Another Knucklehead Build

[mayhugh1]

Pertetha,
The gears are there just because they're in the actual engine. It was a convenient way of getting the proper direction and speed of the camshaft with respect to the crankshaft. In the tester, I was trying to duplicate the engine as closely as possible because I needed a sanity check on the lifter compensations that were making my head spin. In use, the gears do get in the way of measurements sometimes, and I often 'hide' them using a feature of my CAD that allows one to look through selected components while retaining their functionality.
I never checked the LSA of any of my other engines. If I remember correctly, there wasn't much if any at all overlap in the radials, and so I would expect their LSAs to be fairly high. Also, all my comments were for a four stroke sparkplug engine. I've no experience with glow plug or two stroke engines and would have to some thinking about what, if any changes, would have to be made for them. - Terry

 

[mayhugh1]

The CAM software that I use (Sprutcam) has a four axis 'Rotary Machining' operation that's capable of machining camshafts. This particular operation first appeared in my now nearly decade old version of the software, and it was Sprutcam's first continuous 4-axis machining operation. Although probably much improved by now, its initial features were limited, buggy, and not well integrated into the software's framework. After considerable trial and error, I was able to use it to machine the Merlin's camshafts and the counterweights on its crankshaft. I'm still a rank user, but with my Merlin's notes as a starting point, I was soon simulating the two machining operations needed for the Knucklehead's cam.

I used an earlier made practice blank to fine-tune the operations' parameters in actual metal. A problem that's shown up in the past are crowned lobe surfaces created by the dished relief ground into the bottoms of cylindrical end mills. I got around this on the Merlin by having a local tool grinder remove the relief on a couple of my cutters especially for use with this operation. The relief on a stock 1/8" .010" corner radius'd carbide cutter that I had on hand seemed almost insignificant, and so I decided to use it 'out of the box'.

Each lobe was machined using a .005" depth of cut and an effective feed rate of .5 ipm. This feed rate is considerably slower than one would expect for an 1/8" carbide cutter working in a piece of drill rod, especially with such a shallow depth of cut. However, one of the bugs I've come across in this particular operation is an inverse time feed rate miscalculation for certain cutting moves under certain conditions that tend to chip cutter teeth.

The setup that I used required initially positioning the spindle over the center of the lobe to be machined and then lying to the software about its location. The lobe was machined using only z and y axis spindle moves while the rotary turned some 30 continuous revolutions below it. The machining marks left on the part as it came off the mill look much worse in the microscope photo than they actually are. They, along with any visible traces of a crown, were easily removed with 1500 grit paper.

Although the software is capable of generating code to machine all four lobes in one continuous operation, I machined them one at a time and manually reset the rotary and repositioned the spindle after each. Although unfounded, I was concerned about running into some obscure Mach 3 bug caused by the tens of thousands of degrees that would have accumulated on the rotary.

Both previously prepared cam blanks made it all the way to becoming finished camshafts, and so I ended up with a spare for my keychain. After an hour at 1465F they were quenched in oil and tempered at 365F. To prevent scaling during the high temperature cycle, the parts were enclosed in argon-filled stainless steel foil bags closed with double-folded seams.

Both cams were trial-fitted inside the engine and the lifter contact point locations verified with a dental mirror. With the lobes finally at their proper diameters, all lifters appeared to be riding on their lobes where they should be. With finger pressure on the lifters, the follower motions on the lobes of both cams felt silky smooth. One cam was selected for use, and a spacer between it and the gear box cover bearing was machined to remove the camshaft's thrust clearance.

I've included some photos that may be of interest. My next step is to machine a set of pushrods. - Terry

 

[tornitore45]

What provision do you have to phase the cam to interpolate between gear teeth?

 

[mayhugh1]

What provision do you have to phase the cam to interpolate between gear teeth?

See the photos in post #233

 

[tornitore45]

I see the gear is pretty busy.

 

[mayhugh1]

At 5/32" diameter, the relatively short (2-1/2") pushrods in the Knucklehead drawing package are pretty substantial. Even though I'd planned to use smaller diameter pushrods, I kept the 5/32" for the machined sockets in the 7075 rocker arms to take advantage of the wear resistance it would provide. Smaller diameter rods should look a little better, have less mass, and speed up oil return inside the pushrod covers. There will also be some extra wiggle room for the rods inside the covers.

Before machining them, I wanted an accurate measurement of each pushrod's maximum allowed length. There's easily enough range in the lash adjusters so the rods can all be made identical, but the covers provide only limited access for their adjustments. Customizing the length of each pushrod to its particular location will insure the adjusters stay inside their access windows.

