Wednesday, 15 January 2025

Making a K-17/5 Cam for KTT 55 – Part 1: Reverse Engineering an Existing Cam

Cams in good condition for early cammy Velos are hard to come by. I had a number of decent K-17/2 cams (which superseded the early K-17 cam) that were reground by Newman Cams and used on both the AJcette and the V-Twin (see here) and I have one more of these but, ideally, I would like to use a K-17/5 cam. There are 2 reasons for this: the later K-17/5 cam is considered superior in the respect that it has a smaller base circle diameter and consequently a lower average contact speed with the rocker skid – which is better for the overall mechanical integrity of the cam/rocker combination and, secondly, and this is my own opinion, it is much easier to work with in the respect that it has a threaded portion that allows it to be pulled off the camshaft much more easily that the K-17/2, which has no such threaded section.

A K-17/5, on the left, and K-17/2 cam are shown in the picture below for comparison.

Just for the record, the K17/2 cam (which is often stamped 24C) needs a Sykes-Pickavant type bearing puller to get behind the cam to remove it from the camshaft. This is what mine looks like.

They are pretty cheap and worth every penny to avoid damage to the large K-12 bronze bearing,

I only have one K-17/5 cam, shown in a different camera angle below, with the internally-threaded puller that I made when I took it off a camshaft long ago.


This cam was reground some years ago by Newman and is in excellent condition. However, I only have this one and decided that, as I would like to fit K-17/5 cams on more than one of my engines, I would make another by replicating this one. I have never made a cam before and this has proved to be an enlightening experience.

The top-down approach that I adopted to copy this is:

1)    Measure the existing cam

2)    Machine a blank (ie all the “round bits” and the thread) on the lathe using O1 tool steel

3)    Machine the inlet and exhaust cam profiles on the milling machine

4)    Use an external machine shop to put in the keyway for the woodruff key (I can’t do this with my equipment)

5)    Clean up as necessary and heat treat.

Simple really.

However, the question then arises as to what exactly to measure and how will it be machined?

The picture below shows a depiction of the cam positioned under an end mill with the Z axis aligned with the vertical axis of the milling cutter and the Y axis being the direction of a traverse of the table beneath the cutter.

To machine the cam from a blank, ie a round section, the cam is now rotated in the rotary indexer or dividing head by some angle, θ, the cutter is lowered by an amount Z and the table of the milling machine traversed in the Y direction so that the end of the cutter will remove material at the highest point and which is tangential to the cam surface, shown below,


….and then continue to rotate the cam incrementally through one revolution and remove material until the cam profile has been completely machined – with the equipment that I have this is the only realistic way I can machine the cam.

The key question is: what is the depth-of-cut for any given value of θ?

The depth-of-cut does not have a simple relationship to the cam profile, as will be discussed in a later blog. I do not have the cam lift profile so it wouldn’t help anyway.

To machine a new cam by copying an existing one the depth-of-cut needs to be measured for small increments in θ - I have chosen to measure and subsequently machine the cam using 20 of rotation. Yes, this is a very small increment but this is the level of definition that is needed to capture accurately the periods of maximum acceleration (and deceleration) and to avoid significant "flats" on the nominally smooth contact surface of the cam.

To measure and locate the position of the tangent point I made an electrical probe, shown below (together with a mandrel to hold the cam), in which there is a 1/8” diameter hardened and tempered pointed silver steel probe that is inserted into a ½” diameter shank and held in place with epoxy resin so that it is electrically isolated from the shank.


The cam/mandrel/rotary indexer was then set up on-centre in the milling machine and the probe connected to a multimeter such that an electrical circuit is made when the probe point touches the cam. The multimeter has a setting that provides both a measurement of resistance and sounds a buzzer when contact is made.


I do not have a DRO to give a digital reading for the vertical position of the probe and so I needed to rely on the micrometer reading


graduated in 0.001” and which I found measurements to be very repeatable.

The process for deriving the depth-of-cut data is as follows:

      1)    Rotate the cam through an angle Δθ

 2)    Lower the probe carefully and traverse the table in the Y direction until electrical contact is made.

 3)    Note the readings of the vertical position, Z, and the Y coordinate (referred to below as the Yoffset )

 4)    Repeat steps 1- 3 for 3600 of cam rotation

As I mentioned above, I have chosen Δθ = 20 of cam rotation to take measurements. The cam nearer the threaded end is the exhaust cam and I positioned this so that 00 is at (approximately) peak lift and then rotated clockwise to 2, 4, 6 degrees ….etc until the base circle is reached. This was then repeated from the same start position but rotated in the opposite direction (360, 358, 356, …etc) to complete the data collection for the exhaust cam. Without repositioning the cam, the inlet cam was then measured in the same way; it is very important that the same angular coordinate system is used to maintain the overall relationship between the inlet and exhaust cams.

