Camshaft Design- New 1

[EngJW]

I would like to start this thread over again. The last one had drifted too far off topic. My proposal for the topics would be the following:

1. Cam profile design and ramp design
2. Cam and lifter materials
3. Camshaft manufacturing
4. Valve spring design, materials, manufacturing
5. Valve train dynamics

I don't think we should get into applications (what cam should I put in my car?) or valves and ports. That should be the subject of another thread.

Most of the above subject matter is in the hands of a few specialty companies and consultants, but I doubt if a lot of it is proprietary. Engineers involved in engine design and testing could benefit from at least a working knowledge in these areas. If a problem comes up, it would be good to be able to take a first shot at it before turning to consultants, especially if your company does not have deep pockets.

Thanks,
John Woodward

 

[patprimmer]

Thanks John for bringing some clarity and sanity back to this subject. I shall follow it with interest. I hope it is not the subject of multiple hijacks again

Regards
pat pprimmer@acay.com.au
eng-tips, by professional engineers for professional engineers
Please see FAQ731-376 for tips on how to make the best use of Eng-Tips Fora.

 

[GraviMan]

I missed the last thread, but here are my initial thoughts:

1. Cam profile design and ramp design

The ideal is to minimise return spring stiffness (friction), without allowing follower to cam contact loss. Most seem to offer good timing characteristics with elliptical shaped risers.

2. Cam and lifter materials

Tribology plays as much a part in this as any. If the oil film is thick enough, there will be no metal to metal contact.

3. Camshaft manufacturing

This should be interesting. I have seen hydroformed cams, which struck me as a good way to save weight. I imagine only limited machining was necessary.

4. Valve spring design, materials, manufacturing

To some extent, this will depend on point 1.

5. Valve train dynamics

Are you interested in vibration or fluid dynamics? I imagine the ideal motion to be a cosine pulse, with a dwell between events. The cam profile ideally reproduces this.The tricky part is optimising the timing.

Are you considering VVT mechanisms, desmodromic, and actuated cams?

Mart

 

[willeng]

John, this is a very broad subject & basically the design, the ramp, materials, manufacturing, spring design, materials & manufacturing again & valve train dynamics are application based. From a daily driver to an all out strip engine & everything in between.

Each specific configuration of engine will require a certain design & is dependant on the bore, stroke, rod ratio & the flow characteristics of the induction & the exhaust system.

The valve timing events would be a good place to start & then again is application based.

Taking into consideration that an engine to produce good power needs pressure, flow & rpm, this would be a good talking point to start things off, what timing events are the most important & why.

 

[EngJW]

I think the application would determine the intake and exhaust duration, the lobe separation angle, and the amount of advance or retard. Or in other words, the valve opening and closing points. Perhaps lift is influenced by the port design and the amount of room available in the chamber for lifting the valve. Unless I have it all backwards.

My interest has been in how to translate the required lift and duration into a profile that will not cause dynamics problems. At my place of work we have been having valve and seat wear problems and we have been experimenting with different profiles and spring rates. More on that another time.

Willeng- there was a good SAE paper one time that examined all the points. The intake closing and exhaust opening points were said to be the most important. More recently I have seen mention of the overlap period being used to control emissions. More overlap at low speeds acts like built-in egr. I always thought overlap was used for high speeds.

Till next time,
John Woodward

 

[PSlem]

I had what I thought was a great idea several years ago, a dwell type cam drive. If the cam had an arm with a pivot at the end and attached to that another arm with a roller at one end 90* to the cam arm and a sector gear at the other centered on the pivot, and the timing sprocket, which is free to spin on the cam, had a sector gear that meshed with the one on the arm, then the cam would be driven by the timing sprocket through the arm with the roller and sector gear. However if the roller is allowed to pivot out or in, the timing would be advanced or retarded. By mounting this roller between two non-rotating concave plates that transition from elliptical near the center to circular at the edges, we introduce two dwell periods per rotation, and by varying the distance, can control the amount of dwell. Thus we can shorten duration at low speeds and return it to normal at higher speeds where the roller is on the constant radius. The second event would occur while the valve is shut so would have no effect. This is only suited for DOHC 1,2, or 4 cylinder engines. Valvetrain stability is the same at high speeds and the only down side is additional load on the timing chain from acceleration.

