Another Knucklehead Build

[kvom]

I assume that the amount of lift for each lobe is calculated vs. the angle of the lifter. Or does it make a difference in a model?

 

[mayhugh1]

I assume that the amount of lift for each lobe is calculated vs. the angle of the lifter. Or does it make a difference in a model?

kvom,
Theoretically, it should. But as you say, it shouldn't make any difference in a model. - Terry

 

[Rozlo]

I've just read all 13 pages of this build to catch up and all I can say is Simply Awesome. This is a work of art. I would think it would take some serious time to machine without CNC. Makes me want to go out and buy a CNC machine. But not yet I'm still a beginner. I will most definitely keep up with this build. Thanks for sharing.

 

[rodue]

I was too cheep to purchase the bearings, so I increased the size to .25 and replaced it with drill rod roller It dose great a problem on insulation but I have conquered that. I will have to see how well it works. I am going to have to add a spring to the push rod it doesn't want to return without help. As I said earlier a square broch I think would work better on the lift.

 

[mayhugh1]

Work on the lifters began by machining a set of guide bushings for them. These bushings provide bearing surfaces for the lifters inside the guide blocks and the roof of the gear box. They were machined from SAE-660 bearing bronze for smooth sliding fits with the lifters.

It's important that oil returning through the pushrod covers from the rocker boxes be allowed to drain quickly back into the gear box to avoid filling up the covers with oil. Four grooves were machined along the length of each bushing to facilitate this return. Three grooves were milled around the bushing's o.d., and one was broached through its i.d. Oil returning through the broached groove will lubricate the body of the lifter inside the bushing before finding its way (hopefully) to the cam. The bushings were dimpled for setscrews that will secure them to the guide blocks. They'll be positioned far enough below the top surfaces of the guide blocks so the ends of the pushrod covers don't interfere with the oil flow.

The lifters and lash adjusters were machined from O-1 drill rod. Both have wrench hex's machined into their tops as aids during valve adjustments. The drawings specify 6-32 for the threaded sections which didn't permit the use of jam nuts due to lack of space inside the pushrod covers. I reduced the thread size to 4-40 which allows the use of small pattern jam nuts. The tops of the adjusters were machined with polished hemispherical cups to accommodate the ends of the pushrods.

All four lifters are identical except for their integral tappets. Each was machined for either a two, three, or four degree crown depending upon its particular location. The locations were engraved into the flats of their hex's in order to avoid confusion during assembly. The CAD drawing shows the design of the 4 degree crown as an example, and a microscope photo shows the actual machined result. What would otherwise have been a pointed center was blended into the crown using a shallow contour. In order to provide wrench access when doing valve adjustments, the lifter length was selected so that while resting on the flank of the final cam, its hex will be just above the top surface of its guide block.

Some experimentation with tool selection and its feeds and speeds was done while CNC machining the tappets so they would come off the lathe with accurate and extra fine finishes that wouldn't require a lot of risky manual polishing. The process was tedious, but I was able to go from the lathe to 1500 grit paper before finishing with buffing compound. The lifters were quenched in oil after a one hour soak at 1450F and then tempered at 350F. (The cam will be tempered at 450F.) The lash adjusters weren't heat treated.

With the lifters temporarily installed in the engine in their bushings, I was able to check their contact patches (or more correctly, their contact lines) on the inked lobes of one of the cam blanks machined earlier. Theoretically, these lines should have been about .027" wide and located on the inside halves of the lobes. Measurements of the two and three degree lifters showed the widths to be between .020" and .026" and a little closer to the lobe centerlines than expected. The contact line for the four degree lifter was only .014" wide, but almost .005" of its missing width can be attributed to a geometry issue created by the excess starting stock on the cam blank. - Terry

 

[Scott_M]

That'll work !
Nicely done.
Terry your engineering skills are as impressive as your machining skills. You set the bar pretty high for the rest of us. This is greatly appreciated and shows what can be accomplished when you really give it your all.
Thanks

Scott

 

[kvom]

Any reason that the lifters wouldn't want to be centered on the cams?

 

[mayhugh1]

Scott, thanks for the comment. It's much appreciated.

Kvom, no reason that I can think of. After I get an actual cam machined that I'm happy with, I may make a new blank with the lobes repositioned. I'll want to see the contact areas on a finished cam before I do that, however.

Terry

 

[tornitore45]

Terry, what is the diameter of the hemispherical cavity? What is the diameter of the push rod section impinging in the adjuster?

I have a similar situation in the Edward 5 radial. The drawing calls for a 0.094 rod with 0.075 ball ends working into a 0.080 hemispherical cavity 0.062 deep.
On paper it looks wonderful but after machining such small features it "feels" like there is not enough captivity to keep the rods in place.
Your picture looks like the cavity is much more deep and "secure" than what I see on mine.

