E: Engine Camshaft

[Cyborg]

Spintron....

A fairly large percentage of the warped brake rotors that I came across were caused by folks jamming the wheel nuts on with an air impact driver. Grossly over torqued they would distort the hub and that in turn would distort the rotor.

 

[Deleted member 3831]

TT
sorry I didn't reply to your query re the difference between cam profile data and how it can be related to lift at the valve. I must stress the graphics are cam lift data only.
I would not start to work from the cam, but would start from the valve lift design that all my calculations and simulations (both static and dynamic) indicate would be capable of meeting the design performance target. This valve lift design is then processed through the valve train of the engine in question to arrive at the cam, where a profile will be calculated.
It is of course necessary to have complete data for the valve train, dimensionally and with X and Y co ordinates where appropriate, together with all weights and inertia and stiffness values for the moving components, which will also include full valve spring data , and this can take some considerable time to collect if the data is to be accurate. which it must be.
With some engines I have to build in a provision to cover an over rev through a missed gearchange or similar, and it often involves a number of detail changes in order to arrive finally at a satisfactory outcome.
Cyborg
a great little Spintron movie, but the end comments re valve seating bounce where poor springs are blamed, is not strictly correct. Bounce can be the result of a badly designed closing side of the cam, and with a better design the springs would be OK. No matter, the spring oscillations are shown in the output graphics of my dynamic simulation, and when I get it wrong I see the bounce , just as the Spintron shows it.

 

[Cyborg]

I wondered about the spring comment. This is all way above my pay grade, but after fiddling around with some of Honda's smaller multi cylinder engines, it becomes apparent that they don't achieve that high RPM, by throwing stronger valve springs at them. It seems to me that stronger springs are often used to try and correct a problem elsewhere..... Like my neighbor with the drag boat that changed cams as frequently as oil.
I always thought that if the opening and closing velocity was pretty much equal, that would create a more fluid/harmonic situation with less bouncing around. Now I'll exit stage right, before I dig myself in any deeper.

 

[greg brillus]

I have spoken to Ken Horner several times about the valve train in a Vincent, and there is always an expression of despair on his face. The biggest limitation is available room, the cams/followers are too small to take any real heavy loading. They try and get around this using the same pivoting cam followers but with a roller mounted at or near the tip and symmetrical cams. They then use rocker arms with a much larger ratio than the stock 1 : 1 and I'm guessing they must get at least 600 lift at the inlet valves. The rework on a standard head type Vincent engine to get that kind of lift is very difficult, along with everything else, this is a major limiting factor for most owners even if they want to build an all out race engine. These engines were designed to be large valve low lift valve train with pretty low valve spring rates. The cost of trying to achieve any of this is quite high and the numbers interested in doing so are small, given there are very few Vincent racers out there anymore.

 

[oexing]

Publishing one´s measuring machine´s accuracies may be veeery impressive but at these numbers all this does to me is rolling my eyes . . . . I have limited resolutions on my test stand for cams to 1 micron (1/25 of one thou) and 1/4 of one degree on the rotary encoder. I am not interested in anything better as I know from my professional life being retired toolmaker , accuracies in REAL life are quite different from what academic people believe in their imagination. All these numbers may have some meaning for computer calculations to arrive at some result in the CAD exercise but in the toolroom you are happy to see one thou or half of it close to the desired size, no way near 8 decimal places in existence. In fact, when you get a new cam from any company, you are lucky to see a base circle with less than two thou runout, all other base circles with runouts all over the place.
Anyway, this is not our concern, we just had a look at cam lobe shapes for comparison and evaluation from other cam data bases in order to be able to assist buyers in finding suitable camshafts. I do have a few programs from Perfrmance Trend, cam analyser and cam grinder. BUT, big question is,how do you arrive at the cam shapes ?? And right here all postings are veeery hazy, it is quite obvious that the start has to be the valve lift diagramm. I can provide 40 different computer calculated files , different in lift numbers, acceleration , and duration, three decimal places - in millimeters, one micron that is.
Now the REAL hard work is to model all levers, rocker finger shapes, cam follower details, into a CAD program that handles these kinematics to transform in the end valve lift numbers into cam lobe shapes !! I was toolmaker in life , with all traditional skills, but CAD was not one of these. So is there a chance to get any info here on suitable programs for such calculations , Fusion 360 , not so much ?? I believe Performance Trend can analyse dynamics in the valve train by simulation but only AFTER a cam was fabricated, it does nothing for designing a cam.
So the procedure to arrive at a particular cam shape , starting with valve lift numbers , is definitely my big question but I guess that haziness about details will continue . . .
Valve bounce and other adverse effects are seen mainly in race type engines and I believe there is no universal solution in all cases and has to be dealt with individually. Most Vincenteers will be more interested in smooth operation by mild acceleration and quieting ramps so no radical fast wearing shapes acceptable.