In order to determine the lengths, I made an adjustable pushrod to use as a measuring stick. After cutting a set of four pushrod blanks to the lengths measured (minus .015"), spherical ends were machined on each. The diameters between the ends were turned down a bit at a time after inching the rods out of the lathe collet, but to remove the last quarter inch or so, the part had to be flipped around. A couple layers of shrink tubing over the already turned-down areas provided remarkably low TIR (.002") surfaces for the collet to grip while removing the remaining material.

It came as no surprise that I'd have to make yet another degree wheel. I've not been able to come up with a 'one fits all' design, and I have almost as many degree wheels laying around as I have completed engines. The Knucklehead's version is attached to the engine using a machined Delrin center section that's finger-pressed into a recess in the flywheel.

While thinking through the process of timing the camshaft, it occurred to me that a screwdriver slot in the end of the camshaft might be useful for fine tuning the the cam in its adjustable gear. The slot would have been a lot easier to machine before the cam was hardened, though. After rubbing the teeth off a HSS slitting wheel, I remembered about the heat treatment and set up a tiny abrasive wheel in the mill to grind the slot. After completing it, I really didn't find it all that useful.

Camshaft timing began by rotating the flywheel until its TDC mark was adjacent to the 'F' mark on the crankcase. As explained much earlier, the alignment of these two marks indicates the front cylinder is at TDC. While standing on the gearbox side of the engine with the degree wheel attached to the flywheel, a stationary pointer was set up adjacent to a convenient angle on the degree wheel whose value was recorded. The goal was to install the camshaft so the center of the front intake lobe occurred 107.5 degrees after the front cylinder's TDC. As mentioned earlier, this 107.5 degrees is the cam's centerline angle. This relationship with the cam's centerline insures fuel will be sucked into the front cylinder during the piston's downstroke. From the gear box side of the engine, the crankshaft rotates CW when the engine is running. The cam must be installed so that when the degree wheel is rotated 107.5 degrees CW from its recorded position, the front intake lifter is sitting on the center of its lobe.

On this particular engine, determining the center of the intake lobe requires only the front intake lifter (no pushrod) to be installed. With its lash adjuster temporarily replaced with a machined-flat screw for use as a DTI measuring surface, a very sensitive indication of the lobe's center is available. The cam's position was quickly determined to within a single gear tooth, but this was only good to some 30 crankshaft degrees. Using the slotted gear to resolve it further was much more tedious and seemingly chaotic, but I eventually managed to land on 108 degrees.

Using .004" lobe lift points, measurements using the same DTI setup showed the front intake opening 4 deg BTDC and closing 30 deg ABDC compared with the 5 deg and 40 deg target values. Similar measurements on the rear intake yielded 4 deg ATDC (not BTDC) and 39 deg ABDC.

Measurements on the front exhaust lifter showed it opening 65 deg BBDC and closing 4 deg ATDC compared with target values of 65 deg and 5 deg. The angle between the front and rear intake lobe centers was measured to be 322 degrees compared with its target value of 315 degrees.

Using half the distance between the .004" lift points to define the lobe centerlines gave a slightly different result, but it wasn't at all clear it was any better than measuring the lobe peaks. After making my peace with the measurements, the five mounting screws in the cam's slotted gear were torqued down for the final time.

The crankshaft gear is fixed to the crankshaft using a pair of dowel pins whose holes were cross-drilled in such a manner to insure the two can be assembled in only one way. With the timing completed, witness marks added to the crank and cam gears will now indicate the cam's proper installation during final assembly.

I thought I'd now be ready to machine the last parts needed for final assembly - the piston rings. However, after machining the pushrods, I ran into a problem I should have realized earlier: it's not possible to install a pushrod simultaneously with its cover without significant engine disassembly. Either can be installed separately but not together. My cover design needs to be changed so the lengths can be shortened another 3/8" while they're being installed with a pushrod inside them.- Terry

 

[Scott_M]

Hi Terry
Bummer on the push rod tubes !
Question ?
If you have not made the rings yet then I am assuming the whole top end needs to be disassembled to install them. Is that not sufficient to get the rods and tubes in place ?
Or is it because "it is just not right " Which I totally understand
Or will you not have enough room to set valve lash with the pushrod tubes fully collapsed ?

Scott

 

[mayhugh1]

Scott,
The top end will have to come apart to install the rings, but I just don't like the idea of the pushrods getting captured during the reassembly. Like you say, it's just not right. - Terry

 

[Scott_M]

"It's just not right" I kind of thought that was why

And I forgot to mention, as always , beautiful work !

