Another Knucklehead Build

[mayhugh1]

The full-size Knucklehead uses three caged roller bearings on the big ends of its rods (the rollers ride on the hardened crankpin) and ordinary bronze bearings on the small ends. The widths of the outer cages are narrower than the center one, and so the rods' big ends wind up with equal bearing contact areas.

The Draw-Tech version uses four identical ball bearings on its crankpin and a single caged roller bearing on the small end of each rod. (The rollers will ride on hardened wrist pins.) The widths of the ball bearings are such that the big end of each rod is supported by its own pair of bearings, but because they're not rollers their effective contact areas will be much less than what is provided by the rollers on the small ends.

The SCE-45 roller bearings that I received from McMaster-Carr measured .4395" in diameter - a full two thousandths over its recommended bore. The thin metal shell on this type of bearing can sometimes be out of round and difficult to measure until it's pressed into place, but my measurements showed these were perfectly round. In addition, gage pins showed their shaft diameter wanted to be .252" instead of .250". I couldn't see how the bearing's rolled-over steel edges were going to allow them to be compressed a full two thousandths during the pressing operation, and so I performed a test.

I drilled and reamed a .4375" test hole in a piece of 3/4" thick scrap aluminum. An equivalent of the recommended installation tool was used to install one of the bearings and it was pressed in using the recommended outer edge of its lettered rim. The bearing went in, but it required a breaker bar on my one ton arbor press. A .250" gage pin was still a loose fit in the installed bearing, but a .251" pin fit reasonably well which meant the shell had indeed been compressed some.

I was afraid that a .002" interference fit might be damaging to the small ends of the aluminum pushrods after they had accumulated a number of hours of machining time. A cock-eyed installation due to galling would be a problem. A roller bearing is probably less tolerant of rod and piston machining errors since issues with twist that would otherwise 'wear in' a bronze bearing might break a roller. Bronze bearings were certainly an option, but I decided to continue on with the rollers.

My plan for machining the rods was to cut them concurrently from a single piece of 7075. Their locations inside the workpiece would be equally distant between the top and bottom surfaces of the workpiece. The first operation was to drill and ream the four bearing holes completely through the workpiece.

I didn't have a suitable reamer for the oversize rollers, and so I used a barrel lap and Timesaver lapping compound to increase the bores from .4375" to .4390" for a .0005" interference fit which was all I was willing to chance. Since I had to lap the small end bores, I reamed the big ends a thousandth under and lapped them as well. Timesaver cuts aluminum fairly quickly, and so care had to be taken to not pass up the target. It won't embed and cause problems later, but in this application there won't be any relative motion between the fitted surfaces, anyway.

After drilling, reaming, and lapping the holes for the bearings, a trough was machined through the topside of the workpiece and down to within .100" of its bottom to define the perimeters of the rods. This allowed the rods to be completely machined except for their bottom faces which remained attached to the workpiece.

When the topside machining was completed, the troughs were filled with Devcon 5-Minute epoxy and allowed to cure overnight. This epoxy kept the rods within the troughs attached to the workpiece so it could be flipped over and the rods machined free of it. Machining the entire peripheries of the rods from the topside avoided a seam between the top and bottom side operations. The finished rods were finally removed from the workpiece by heating it to 300F and pushing them out free while still hot.

The last operation removed stock from between the tines of the fork rod in order to provide clearance for the knife rod. The bores were temporarily plugged and then the rods bead blasted to make their surfaces appear similar to those on the full-size rods. Finally, the bearings were pressed in with what felt like just the right amount of force, and the crankpins will be turned to .2515". - Terry

 

[natalefr]

AWESOME !

 

[kvom]

What's the reason for the cross slots around the bearings?

I'm currently making some parts using this Devcon epoxy, a technique I discovered from you on the radial build. How do you get the epoxy into just the milled slots and not on the part face itself. Also, the color of the epoxy that I got from Lowes is a pearl color when mixed, not like that in your photos.

 

[mayhugh1]

What's the reason for the cross slots around the bearings?