Engineering is the art of not constructing...of doing that well with one dollar what any bungler can do well with two after a fashion.

 

[mvf4sp]

Hello everybody

Well the design of a camshaft is a broad topic. But there are some essentials. First you have to have in mind in which engine you will fit the camshaft, if it requires torque in the low range or in the large one. With this in mind the experts use their experience in the design of the features of the cam, lobe separation, overlap, lift etc. In cam design there's a rule that says that the derivate of the acceleration must be finit, it doesn't matter if it's not continous. Another rule I've read is about the lobe separation, in an engine bigger tha 350 ci, add 5 degrees in lobe separation for every 25 ci up in displacement, in engines with less than 350 ci, take out 5 degrees in lobe separation for every 25 ci

For an engine cam, it was demonstrated that the best function for the displacement going up or down is a 3-4-5-6 polynomial function because it offers the minimal values of acceleration and in consecuence force.

The materials most commonly used are hard ones such as cast iron, gray cast iron or steels with mid content of charcoal.

The best way for machining cams is using CNC as far as I know.

There is more info in Norton's book "Design of Machinery" about cam design and cam dynamics

 

[willeng]

John:

In your last post you stated that,
At my place of work we have been having valve and seat wear problems and we have been experimenting with different profiles and spring rates.

What test engine is this in, pushrod, OHC etc?

 

[EngJW]

Willeng-

We have an industrial air-cooled V4, ohv. This engine has been in production for years. We started with cast iron seats and Stellite valves and wore out the seats, so we changed to nickel based seats with the Stellite valves, and now we wear out the valves. The weird thing is that we have this problem on only one cylinder, which suggests a heat problem, but other cylinders seem to run hotter.

The problem was accidently discovered while checking the valve clearance after several hours of hard running. There was none left. The standard procedure was to turn the engine over until one valve was open and the other closed, then set the lash. However, it was found that after rotating the engine around again to tdc compression, that the lash decreased by several thousandths. We suspected base circle runout or valve train deflection, but could not find any.

We are going to send the engine out to a local cam company and have them run it on their Spintron machine next.

John Woodward

 

[ghuang429]

Hi, sorry for a newbie question.

I have some questions regarding the manufacture of over-head cams.

Specifically does anyone have any recommendations as to:

1. Where to test the camshaft cores (I guess wear resistance, overall material quality, etc.)?

2. Where to grind the camshafts (United states)

3. And where to test the accuracy of the grinds?

4. Is there anything else I'm missing?

Thanks in advance!

 

[EngJW]

qhuang-

As far as material, grinding, and measurement, I don't think there is a difference between ohv and ohc cams so any cam grinder or manufacturer should be able to do them. Design is a different matter, however. The mechanical position of the lobes will be different from an ohv, depending on such things as type of drive, angle between valves, number of cams (sohc or dohc). The concepts of advance, retard, and lobe separation angle should apply equally. Another factor is the amount of lift that can be squeezed into the profile without getting excessive accelerations, since there may not be any rocker arms to multiply the lift.

John Woodward

 

[hughnew]

I am new at this so you will have to excuse any silly mistakes.
I was looking at an earlier thread and there was mention of a cam design program in VB.
I wrote a program in fortran about 20 years ago and have been trying to bring it to life by rewriting in qbasic. Am having some success but was wondering if an alternative program was available for the design of polynomial cams?.
The original program was used to design some cams, mainly for 4-valve engines including the weslake speedway engine.
It is many years since I was involved in I.C. engines so am a bit rusty.
Any help would be much appresiated.
Hughnew

hugh@newlyn.com

 

[EngJW]

We had Fortran in college back in the 1970s. I would be surprised if anyone uses it now. Next came Qbasic. It used to be supplied with DOS and I don't know whatever happened to it. It was much easier to learn than Fortran and did the same thing, but its input and output formats are primitive by todays standards. VB adds all the Windows features so that you can have some nice input and output screens, and you can also make graphs. Unfortunately, I couldn't begin to tell anyone how to learn VB. The books I looked at were useless. I mostly found examples of things similar to what I wanted to do and then did a lot of copying and pasting. Lots of trial and error. My programs are probably not efficient; as long as they work and don't tax the computer I don't care.