Scott

 

[rodue]

I was studying you plans on the carburetor and I didn't see the float, did you do away with it. The float is a problem to make it light enough to float in fuel. I am eager to see your engine running. Rodue

 

[mayhugh1]

I was studying you plans on the carburetor and I didn't see the float, did you do away with it. The float is a problem to make it light enough to float in fuel. I am eager to see your engine running. Rodue

Instead of a float, I'm using an electric fuel pump located inside the fuel tank to drive a recirculating loop that will maintain a constant level of fuel in the carburetor bowl. - Terry

 

[michelko]

Oh man... this is realy a piece of art!!! Awesome work.

Michael

 

[mayhugh1]

Salvaging my original pushrod covers seemed hopeless, and so I spent a few days looking at totally different cover designs. These all wound up as dead ends, and so I took a few days off to reset my thinking. Eventually, I returned to my original design with some modifications that looked promising even though I still couldn't reuse any of the existing parts.

The new upper covers are similar to the old ones including the Kynar oil seals. Of the three parts making up a complete assembly, the top halves are still the more interesting pieces and relatively straightforward to machine. The lower halves should have been trivial to make, but I made things difficult by making their o.d.'s as small as possible. After some trial and error, I found I could safely reduce their wall thicknesses to as little as .010" over the roots of their internal threads. I initially tap-threaded a few parts after profiling their o.d.'s, but manual threading invariably caused the finished parts to swell as much as .003". This created clearance problems for the close-fitting center covers that will have to freely slide over them.

The solution, of course, was to thread the lower covers before profiling their exteriors, but the resulting thin-wall parts with their unfortunate shapes had too much stick-out and created severe chatter and surface finish issues. My goto solution for this problem is to pack the part with modeling clay, but this time it didn't help. The chatter was eventually squelched by threading a short piece of loose-fitting threaded steel rod inside the part to absorb the oscillations while its exterior was being turned.

The center cover is also an internally threaded thin-wall part. However, being turned from stainless steel, and having only a few internal threads, there was no noticeable o.d. distortion. The center and upper covers were permanently joined together with a grease-stick type thread locker that reduced the three part cover assemblies to two parts. This was done to make their installations a little easier, especially in the limited space around the intake lifters.

The machining steps for all three parts took some time to get just right, and in the process I wound up making several extra sets. I still don't feel comfortable about permanently joining the center and upper covers, but trial installations in the engine showed it provided some welcomed advantage. Before assembly, the aluminum-on-aluminum threads were dusted with powdered graphite for a butter smooth fit.

The larger lash adjustment windows available with the new covers no longer require the pushrods to be sized to their particular position in the guide blocks, but I plan to use the ones I've already made. To wrap up the pushrod/cover work, I also cut a pair of teflon gaskets to seal the guide blocks to the gear box. The photos show the new covers and some comparisons of them with the originals. - Terry

 

[mayhugh1]

I use George Trimble's method to make my piston rings. Although the multi-cylinder engines that I've been building have brought my to-date total up to some 200 rings, I seem to learn something new with every batch I make. The only change I've made to his process is to use a normalization temperature of 975F rather than his originally recommended 1475F.

As part of my own process, before any ring is installed on a piston, I check its fit in a cylinder using a 250 lumen flashlight. This test weeds out marginal rings before they wind up inside an engine. A ring that has any light leaking between it and its cylinder wall, other than through its running gap, is discarded.

My yields have generally been dismal because of circularity issues that typically show up during the blank's machining. I use class 40 gray cast iron from a number of sources that's been laying around in my shop for years. The errors have been unpredictable, seemingly uncontrollable, and sometimes as high as eight tenths over portions of the finally finished blanks. Typically, less than half of any blank will pass my acceptance criteria of two tenths, and occasionally an entire blank will be scrapped. Once the rings have been parted off, they're nearly impossible to evaluate until fully finished and light tested. After slicing candidates from the well-behaved portions of the blanks, my yields for the rest of the ring-making process are typically 80%-90%.

To start a large batch of rings, I usually prepare several blanks from different sources. Since I needed only four rings plus a few spares for the Knucklehead, I started a single blank from material that I'd used before. The Knucklehead's rings require a 1.0632" final o.d., and so I started with 1-1/4" diameter raw material. Cast iron rounds are manufactured oversize, and even though I might have gotten away with a smaller starting diameter, I'm leery of the castings' outer 1/8" or so.