 

[mayhugh1]

Mauro,
The diameter of hemispherical cavities are .157" and they're .087" deep. The diameter of pushrod balls will be slightly under that. I haven't decided in the diameter of the pushrods themselves yet - probably .125" diameter though. - Terry

 

[tornitore45]

OK that explains it. Much larger that the stuff I am dealing with.
I am toying with the idea of using a slender cone with round point into a conical hole to provide a deeper engagement for the same diameter.

 

[mayhugh1]

Harley's world famous irregular firing sequence is the result of using a single pin crankshaft in a 45 degree V-twin. The rear cylinder fires 315 crankshaft degrees (360-45) after the front cylinder fires, and the 4-stroke cycle completes with the front cylinder preparing to fire again after the engine rotates an additional 405 degrees (360+45).

Similar to the full-size Knucklehead, the model engine uses a single four-lobe camshaft albeit with different lobe assignments. The lifter block angle separations in the model are 46 degrees for the intake lifters and 56 degrees for the exhaust lifters. (These angles may be different from those in the full-size engine.) The intake lobes on the model engine's camshaft should therefore be separated by 111.5 camshaft degrees (315/2 - 46) since the camshaft rotates at 1/2 the speed and in the opposite direction to the crankshaft. Similarly, the exhaust cam lobes should be separated by 101.5 camshaft degrees (315/2 - 56).

An end view of the camshaft in the drawing in the download shows the intake lobes for the front and rear cylinders located suspiciously close together and likewise for the exhaust lobes. Accounting for the angles between the lifters, a little math shows the cam in the drawings will open both intake valves within 16 degrees of one another, and it will also open both exhaust valves opening simultaneously.

Unfortunately, the cam provided in the drawings isn't actually for a Knucklehead but instead provides an obscure 'Big Bang' timing that blends both cylinders into a single behemoth with roughly twice the volume. A V-twin configured this way will have a distinctively loud and hard pounding power stroke but most likely won't perform well with a single carb setup. The cam in the drawings also orders the valve events in such a way to cause the engine to run in a direction opposite to that of a Harley. Upon realizing this earlier in the build, I modified the designs of the oil pump and distributor and designed the starter clutch to work with the engine running in the proper Harley direction which is clockwise when viewed from the gear box side of the engine.

The 'Big Bang' cam provides intake valve lifts of .050" (.060" lobe lift x .84 rocker ratio) and exhaust valve lifts of .070" (.065" lobe lift x 1.08 rocker ratio). Since both valve seat diameters are .510", the lift-to-diameter ratio for the intake valves is .096, and for the exhaust valves it is .140. When valves have been properly sized for the needs of an engine, a lift-to-diameter ratio of .250 is generally considered ideal. At peak lift, it's the ratio that corresponds to equal face and gap areas which means lift isn't limiting flow. The Knucklehead's valves are probably larger than needed, and so throttling their flow with lift most likely won't be an issue. The model's maximum allowable lift to insure the engine remains interference-free with its original low compression pistons is .100", and so a replacement cam could possibly provide a bit more lift.

Harley's original Knucklehead cam specs are surprisingly difficult to find, but a number of aftermarket suppliers offer what they claim to be a 'stock' equivalent:

Intake (234 deg duration): opens 13 deg BTDC
closes 41 deg ABDC
Exhaust (240 deg duration): opens 44 deg BBDC
closes 16 deg ATDC

A widely available 'S' grind is also available from the aftermarket:

Intake (262 deg duration): opens 27 deg BTDC
closes 55 deg ABDC
Exhaust (262 deg duration): opens 55 deg BBDC
closes 27 deg ATDC

where I assume the 'S' stands for 'symmetric'. All these numbers are, of course, in crankshaft degrees.

Overlap, the crankshaft angle over which the intake and exhaust valves are simultaneously open, is an important cam spec because it controls the rpm range over which the engine will make its power. Rather than deal with a lot of opening/closing angles when comparing cams, the lobe separation angle (LSA) is often used instead. LSA can be calculated by subtracting the overlap from the average of the cam's intake and exhaust durations and then dividing the result by two in order to express it in cam degrees. A rough dividing line between mild and wild is an LSA of 110-112 degrees. Both of the above cams happen to have LSA's of 104 degrees which, surprisingly, is on wild side.

The next step is to come up with a suitable replacement camshaft for the Knucklehead. The goal is a design that correctly compensates for the model's lifter block angles and produces an engine running in the proper direction with conventional 315/405 Harley timing. - Terry

note: when comparing the camshafts in the two photos, remember they were designed to turn in opposite directions.