I'm currently making some parts using this Devcon epoxy, a technique I discovered from you on the radial build. How do you get the epoxy into just the milled slots and not on the part face itself. Also, the color of the epoxy that I got from Lowes is a pearl color when mixed, not like that in your photos.

The slots are to provide an easy path for oil flow into the bearings which are open style bearings.
I use plastic 1oml or 50 ml syringes to direct the epoxy. You can buy them by the case cheaply online.
There are two kinds of Devcon 5 minute epoxy. One is a gel which doesn't flow out as much as the regular stuff. It is a pearl color when mixed. I've used it almost exclusively in the past when I didn't have well defined troughs to fill since it was a little more efficient to use and tended to stay put a little better than the regular stuff which I used above. The regular stuff is a little more clear when mixed. It flows more easily and is more convenient to use when you have deep troughs because it will flow out and fill them more easily than the gel. I used to machine my parts half-way through each side of their workpiece and this would leave a bit of a seam that had to be cleaned up later where the two operations met. This style of machining didn't involve deep troughs and was more appropriate for the gel. After learning I could get rid of the seams by milling the entire part from one side, Instarted doing that, but I now had deep trough($) that had to be filled and for this I used the non-gel.

I think I started noticing, a year or two ago, a change in their formulations because I use to get the parts to release quickly and easily with a heat gun. Now I put them in my shop oven at 300F and, wearing oven gloves, push the parts free while they're hot. This might also be due to the fact that I've been cleaning the parts better than I used to. The non-gel stuff has less strength than the gel and so when I began using it it I started being more careful about cleaning the parts before applying it. I use Simple Green and then rinse with water and then use dish soap and hot water to clean the cutting fluids from the parts before applying the epoxy. I became anal about the cleaning after once having a close call where a part pushed free of its workpiece before heating. - Terry

 

[kvom]

My current experiment went a bit south. Not having the syringe, I just covered the entire part. Cured overnight, and then face milled the bottom to reveal the part. After half an hour at 400F the epoxy had turned orange like your photo, and while I was able to remove the part, there was still epoxy clinging to the part (Mic6 Al) that resisted coming off (tried to slice it off with a box cutter - fail). I put the part back in the oven a 500F for half an hour, which resulted in the epoxy turning black and adhering even worse. Finally tried to burn it off with a torch, which got most of it off, but still spots that resist light sanding or scotch brite. Probably bead blast would be best, but I haven't the gear. Parts are usable but not shiny.

I'll order the syringes for the next try.

 

[mayhugh1]

I've never heated the stuff over 300F and so you might have baked it onto your parts. You do need to push the parts free while still hot right out of the oven. I use 'Ove Glove' oven mitts to handle the hot parts. Slopping the epoxy over the part shouldn't be a problem as the epoxy can be rubbed off the surfaces where you don't want it. If it's still sticking to your part, put it back in the oven for another ten or fifteen minutes but at only 300F, and it should rub right off.
The connecting rods above pushed right out on the first try, but there was a bit of epoxy sticking to the sides of one of the rods. While still hot I rubbed most of it off with a wood popsicle stick. There was still a bit that wouldn't come off, and so I put it back in the 300F oven for a short while and, on the second try, it rubbed right off with my gloved hand. - Terry

 

[kvom]

Thanks. I ordered syringes and will remake the parts. This is a valuable technique to master.

 

[johnmcc69]

Great techniques Terry, & a wonderful thread.

John

 

[mayhugh1]

After making pistons by the handfuls for my other engines, it didn't feel right to make only a couple for the Knucklehead. So, I machined two for a stock 5.3 compression ratio and a second set for a slightly higher 6.5 c.r. The higher compression pistons will probably need to be notched for valve clearance, but that'll come later after the camshaft lift is determined.

If I've correctly interpreted the Draw-Tech piston drawing, it shows six equally spaced holes drilled radially through each of the three ring grooves. Many pistons have oil relief holes in their oil ring groove (along with some kind of path through the oil ring itself), but I don't understand the reason for drilling holes through the compression ring grooves.