A 4" drop was turned down to 1.109" so it could be held in a collet during the blank's machining. The blank was drilled through .625" and a portion of one end bored out to the rings' final i.d. This left a quarter inch wall thickness for work-holding and about .050" of o.d. stock for later removal. The blank was then heated to 700F for three hours and allowed to slowly cool overnight in hopes of removing residual stresses. (I began including this heat soak step a few years ago to combat an occasional issue I noticed with the blanks continuing to change dimensions days after their machining had been completed.) Two days later, the blank's o.d. was finished and polished to its final diameter plus two tenths using 800 grit paper. During its machining, quadrature measurements of the blank's diameter were continually recorded along every half inch of its length.

To my surprise, the errors never exceeded a tenth or so. This totally unexpected outcome may have been the result of finishing the blank's i.d. before the heat soak. In the past, I've left equal amounts of i.d. and o.d. stock on the blank for removal afterward.

Rather than waste the material, I decided to finish all 24 rings that I was able to slice from the blank using a .019" wide carbide parting insert. Parting tends to raise burrs on the corners of the i.d.'s, and these were broken using a 1/4" diameter ceramic file. The rings were parted a thousandth over the width of the ring groove. Using 600 grit grinding grease on a glass plate and a simple fixture to hold onto the rings, both sides were equally lapped to obtain a .001" ring groove clearance. During combustion, one of these surfaces will end up sealing against the lower wall of the piston groove and provide an important component of the combustion chamber's overall seal.

The Trimble article emphasizes the need for a straight radial break in each ring to properly contact with the spreader dowel, and a shop-made cleaver is recommended. Although 'good enough' results might be obtained by simply snapping the rings, I constructed a cleaver several years ago. After lapping, each ring was cleaved in preparation for heat treatment.

The Trimble articles also describe the construction of the fixture required to support the rings during their heat treatment. Equations were provided for the dimensions of a mandrel and a spreader dowel that are its key components. The fixture isn't difficult to make, but its dimensions are specific to a particular ring diameter, and this one will wind up in a drawer along side the other three I've made.

Although I've typically turned these fixtures from free machining alloys such as 12L14 or 303 stainless, this time I used plain hot-rolled steel. I always seal the fixture'd rings in an argon-filled stainless bag for protection during heat treatment, but the contents invariably wind up covered in a mysterious deposit. Although it isn't difficult to remove, it's an annoying extra step that I've begun suspecting may be related to the sulphur and lead that are alloyed into the free-machining steels I've been using. - Terry

 

[petertha]

Thanks for summarizing everything Terry. The sketch & worked through example was particularly useful as I see I misinterpreted one of the clearance parameters from your 18-cyl build.

BTW all, OLM now sells Durabar cast iron.
https://www.onlinemetals.com/en/search/results?text=cast+iron
I'm not sure if its because this CI its a recent add or their new website teething pains but I see round solid rod available as typical hobbyist cut lengths (ie stock for liners & rings). But some of the other shapes only appear as stock (long) lengths, so might have to inquire on that stock.

 

[petertha]

Terry can you elaborate on these insert blocks on the side of the slot. I thought maybe something like adjusting gib sliders but I don't see a set screw or anything.

 

[petertha]

One parameter I didn't see was running gap. If I understand, you make a cleaved split in the ring (call that a zero width kerf). Then the heat set over the calculated dowel diameter. Then you remove some additional amount from the open ends that would yield the running gap once compressed in the bore? If so how much & is that also a formula based on bore or something?

 

[mayhugh1]

Terry can you elaborate on these insert blocks on the side of the slot. I thought maybe something like adjusting gib sliders but I don't see a set screw or anything.

Petertha,
They're just filler shims used to precisely align the cutting tips of the two HSS blades.

Terry

 

[mayhugh1]

One parameter I didn't see was running gap. If I understand, you make a cleaved split in the ring (call that a zero width kerf). Then the heat set over the calculated dowel diameter. Then you remove some additional amount from the open ends that would yield the running gap once compressed in the bore? If so how much & is that also a formula based on bore or something?

Peter,
The running gap must be put in after the heat treatment. I generally use .004", which is more than enough and not at all critical. It's just there to make sure the ring doesn't close up and bind in the cylinder due to temperature expansion. I did the calculations for a previous build and found, contrary to popular opinion, it isn't a noticeable leak. - Terry

 

[Rustkolector]

Terry,
Thanks for your comments and details on the Trimble ring making procedure. I have carefully built the same ring cleaver that you use and I consider the ring breaks created as very straight radial breaks, but only at the surfaces contacted by the knife points. Between the knife contact points the break surface is rather jagged, and occasionally has a few protruding points. Enough so that it looks like the break surface needs a few strokes with a fine file (which I give them) to give a more even and square contact with the fixture dowel. What do your cleaved surfaces look like?
Jeff