E: Engine Camshaft

timetraveller

Well Known and Active Forum User
VOC Member
The reason that I find this problematic is that Grey One thinks in cam profiles whereas I always think in terms of valve lift. This is clearly exemplifies by Grey One's graph in #49 above. Look at the overlap. There is almost no overlap and the lift in the centre of what there is is negligible. The lift of the valves with any Mk2 in the centre of the overlap is likely to be about 180 thou with about 105 degrees of overlap!
I agree with Grey One that when designing a cam one needs lots of decimal places. Looking back through my notes I see that I worked to four or five decimal places but my criteria were that all functions right down to the jerk and jounce were smoothly varying. I will show a graph below to illustrate that, but getting the cams manufactured to four decimal places was more difficult than I expected and, in fact, was not achieved.
26314

Looks wonderful as just numbers and graphs, the reality is much worse.
 
D

Deleted member 3831

Guest
TT
if you had 1 degree increments for your measured data, your graphs would probably shown up a few discrepancies. I don't know because obviously I am not privvy to your design details, but that's my guess.
I start from measured cam data when I want to work out what it will be when realised at the valve.
Cam data in a somewhat different form is also used for manufacturing.

The data resulting from a valve lift design exercise has to be converted by running it backwards through the valve train to find what the cam profile must be. Before this point is reached some design subtleties have to be indulged in, for dynamic events are not the same as the static ones.

To give the exhaust opening as an example. If the engine is turned over slowly by hand (is this possible?) the exhaust cam has to initially load up the valve train components, and increase loading to the point where the seated pressure of the exhaust springs are neutralised, and only at that point will the next movement of the cam initiate the first fraction of valve lift. This will be some small distance along the first opening (or quieting) ramp. The moment the engine fires this situation is drastically altered, for now there is also the residual cylinder pressure resulting from combustion that adds to the resistance opposing the cam.

This additional pressure is considerable, and the forces involved place severe loads on the valve train that most always results in some pushrod buckling, thereby delaying the actual point of exhaust valve opening by as much as 20 camshaft degrees or more. The buckled pushrod is now a form of spring, and eventually it will spring back to more or less it's original form, quite often as it does so pushing the valve to open further than the cam dictates.
This is not the whole story, just the exhust valve opening given as a typical example of the situation that exists with any pushrod valve train system
What can one do? As I mentioned earlier to Greg, its a choice between a rock and a hard place.
For example a steel pushrod as large in diameter as can be accommodated and you have the difference in expansion rates to worry about. A solid steel pushrod will give the optimum stiffness but carry the maximum weight penalty.
This above applies mainly to a racing engine, touring requirements usually involve running at considerably lower engine speeds, so the problems are reduced in their magnitude, but they are still there.
 

timetraveller

Well Known and Active Forum User
VOC Member
The way I did it was somewhat convoluted. Remember that not only am I not a cam designer, I am not even an engineer, just an astronomer. I started off with several criteria. The first was that I wanted a cam which gave good mid range power and peak power at 6,000 rpm. Secondly I wanted 0.45" lift, not about 0.36". Thirdly I wanted to do this with the acceleration of the valves never exceeding that of a Mk 2 cam. Forth, I wanted to design around a curved base follower as I had convinced myself that I could not get a symmetrical valve lift curve with the Vincent geometry and I wanted the symmetrical curve so that I could get a flat based acceleration curve. The extra lift was to be obtained by ratioed rockers, not on the cam itself as there are restrictions within the tappet covers. I used a spread sheet to get the curves shown above and then modelled the Vincent timing system in ACAD. Given that I then knew what I wanted from the spread sheet all I had to do was move the model of the curve based cam follower enough to move the pushrod the required amount and in the ACAD model drag the cam profile at one degree intervals up to meet it. The points around the cam were connected by a spline function to ensure a smooth and continuous surface. I did wonder about trying to make the cam a 'polydyne' but having measured the compression of the valve train by applying a lever to the outer end of the rocker it did not seem worthwhile. I also realised that although I could have designed for valve train compression at any one value for the revs it would not have been correct at others and this was meant to be a road cam, used over a wide range of revs, not a race cam to be kept in a restricted rev range. I had never realised that the maximum acceleration of the Vincent valve is equivalent to 200G. Hence the need for strong valve springs. I wanted to reduce that requirement.
I realise that to a professional cam designer this will seem a very amateur approach but we chaps who mess about at the bottom of the food chain just have to do the best that we can.
Thank you for taking the time to respond to these postings.
 