The intent may have been to use combustion gases to pump oil into the connecting rod's small end bearing, but holes in the top groove will also leak compression and reduce the forces sealing the ring to the cylinder wall during the power stroke. And, without some kind of path through the bottom ring itself, it's not clear that holes in the bottom groove will do anything. I wasn't able to determine the rings' radial back clearances from the drawing, but the axial groove clearances specified as nominally zero also confused me. I may have misinterpreted the original design intent, but since I didn't understand it, I opted to redesign the pistons.

I started by reducing their o.d. by .003" except for the short portion inside the cylinder above the top ring which was reduced by another .003" to allow for temperature expansion. The number of rings was reduced from three to two, and an oil collection groove was added immediately below the second ring. Relief holes drilled through this groove will direct oil scraped by the lower ring into the interior of the piston. The essentially zero clearance around the connecting rod in the original design was increased to .040" on each side of the rod. The photos show a few other cosmetic changes intended to make the pistons look a little more like the full-scale versions, but it turns out there are quite a variety of them.

A couple other clearances will define the design of the rings when they are machined later. These include a .001" axial groove clearance as well as a .006" radial back clearance behind them.

The photos include CAD models of both the 'high' and low compression pistons. A cross-sectional drawing that includes the head's hemispherical combustion chamber shows a comparison of the piston dome heights inside the head at TDC.

The piston blanks were prepared by turning their finished o.d.'s to .003" less than the inside diameter of the cylinders. This is the clearance used in my radial engines, but since the heads in this engine won't be as effectively cooled (no prop wash), the piston diameter inside the cylinder above the top ring was reduced an additional .003".

With a piston blank installed in a collet block, the hole for the wrist pin was drilled and reamed before its interior was machined for the connecting rod. The blanks were then moved to the lathe where their domed tops were turned and the ring grooves cut. A chatter-free finish on the lower wall of the ring grooves is especially important because it's a sealing surface for the compression ring(s) during the power stroke. The oil relief holes will be drilled later when the mill's fourth axis is set up to machine the cam box gears.

The wrist pins, which are press-fits in the pistons, were turned from O-1 drill rod. One end of each pin (the heat darkened end in the photos) was finished a couple tenths oversize to create the interference for the fit. Before heat treatment, their ends were drilled and reamed for soft round-head aluminum rivets that will protect the cylinders from the edges of the hardened pins in case they move inside the piston. The wrist pins were oil quenched after being heated to 1475F and tempered at 350F prior to pressing in the rivets. Heat treatment was needed because the roller bearings in the small ends of the connecting rods will ride on them. - Terry

 

[e.picler]

Wonderful work!

As always your machining quality is amazing, as well as the photos and CAD details.
Congratulations!

Edi

 

[mayhugh1]

The pushrod assemblies are similar to those in the full-size engine and, with about a dozen parts each, they're fairly complex. I'm currently modeling their parts so I can build a virtual assembly and better understand the order in which they'll need to be assembled and also how lash adjustments were intended be made.

The guide blocks are the cornerstones in these assemblies. They couple the roller lifters inside the cam box to the pushrods which in turn operate the rocker arm assemblies inside the rocker boxes. The head assemblies have already been completed including the double-angled rocker access holes drilled through the bottoms of the rocker boxes.

The pushrod cover tubes are three-piece assemblies. An upper tube is expected to slide into the bottom of the rocker box, and a lower tube slides into the top of the guide block. A third tube covers the gap between them. Ideally, in each assembly, the axes of all three tubes will align with the holes in the guide block and rocker box and provide the necessary clearance around the pushrod to accommodate its changing axis while operating the rocker arm. The compound angles for each pushrod assembly are different, and poor fit-ups with the rocker boxes will create oil leaks.

The guide blocks are among the Knucklehead's most difficult parts to machine because of the precision intended for the angles of the guide holes that must be drilled through them. Errors of just a fraction of a degree will affect the fits of the parts within the assembly and create oil leaks. Many of the already completed parts affect the locations and/or orientations of the rocker access holes, and machining errors in them will also affect these alignments.

