out. Talk about H/P obsoleting the "Slide Rule"...

Obsoleted way faster than "buggy whips"...

pdq67

[groberts101]

Kinda interesting. Not to muddy this thread up with even more complexity but the mention of spline-based methods reminded me of this saved link. Tons of history, info, and even some downloads mixed in there too.

http://www.tildentechnologies.com/Cams/CamDesign.html

http://www.tildentechnologies.com/Cams/CamSprings.html

http://www.tildentechnologies.com/Numerics/index.html

[Rick Finsta]

So far I have found that I get the lift curve first, then set the jerk to a constant, and then start smoothing the velocity curve using the acceleration and the lines are coming together pretty nicely.

[SchmidtMotorWorks]

So far I have found that I get the lift curve first, then set the jerk to a constant, and then start smoothing the velocity curve using the acceleration and the lines are coming together pretty nicely.

You have the right idea.

[SchmidtMotorWorks]

IBM 360A1... Or whatever UMC's big engineering computer was?

Anyway, stacks of cards, one card per line!

I was taught to use SIMULQ and FPLOT to calculate/graph N+1 data points into an Nth order polynomial back in the early '70's in ME School.

At least I think I was?? Damn, haven't thought about this in quite a while...

And I can remember seeing a Marchant mechanical calculator in my Nuke class and shortly thereafter, the H/P-35 came out. Talk about H/P obsoleting the "Slide Rule"...

Obsoleted way faster than "buggy whips"...

pdq67

You got it, a lot of people don't realize what it meant to work with punch cards, I never did it myself but saw my 10 year older brother do it. I doubt that I would of had the patience for it. I like computers because it allows us to try things quickly and find mistakes and fix them quickly.

I suppose punch cards seemed quick compared to pencil and paper.

As I recall, those first HP calculators were very expensive, in the range of a good used car.

[Runit]

Batch processed punch card Fortran on a CDC3600, then IBM 1401, 7094,and 360-50. A LONG time ago. The 360 was used by the Cyclotron across the street too and that sometimes screwed with your program run. The batch printer would stop printing output and open mysteriously.

[SchmidtMotorWorks]

here is an image of one of UD Harolds Acceleration curves.
You can use this as a hint for what the shape of acceleration curve is.

[Runit]

It was so long ago that there was still a guy in the computer center that had operated the IBM machine on display in the lobby. It was one of the very first ones, maybe the 3rd or 4th. Lots of vacuum tubes and jumpers and switches. The guy said it rang a bell when it was finished with a program!