One practical point on this measurement method: it is not possible to detect one single, unique point of electrical contact for any given position of the cam and, in practice, the buzzer will sound for 2 positions – one approaching for increasing “Y” and the other when the table is traversed in the opposite direction, ie decreasing “Y”.

The situation is shown below.

The 2 measurements were averaged to give a value of Yoffset midway between the readings. I found that the vertical coordinate could be repeatably measured to 0.0005” and the horizontal position to a couple of thou, depending on the angular position of the cam.

It should be stressed that the Y measurement of the tangent point is not required to machine the cam; it is only required to derive the cam lift profile as described below.

Here, the following variables are defined:

Θ the prescribed angle of rotation of the rotary indexer

Z the measured depth-of-cut

Yoffset the measured offset of the location of the tangent point

Φ the angle between the Z axis and the location of the tangent point

L the cam lift at rotation angle Θ

Rbase the radius of the base circle

Llift,max the maximum cam lift

The following relationships allow the cam profile, ie the relationship between the cam lift and the angle of rotation, to be determined:

tan Φ = Yoffset / (Llift,max + Rbase - Z)

from which

Φ = tan-1 (Yoffset / (Llift,max + Rbase - Z))

The angle Φ can also be defined in terms of the cam lift, L, as:

cos Φ = (Llift,max + Rbase - Z) / (L + Rbase)

giving the cam lift:

L = (Llift,max + Rbase - Z) / cos Φ - Rbase

And so, after many hours of taking measurements, I have ended up with the following information, which has been input and plotted from an Excel spreadsheet.

The raw data from the exhaust cam, the first one measured, looks like this.

 


I have indicated the locations of the exhaust opening and closing and the direction of rotation of the cam in the engine – the same angular coordinate system applies to all of the following pictures.

The data from the inlet cam is shown below.



The derived cam lift profiles are shown below.



Now, I realize that this is not the traditional way in which cam profiles are shown but this is what you get when the angular coordinate system is referenced to the position of maximum exhaust cam lift at zero degrees – and, to machine both cam profiles, they need to be in the same coordinate system. They are also shown with the direction of cam rotation in the engine going from 3600 to 00.

It is interesting to compare what I have measured with published data about Velocette K-series cams, although as you will see, it is of limited use. The following table is reproduced from Fishtail #80 (April 1969).

The data in the above table is in degrees crank angle and the clearance is measured at the valve. To compare the periods with my measured cam data the clearance should be modified by the rocker ratio, which is pretty close to 1 and makes very little difference (see below)

and the period divided by 2 to convert from crank degrees to cam degrees.

In the table below I have compared the periods, in cam degrees, for both the K17/5 and K17/4 cams using the valve timing data from the table with the measured data from my cam.

 

K 17/5 Cam (Fishtail)

                     Period Period

                    (Crank) (Cam)

K 17/4 Cam (Fishtail)

                     Period Period

                    (Crank) (Cam)

Measured

Period

IVO

390 BTDC

2880

1440

430 BTDC

2930

1470

1600

IVC

690 ABDC

700 ABDC

EVO

600 BBDC

2800

1400

680 BBDC

2960

1480

1600

EVC

400 ATDC

480 ATDC

The differences are almost certainly due to the fact that, because both the rocker skid and cam are curved surfaces and the skid does not bear directly onto the top of the cam, the contact line between them does not stay in the same place as the cam rotates and the rocker lifts.

It is also possible to compare the phasing of the cams. ie the angular difference between the peaks of the inlet and exhaust (strictly, these are not necessarily the locations of the peak lift but rather the mid points between the opening and closing). Using the valve timing given in FT #80 I end up with 102.50 cam angle difference between the peaks for the data quoted for the K17/5 cam; if I do the same thing with my data then it gives only 620.

Why such a big difference? The explanation is because taking the difference in the angular locations of the inlet and exhaust mid-points on the cam does not translate in a simple way to the valve timing. The contact lines of the rocker skids are not on the centreline of the camshaft when everything is installed in the cambox – see the picture below (this cambox is from another engine that I’m working on). 


I have put 3 lines onto this picture to show the centreline of the camshaft and, on either side and as best as I can judge, the position of the contact lines of the inlet and exhaust rocker skids with their respective cams which clearly shows they are not coplanar. It is for this reason that the phasing of the inlet and exhaust valve timings do not have a simple relationship to the cam profiles with the inlet valve opening earlier and the exhaust closing later than that suggested by the measured cam phase angle.

As the machining data for both inlet and exhaust cams profiles is now known the last exercise is to make a drawing for a blank. The blank is essentially the complete cam but without the profiles, shown below.


The one detail not shown is the keyway: this is 1/8” wide and 1/16” deep.

All the required information is now collated to be able to start cutting metal.

No comments:

Post a Comment