The axes of the guide block through-holes are provided in the drawings with a pair of angles (to two decimal places) and a touch-off point referenced on the bottom surface of the block. An error in a guide block angle of just a tenth of a degree will cause the cover tube assembly to miss its target in the rocker box by .005". Realistically, an error of half a degree can probably be tolerated before pushrod clearance becomes an issue. Cosmetically, the center cover tube is capable of hiding a minor mismatch between the axes of the upper and lower tubes.

Another consideration for the guide block through-holes involves the camshaft. The lobes on the stock camshaft are only .080" wide (the rollers are .064" wide), and the cam will be located approximately 1-1/2" below the guide block. This means that a 0.1 degree error in a guide block angle will move its roller contact .003" across its lobe. (Another set of angles coming up later will define the orientations of the rollers so they ride flat on the cam lobes.) Some errors can be accommodated by moving the locations of the lobes along the camshaft as needed to keep the rollers centered. However, the lobes have to remain sufficiently separated to avoid interference between their roller guides - especially those in opposing guide blocks.

A CAD model with a couple views from inside the cam box shows the relationships among the camshaft and the guide blocks using pairs of pointed test rods inserted through the guide blocks and contacting the camshaft. The guide blocks were modeled using the dimensions provided in their drawings, but they sit on top of the model of my slightly enlarged cam box. The camshaft lobes are, for now, just simple circular disks, but they are spaced as defined in the camshaft drawing. The model shows the center distances between the test rods being spot on the centers of the lobes.

The guide blocks were machined from 7075 aluminum. Each of the holes had to be spotted, drilled, and reamed in separate operations using setups that allowed the x and y tilts of the blocks to be adjusted. I used an electronic angle gauge to roughly set each axis, but the final settings had to be tuned with a z-axis dial indicator in conjunction with the mill's DROs. The angle gage was barely able to handle even a single degree of accuracy. Readings and resolution in one axis were unpredictably affected by the angle of its other axis, and so the angles had to be independently measured and adjusted.

I spent a lot of time carefully tweaking each setup, but in the end I could do no better than a tenth or two of an indicated degree. The angles invariably and unpredictably changed when the various lockdowns were tightened, and the setups generally lacked rigidity. With drill starting and wandering added into the mix, I doubt the holes ended up within a half degree of where they were supposed to be.

My biggest frustration with the setups, though, was with what seemed like never-ending confusion about which direction to tilt the axes. Working from the bottoms of the parts and with such small angles made sanity checks difficult. I had everything ready to go more than once only to discover that the entire setup was backwards. One time, after completely reversing it, I realized everything had been correct the first time. Anyone watching my clown show couldn't have helped but laugh hysterically. After nearly a full week, I had four scrapped guide blocks and a single pair of candidate block to show for my efforts.

With the finished guide blocks sitting flat on the top surfaces of the cam box, I checked them using test rods that were closely fit into the blocks but had top ends that were turned down by .010" for trial fitting into the rocker boxes. The errors were a bit too large for both test rods to enter the front rocker box, but I was able to get both into the rear rocker box.

Both guide blocks had to be moved a mysterious .045" away from the inside edge of the cam box to achieve the best overall fits of their test rods inside the rocker boxes. The eighth inch by which I radially increased the periphery of the cam box to accommodate its o-ring accounted for only .005" of this. Since both blocks were identically affected, I initially thought the difference was due to some other change that I made. The CAD models don't support this, though, and my assembly model shows the guide blocks properly located .005" from the edge of cam box. My suspicion is that there was a systematic error related to the touch-off points on the bottoms of the tilted blocks.

One of the CAD drawings shows a cam box gouge that would have been required for clearance of the roller guide in the rear intake position. This is a 'feature' of the stock design, and my CAD model predicts it should vanish as a fortuitous result of the .045" translation.

The next step would have been to transfer the 'best fit' guide block mounting hole locations to the cam box. With all the work already put into the cam box, I wanted to make sure there wasn't going to be an issue with the camshaft. I machined a test fixture consisting of a duplicate of the top-end portion of the cam box along with a dummy camshaft. Transferring the mounting hole locations to the cam box CAD model automatically transferred them to the fixture, and they became part of its machining. I added generous clearances around the expected locations of the test rods so I could duplicate the virtual camshaft test done earlier. The resulting measured spacing between the centers of the rods at their points of contact on the dummy camshaft were within .010" of their theoretical values. Their absolute distances with respect to the edge of the cam box were, on average, offset by the same .045". Fortunately, I could see no issues that couldn't be accommodated with minor camshaft modifications.