 

[mayhugh1]

For a 'mostly display' model engine, I'd tend to avoid camshafts with large overlaps and narrow LSA's that are responsible for poor idle quality and high rpm power bands. My goal for the Knucklehead's cam is an easy-to-start engine capable of low rpm idling. It needs to make only enough power to overcome its own losses if that means minimizing the amount of generated heat. For my cam I've decided upon a small overlap in order to maximize manifold vacuum for improved starting and idling. I also plan to open the exhaust valve a little earlier than normal to make the engine a little louder. At the expense of some power, it might help the engine shed heat through the exhaust for longer run times. My target specs are:

Intake (225 deg duration): opens 5 deg BTDC
closes 40 deg ABDC
Exhaust (250 deg duration): opens 65 deg BBDC
closes 5 deg ATDC
intake lobe lift: .090" (.075" @ valve)
exhaust lobe lift: .070" (.075" @ valve)

The lobe separation angle for this cam works out to be 114 degrees which puts it in the mild category.

For verifiable measurements, a specific lift at the 'open' and 'close' angles needs to be specified. Full-size cams are always specified this way with .020" and .053" being typical for motorcycle cams. Since there doesn't appear to be a standard for model engines (zero is used as often as not), I've selected .004" lobe lift to define the 'open' and 'close' angles for my cam.

When designing a cam lobe, the angle between the starting points of the lobe's ramps needs to account for the lift used to specify its duration in crankshaft degrees. In order to determine these points, I used a virtual cam tester that I created in SolidWorks specifically for the Knucklehead.

The tester was designed to work with the CAD/CAM model that will eventually be used to machine the camshaft so measurements can be made as though the actual cam was operating inside the actual engine. A degree wheel, attached to a virtual crankshaft, is geared to the camshaft under test. Lifters operating at angles matching those in the actual engine ride on the cam's lobes just as they will inside the engine. I began work on the tester while still planning to use roller lifters. I didn't bother modifying it after shifting to non-roller lifters since the differences between the two in the tester appears to be insignificant. Although I couldn't figure out how to make 'live' measurements inside a SolidWorks assembly, the lifters can be individually measured with respect to a reference surface and their precise lifts recorded for any crankshaft angle.

The tester's small red arrow is a reference pointer for the degree wheel that can be moved anywhere around its circumference. The 315 degree outer collar can also be independently positioned around the degree wheel and used to indicate the relative locations of the tester's virtual cylinders in crankshaft degrees. My original reason for creating the tester was to have a tool to sanity check the dizzying compensations that will have to be made to the angles between the cam lobes to accommodate the unequal angles between the intake and exhaust lifters.

Using the tester to iteratively design the cam's intake lobes eventually showed the angle between the starting points of the lobe's ramps had to be 127 degrees to obtain a crankshaft duration of 225 degrees at .004" lobe lift. This was found by measuring the angles on either side of the lobe's center where the lift measured .004". The actual angle between the lobe's ramps was increased in small steps until the specified duration was finally achieved. A similar procedure was used to design the exhaust lobe. Final dimensioned drawings of both lobes are shown in the photos.

After the contours of the intake and exhaust lobes were defined, their angular positions relative to one another needed to be determined. Since the intake lifters are 46 degrees apart, the cam will be machined so the centers of the intake lobes are 111.5 degrees apart (i.e. 315/2 -46). Since the exhaust lifters are 56 degrees apart, the exhaust lobe centers will be 315/2-56 = 101.5 degrees apart.

The final angles needed to complete the cam design are the separation angles between the lobes of each intake/exhaust pair. For this engine, these are somewhat tricky. If the angle between the intake lifters were identical to the angle between the exhaust lifters, both separation angles would be the cam's LSA or 114 degrees. Since the angle between the exhaust lifters is 10 degrees greater than the angle between the intake lifters, this difference will alter the separation angles.

While standing on the gear box side of the engine with the camshaft rotating CCW, the rear exhaust lifter is located 5 degrees past the rear intake lifter. This 5 degrees must be added to the required LSA when computing the required angle between the rear intake and the rear exhaust lobes (i.e. 114 + 5 = 119 degrees). The situation is reversed for the front lifter pair. In that case the front intake lifter is located 5 degrees past the front exhaust lifter, and this 5 degrees must be subtracted from the required LSA when computing the required angle between the front exhaust and front intake lobes (i.e. 114 - 5 = 109 degrees). With these compensations applied to the virtual camshaft, the LSA measured 114 degrees for both the front and rear virtual cylinders in the tester.

One last angle of importance, the centerline angle, isn't needed for the camshaft's design nor its machining but is required to install and time it to the crankshaft. The intake centerline is the point of highest lift on an intake lobe. The centerline angle, illustrated in one of the photos, is the number of crankshaft degrees between the intake centerline and its cylinder's TDC. Either the front or rear intake lobe and cylinder TDC can be used. For my cam, I'll set the centerline angle of the cam's front intake lobe 107.5 degrees (225/2 - 5) past the front cylinder's TDC. The rear cylinder can just as easily be used.