The polynomial equations came from a number of older SAE papers. The methods of solving them came from a library book about matrix algebra (I don't know, must have cut class the day that was taught).

There are some programs that you can buy. Andrews is one of them. There are some posts in this thread about others.

These comments are just general. If you start on this project, post again and perhaps I can dig up some of my old information.

 

[hughnew]

Thanks John, I will press on with this and may have a look at VB.
Will come back when its running, shouldn't be long I hope.
Hughnew

 

[GregLocock]

Having just paid real money for a copy of Fortran I can assure you it is alive and well. There are an awful lot of programs being used that are still written in Fortran, and a lot more that are written in C, mechanically translated into Fortran, and then compiled.

I agree VB is a horrible mess. For day to day work I now tend to use Scilab (

http://www.scilab.org)

which is an interpreted mathematical toolbox, rather like matlab, but free.

If you like Pascal you might like it's son, Oberon, also by Werth. It is a bit more graphically orientated than Pascal, but it still has a text based interface.

But qbasic is fine for typical engineering projects.



Cheers

Greg Locock

 

[magnograil]

John - A book here at work on Mechanisms and cam design shows building a cam profile from polynomials, probably similar to the SAE papers. Either you work backward from the assumed accelerations to get the displacement, itterating to obtain the duration, or forward from the assumed displacement and solve for the accelerations.
Seems the best approach is to write out the equations for displacement, velocity and acceleration of everything from the geometry, input limits for the loads, lift, duration, spring, RPM, et cetera and start with a simple cosine shaped control curve. Then move the points on the curve until the opening limits are met, the stopping accelleration (reitterating the opening if the stopping requirements cannot be met), then work the closing curve the same way. The control curve would be interpreted with a spline for all the intermediate points required. A motion curve is calculated based upon the geometry (radius tappets, finger follower, inverted bucket). The control curve can be interpreted for whatever accuracy is required by the cam grinder.

 

[EngJW]

Just got a copy of Norton's book on cam design. It was mentioned in another post here recently. I am into chapter 3 now and it looks like a pretty decent book, more readable than most. However, with 610 pages and the rate I am going, this is going to take a long time.

The emphasis in this book is that you build the profile from continuous jerk and acceleration functions in order to control the motion and stresses. The lift curve is the result, not the start.

 

[magnograil]

Working from that direction you constrain the jerk and acceleration at the cam but not the rest of the system. You also do not know what your lift and duration is until you finished piecing the curve together. With a cam and unverted bucket, the valve stress and spring surge can be had directly. If a tappet, pushrod, rocker system then you have to check the stress and surge after you finished the profile.
Either way it takes itteration. Just seems easier to start with the constants you know and work toward the maximum accelerations constrained by the limits. Also, that way you can change parameters (spring pack, masses, geometry) and come up with a new profile quickly.
I have not seen Norton's book but most were written before fast computers were available.

 

[EngJW]

The Norton book is from 2002. The only other one I am really familiar with is Rothbart. Most of the machine design texts just rehash the same old stuff. Turkish wrote an interesting book about pre-1950 automotive cam design. Although there were no computers then, he tried to bridge the gap between cam layout on the drawing board and a mathematical basis where velocity and acceleration could be determined. Some of those old 3-arc and other profiles are still used.

 

[magnograil]

This book is "Mechanisms and Dynamics of Machinery" by Mabie and Reinholtz, Virginia Polytechnic U. To avoid infinite jerk they use a system of cam design developed by Kloomok and Muffley in '55 that uses three cyclic functions: cycloid, harmonic and eighth power polynomial. The curves are pieced together and accelerations matched at the endpoints.
They also suggest constructing a straight line segment jerk curve and deriving acceleration, velocity and displacement from that.
From this you still have to backtrack to find the cam profile since the motion descibed is that of the follower and not the cam surface. The minimum cam radius and follower length have to be determined to find a useable cam profile without cusps or undercuts.