Finally, I drilled the mounting holes in the cam box and installed the guide blocks. A full length test rod was installed in one hole in each block, and a pair of half-length rods in the other. I was able to verify that a cover with an i.d. that was just .010" over the test rods' o.d.'s would slide over the pair and hide the mismatch.

The next step is to complete the modeling of the rest of the pushrod-related components and begin work on them. - Terry

 

[mayhugh1]

Two pairs of guide holes matching those in the guide blocks had to be drilled through the roof of the cam box. The setups weren't pretty, but they were functional and provided the reamer clearance needed under the spindle. Verifying these setups before drilling was much less confusing that those associated with the guide blocks since the blocks themselves were used as clamping guides.

After aligning the fixture plate with the mill's x-axis, its tilt along the y-axis was the only angle that had to be carefully adjusted for each hole. This angle was set equal but opposite to the one being matched in the guide block. With the guide block installed on the cam box, one end of a close fitting length of drill rod was inserted into it and the other end chucked in the spindle. The guide angle along the mill's x-axis was automatically established by the guide block, and the table needed only to be brought forward so the fixture plate could sit flush against the rear of the cam box. After clamping the two together, the guide block was removed so the hole could be spotted and drilled. The block was then replaced and a reamer run through the combination. This process was repeated three more times for the remaining holes.

When the guide hole drilling was completed, the pointed test rods used in the camshaft checking fixture were moved to the cam box. Eyeballed measurements showed the points of contact on a short length of drill rod sitting in the camshaft's approximate location looked as expected. The cam box ended up slightly gouged, after all, by the clearance hole drilled for the rear intake guide.

I needed a break from the pushrod assemblies, and so I tied up a few loose ends on the cam box. The dipstick, a magnetic drain plug, and the mounting locations for them were machined next. Both parts are o-ring'd. The drain plug was machined from 303 stainless and the dipstick's head from aluminum. The stick itself is a length of drill rod that was flame-darkened to make reading the oil level a little easier. - Terry

 

[mayhugh1]

While looking for an excuse to put off working on the pushrod covers for a couple more weeks, I happened to recall the drawing for the Knucklehead's oil pump. Even though it's just another gear pump, the uncommon tooth profile of its gears caught my interest.

A gear pump works by moving fluid through it in the spaces between the roots of its gear teeth and the inside walls of its close fitting housing. Pump gears have few numbers of teeth in order to maximize the volumes of these spaces. For spur gears this usually ends up being between 12 and 14 teeth.

Spur gears are widely used in oil pumps. Their involute profiles keep two or more pairs of teeth meshed at all times. This is helpful for power transfer but, in pump applications, pockets of incompressible fluid become trapped in the spaces between these teeth. The volumes of these trapped spaces continually vary as the gears rotate, and if the pump isn't totally submerged, air can be drawn into a space whose volume is decreasing.

The stresses created while trying to compress an incompressible fluid not only put strain on the pump, but they also rob power from the engine. Commercial pumps use escape grooves that direct this otherwise trapped oil into the pumps' bearings. Model engine pumps often don't include this complication since in practice they usually aren't all that well sealed.

The continuous contact tooth profile used in the Draw-Tech pump is an alternative to the involute profile. It theoretically eliminates any trapped oil and, with even fewer teeth than available from a spur gear, the transport spaces can be larger. Any real performance advantage will be affected by the accuracy of the gear machining which, unfortunately, looks like a nightmare on a manual mill.

I've included a couple models comparing the two profiles in pumps with similar size housings. The involute gears are typical 14t/24P pump gears with .667 inch o.d.'s. The Draw-Tech gears are .625 inches in diameter. The line of continuous flank contact on the Draw-Tech gears traces out a figure eight pattern and prevents oil from being trapped between the teeth. In comparison with the spur gear pump, its transport spaces are also larger.