D

Deleted member 3831

Guest
TT
I actually applaud what you've done in trying to get yourself an understanding of cam design. There can't be many working in astronomy with similar interests that you can converse with I'm sure.

All my life I've been involved with racing where petrol engines of various types provided the motive power, four wheels, three wheels, two wheels and even F1 power boats.
I should stress that what I've discussed on this particular post is based on how I tackle a new design, and the criteria I deem important. If you are happy with your present way of working, keep at it, but I hope I've been able to show that there can be more to it if you're so inclined.

Whereas you are working with a camshaft containing lobes fixed in position, I can make any number of variations in position, and mix and match lobes in a simulator program when working on an engine development project. A final decision will be made and the relevant cam files will go to the company to be ground up, with details attached to advise of the amount of lobe separation desired. These files always start with the maximum lift, or nose position, designated as zero. On a multiple cam shaft, this becomes an indexing point of reference for all the other cams

Getting cams ground up can be a problem because the good cam shops are usually expensive, and are loathe to take on just a couple of cams, they would much prefer to work in 100's. If you have designed your cams for use with a follower that has a radiussed pad, you may well have a negative component in the flanks which might just require a smaller grind wheel than is normally used.

Finaly I would query the number of Vincent cams that are around today, are they copies, if so, copies of what ? If the cam is caimed to be a new one, where did it come from. There are obviously new MK1 and MK2 profiles available, but it seems that by now they are likely to be third or fourth generation copies, and by now some of the original design may well have become corrupted or changed
Is there any such thing as a standard profile of whatever mark that can serve as a master for future cams to be copied from?
 

oexing

Well Known and Active Forum User
VOC Member
Regarding original MK 2 cams, you can clearly see in my photos that I had genuine Andrews MK 2 for copying the cam that I sent TT for his test equipment. These were obviously not Vincent company made from the fifties but that was my point: 30 years ago I was more interested in new sets containing at least some quieting ramps on the lobes that the originals did not have. Andrews had reworked that classic design to the better I believe. So as has been said, our objective was to see if we could recommend these cams made today by Megacycle in supposing they continue to use data from Andrews for their production. TT has confirmed my verdict on these Andrews in so far as he made best use of equipment available to him.
I am too busy with my restaurations and do not spend time on Performance Trend programs these days to do simulations about dynamics in the Vincent valve gear train . Seems the programs do not offer enough input of parameters to be usable. The configuration can accept one rocker in the head but there is another , the cam follower, that is not cared for. So I cannot see to get the program do something useful in its simulation.
Anyway, the haziness about useful software for designing cams that includes all components in the valve train remains intact so this discussion looks a bit pointless to me, sorry .

Vic

26318
 

bmetcalf

Well Known and Active Forum User
VOC Member
Do you use “smoothing “ to account for the roughness of even the finest finish when going out to so many decimal places?
 
D

Deleted member 3831

Guest
Smoothing can be used when making the initial measurements of the profile, but I don't use it because if there is roughness I would rather be aware of it.
Normally it isn't a problem with lift, be it cam or valve, but the third derivative, jerk, can be a terrible mess and impossible to understand unless some smoothing is applied.
Regarding the number of decimal places being used, 8 is the maximum available, but are not that often used. 6 are used frequently, and when differentiating the reason is apparent from a study of the attached graphics.
Note the header which draws attention to one data set differentiated by subtraction, while the other shows the same data set processed by a curve fitting mathematics.
 

Attachments

  • Differentiation 1.jpg
    Differentiation 1.jpg
    70.5 KB · Views: 23
  • Differentiation 2.jpg
    Differentiation 2.jpg
    76.6 KB · Views: 23

timetraveller

Well Known and Active Forum User
VOC Member
When I was designing the cam with the curved based cam follower I plotted the movement of the contact point between the cam and the follower over the whole of the cam lobe. Remember, this was a curved based, 1.5" diameter, follower so it probably has some relevance to this question whereas a flat lever based follower would show something different. I cannot find the graph at the moment, despite looking through 18 different spread sheets, but my memory is that the contact point varies so rapidly, forwards and backwards, across the base of the cam follower that I could see no way in which a roller could possibly change direction quickly enough to avoid skidding. Over to you.
 
Top