The camshaft can be slightly advanced or retarded by altering its centerline angle a few degrees during installation. This can make a small change to an engine's performance which can be of interest to a performance enthusiast. The slotted mounting holes in the cam gear provide for this, but my plan is to install the cam 'straight up'.

Timing the virtual camshaft to the tester's virtual crankshaft is relatively simple. The crankshaft with its attached degree wheel is first rotated to the point of maximum lift of the front intake lifter. This is the center of the front intake lobe's duration and is tagged by setting the red arrow next to the zero degree point on the degree wheel. The outer collar is then rotated until the front TDC arrow is 107.5 CCW degrees behind the red arrow. Then, as the crankshaft is rotated, the virtual cylinders (indicated by their arrows on the collar) will reach their TDC's whenever the degree wheel's zero passes under their respective TDC marker.

The next step is to finally machine the actual camshaft. The last photo in this post contains a worksheet that will be used to do the machining. The angles as they're displayed in the worksheet are more suitable for machining the camshaft than for understanding where the angles came from. For completeness, a follow-up photo shows the pertinent angles between each lobe. - Terry

 

[mayhugh1]

One more photo ...

 

[brotherbear]

Terry
Once again I am compelled to thank you for the time and energy you so freely give the effort to explain your process in a concise understandable way with illustrations, photographs, and thought-provoking dialogue. Without any pretense of hubris, you explain what you are doing what your concerns are, how and when to address those concerns. With an uncompromising skillset, you unashamedly construct jigs, clamps, sacrificial hold-downs and consistently machine multiple parts. Your willingness to share the ocasional error and reconstruction or altered fabrication process.
I have gone from immediately reading your posts to making a note of...and setting aside some time when I can have a cup, sit down and visit Terry's shop, see what he's been up to and learn something as well as just enjoying such quality of workmanship.
Thank you for sharing your efforts with me!

 

[minh-thanh]

Terry
Once again I am compelled to thank you for the time and energy you so freely give the effort to explain your process in a concise understandable way with illustrations, photographs, and thought-provoking dialogue. Without any pretense of hubris, you explain what you are doing what your concerns are, how and when to address those concerns. With an uncompromising skillset, you unashamedly construct jigs, clamps, sacrificial hold-downs and consistently machine multiple parts. Your willingness to share the ocasional error and reconstruction or altered fabrication process.
I have gone from immediately reading your posts to making a note of...and setting aside some time when I can have a cup, sit down and visit Terry's shop, see what he's been up to and learn something as well as just enjoying such quality of workmanship.
Thank you for sharing your efforts with me!

 

[kvom]

I was wondering whether the plans included specs for the cam, and if so did you use these as starting points for refining to your own preferences? Otherwise I'd think very few people could even attempt this build.

 

[mayhugh1]

Kvom,
The plans include specs for a "big bang" cam that fires both cylinders on top of each other and runs the motor backwards. My camshaft, which is intended to run the engine like a Harley, was designed from scratch. I don't know why that cam was used, but I'm not sure that it's the reason why the engine hasn't been attempted by more builders since most wouldn't have realized the issue with it until well into their build. I'm trying to include enough information about mine, though, so others can use it if it actually ends up working as I hope.

The real reason for the engine's lack of popularity is probably its complexity. This has been the most challenging engine I've work on to date which really hasn't been a big surprise to me. The original designer did an incredible job coming up with the plans - a feat I couldn't begin to match. But nothing this complex could be expected to be gotten entirely correct without an actual build or two to debug the paper. It's been identifying and coming up with fixes for all those issues that I've run into that has been so time consuming and something that I appreciate many others aren't interested in doing. - Terry

 

[kvom]

This has been the most challenging engine I've work on to date

Even after the Merlin? wow.

 

[petertha]

Terry> A rough dividing line between mild and wild is an LSA of 110-112 degrees. Both of the above cams happen to have LSA's of 104 degrees which, surprisingly, is on wild side..... The lobe separation angle for this cam works out to be 114 degrees which puts it in the mild category...

I really like your useful tutorial on cams & especially how you simulated the settings in CAD. Thanks for laying it out, its not exactly an easy subject (at least to me).

- Are the gears more for visual or you are also evaluating mechanical attributes? ie. presumably have a numerical gear ratio defined in the motion mates & they could be 2 circular discs for that matter. Or are you actually registering a cam reference point to a tooth reference or figuring out mechanical gear size & tooth pitch etc.?

- I adopted the LSA calculation into my own spreadsheet tool. Just curious - did you ever calculate out what your radials worked out to for comparison? I am getting a low number like 96 deg on this particular (methanol glow) engine. Are the guideline numbers you reference above suited to a particular engine type or maybe influenced by the type of fuel they burn?