Although mostly for its novelty, I thought I'd give the new gear profile a try. I started by turning a pair of test blanks from 932 bronze. The o.d's of their shafts and gear bodies were fully finished on the lathe, and the tiny corner fillets left behind by the lathe tooling were removed using the edge of a sharp parting tool. For the tooth profile I used the exact dimensions provided in the Knucklehead drawing. Before starting, I cut a few test slots with the end mill that I planned to use so I could account for its exact cutting diameter.

After machining the test gears, I drilled/reamed several sets of shaft holes in a test block so I could experimentally determine a proper spacing for them. I found I was able to decrease the .480" center-to-center spacing shown on the pump body drawing to .468", and still have some .004" safety margin. The gear set turned very smoothly with no tight spots even with the reduced spacing. Finally, I machined an identical set but with the proper end shafts for use in the pump housing. They behaved identically to the test gears in the test block. - Terry

 

[sbdtasos]

Hello Terry
excellent work as always
what is the diameter of the gears?
thanks
Tasos

 

[mayhugh1]

Hello Terry
excellent work as always
what is the diameter of the gears?
thanks
Tasos

Tasos,
Their o.d. is .625"
Terry

 

[mayhugh1]

Just a few more parts were needed to finish up the oil pump. In addition to the already finished pump gears, a housing and rear cover plate were needed to fully enclose them. The gear shafts pass through the front and rear faces of the pump, and so it's important that their axes remain normal to the pump's body after assembly to avoid binding the gears inside their close fitting recesses. In the original design, this alignment would be affected by the clearances around the four mounting screws that pass through the body and cover when the pump is secured to the cam box. These clearances would also affect the location of the end of the drive gear's shaft which is intended to slide into a flanged ball bearing that will be pressed into the cam box cover.

I enlarged the housing dimensioned in the drawing in order to move the mounting screws away from the edges of the gear recesses, and I also integrated the elbow for the pick-up tube into the body. But, more importantly, I added a pair of dowel pins to positively locate the rear cover to the pump body.

Construction began by facing two pieces of 360 brass to the finished thicknesses required for the pump's body and cover. Clearance holes for the pump's mounting screws were individually drilled through each workpiece. The identical hole pattern was then tapped into a sacrificial aluminum block so the two workpieces could be stacked and simultaneously machined. After drilling and reaming holes through the entire stack for a pair of dowel pins, they were pressed into place for the remainder of the machining.

With the pump's body and cover doweled together, the holes for the gear shafts were drilled and reamed through the entire stack. The peripheries of the housing and its cover were then machined together. A hole through the side of the pump, required for the inlet and outlet ports, was drilled in a separate setup before machining the gear recesses.

In order to machine the gear recesses, the cover was removed, and the center of each shaft hole was individually indicated. The diameters of the recesses were machined .003" over the o.d.'s of the gears, and each recess was verified for a proper fit with its gear while still set up in the mill. Finally, the pump body was flipped over so a thrust surface could be machined on the front of the pump for a drive gear.

An oil pick-up was fashioned from 3/16" soft drawn copper tubing and soldered to the elbow on the pump housing. Anyone familiar with the drawings might notice that I've reversed the positions of the pump's inlet and outlet. This was done since I plan to replace the Draw-Tech 'Twingle' cam with a more conventional Harley grind. In the process, the direction of the engine will be reversed so it will become the same as that of the full-size engine. As a bonus, the pump's output line won't have to be routed through the crowded area around the distributor.

Since I wanted to also add the oil pump's drive gear, I set the mill up to begin machining the engine's spur gears. I morphed the pair of short crankshaft drive gears in the original drawing to a single long gear that I machined from Stressproof.

Both internal pump gears wind up supported by integral bearings in the front and rear faces of the pump housing, and the pump's drive gear is driven by the crankshaft through a 3:2 gear set. In the original design, the drive gear's extended shaft is also expected to slip into a ball bearing pressed into the cover of the cam box. In actuality this end bearing probably provides little positive benefit, but it would complicate the assembly of the cam box to its cover. After weighing the pros and cons, I decided to shortened the shaft and leave out its end bearing.