Vic

 

[Monkeypants]

Well, I'm out, I'll never make it.
Dan Smith tells me I can't measure to 4 decimals, 3 at best and now I learn that it takes 8 to be worthy?
Good thing the old bike doesn't know that and it runs plenty strong anyhow.
Sometimes what you don't know can't hurt you!

Glen

 

[Deleted member 3831]

Cyborg
valve springs all possess a natural frequency of vibration. it's essential that this does not co incide with that coming from the cam design anywhere within the engines normal operating speed range. In olden days this was not well understood at all, but now it's part of the overall design process, and most usually this frequency clash is arranged to be outside of the engines main operating speed range. If you now add a pushrod into the mix, you really do get problems, for the last thing you ever want it for the natural frequencies of the pushrod and the valve spring(s) to clash. It is all so easy to get into what amounts to a can of worms, and the more one tries to push the limits, the deeper into the can one gets. I've been deep among the worms plenty of times !

Greg
you describe exactly the problem that anyone wishing to seriously boost the performance of a Vincent engine faces. Do you stay within the original outward design envelope, and try to make all modifications within the engine cases, or throw away these concerns and modify as and how necessity determines? Obviously it's for the individual to decide, but Classic Racing rules usually require the outward appearance to be consistent with the original engine, so one ends up being as it were, stuck between a rock and a hard place !

Oexing
your measuring jig appears to have all the essentials for you to obtain meaningfull data.
The value of data to 8 places of decimals is certainly not necessary if only the profile of the cam is the sole interest, but it's when the second and third derivatives are to be investigated that the more detailed data becomes a necessity. The standard for measuring cam data is the Adcole system, very expensive but used throughout industry, and can be tailored to suit each individual users requirements. Almost as good, and certainly far less expensive are Performance Trends Cam Analyser, and Audie Technologies Cam Pro Plus. For either of these you would need their monitor and software program, you have apparently the requisite encoders.
John Andrews also had design software, but since he's been taken over I don't know whether this is still available. He marketed EZcam, a software program designed for measuring and analysis of cam data.

For an understanding of the calculations of the pushrod system, you will find very detailed explanations with all relevant diagrams and mathematics in Jeff Williams previously mentioned book

 

[Flo]

Grey One, can you please qualify thr 8 decimal places in your data e.g the dimension inch or mm. So do you think that you need a precision of +or - 0,00000001 mm or 0.00000005 inch?

Let me make clear that the analysis of an existing cam profile surely will present challenges bordering on the impossible.

 

[greg brillus]

The only other thing I would say about having a performance cam in a Vincent, especially if the closing point of the inlet valve is very late, is that the engines can a be very difficult to start. I'm guessing this is because so much dynamic compression is lost. Hence the need for high comp pistons when big cams are used. I had 2 fully restored shadows with new Mk 2 cams from a known source that were not only very difficult to start but would not idle very well at all, no matter what you did.

 

[Deleted member 3831]

Flo
both, metric or imperial. The full 8 places are not always used, but are available and can be used as and how the linear encoder detects. If you look at the short data list I posted earlier you will see that not all entries were to 8 places, many were 7, and further into the list (not shown) some were 6 and 5 places.

The measuring equipment I use is Audie Technologies Cam Pro Plus, and for some designers who wish to work to the fourth derivative, Snap, such detailed data is essential, (Cam Pro Plus has lift, velocity, accelleration and jerk, no snap, so if I wanted snap it would have to be derived from a separate computation).

Design may be extremely detailed, where necessary to the 8 th decimal place, and when verifying the accuracy of the manufactured cam it's generally accepted that the measuring accuracy should match the design accuracy for depth of data, this for ultra high performance high speed use, where any slight deviation from design could well end up in a mechanical disaster.
Because I have always designed this way, I use the same methods for whatever work I am involved with, be it an ultra high revving multi cylinder race engine, of a much lower revving single cylinder Classic racer.