The cam box will come on and off the crankcase many times during the machining and trial fitting of its several components. I enlarged the through-hole in the cam box for the crankshaft in order to clear the crankshaft gear so it won't have to be constantly removed as well. I also changed the method for pinning the crankshaft gear to its shaft. Since there isn't room inside the cam box to deal with the tight-fitting spring pins recommended in the drawings, I used slip-fitting solid dowels instead which are very easy to remove and re-install. Their through-holes were centered between the gear teeth and, in usage while rotating, their mating gears prevent them from moving out of place. I've found this method to be effective and very convenient for maintenance any time I'm dealing with gears with even numbers of teeth. - Terry

 

[gbritnell]

Hi Terry,
Excellent work as always. I understand the reasoning behind the shape of your oil pump gear teeth and the oil being trapped at the clearance area at the root of the teeth but my question is won't oil also be trapped in your version as the radius on the one tooth slowly becomes tangent to the adjacent tooth there will still be a small area that the oil gets trapped in before the teeth completely mesh?
Yours is much like a trochoidal pump only with external gears.
gbritnell

 

[mayhugh1]

George,
Thanks for your comments. I've included a photo of the one place in the gears' rotation where a compressible trap might look like it would occur, but it appears that a rear escape door opens up at the exact same time that the front door closes and so compression doesn't seem to occur. -Terry

 

[mayhugh1]

Tying up a few more loose ends allowed me to complete some of the engine's final assembly and continue procrastinating over the pushrod assemblies. With the fourth axis still set up under the Tormach, I drilled the oil escape holes through both sets of pistons. Each received fourteen .052" holes drilled radially through its oil groove.

A pair of TDC marks were also engraved on the rear half (flywheel side) of the crankcase. These, in conjunction with an earlier engraved TDC mark on the flywheel, will be used to time the camshaft to the crankshaft and to adjust the valves. These two crankcase marks, "F" and "R", are spaced 45 crankshaft degrees apart which, non-coincidentally, is the angle between the cylinders.

With the timing marks and crankshaft gear completed, the rear oil seal was installed, and the crankcase finally assembled with an o-ring between its halves. The nuts on either end of the tapered crankpin were Loctited and torqued to 60 in-lbs. The final crankshaft TIR, .0005" measured at the cam box cover bearing, improved slightly compared with earlier measurements. Friction added by the oil seal, though, took away the free-wheeling crankshaft that I enjoyed showing off.

With the crankshaft at a point in its rotation where the front piston was at TDC, the flywheel was installed on the crankshaft with its TDC mark aligned with the "F" mark on the crankcase. The camshaft will later be adjusted to define this point as the top of the front cylinder's power stroke. Since a V-twin's TDC can seem rather broad and difficult to precisely locate, I pinpointed it using a dial indicator on the piston since the heads weren't yet installed. For reference, while standing on the flywheel side of the engine with the front cylinder to my left, this engine will run counterclockwise which is in the same direction as the full-size engine.

If the crankshaft is then rotated 315 degrees counterclockwise (360 degrees - 45 degrees), the TDC mark on the flywheel will become aligned with the "R" mark on the crankcase. This is the angle at which the rear cylinder is at TDC in its power stroke. In order to complete a full crankshaft cycle it must be rotated another 405 degrees (360 + 45 degrees). At this angle the flywheel TDC mark will once more be aligned with the "F" mark, and the front cylinder will again be at TDC in its power stroke.

The asymmetrical plug firing angles in the crankshaft's 720 degree cycle, 315 degrees and 405 degrees, are what give the engine its characteristic sound - a sound that Harley tried unsuccessfully for six years to patent. A small annoyance is that the marks wound up on the opposite side of the engine from where all the adjustments will eventually be performed, but this was also true of the full-scale Knucklehead's flywheel marks.

A timing mark previously engraved on the flywheel 20 degrees ahead of the TDC mark will be also used with the "F" mark to adjust the static timing advance in the distributor. Being only a starting point, 20 degrees was a guess on my part since that number seems to work well with most of the model engines that I've built. Although not relevant to the model, the spark advance in the full-scale engine was closer to 30 degrees.

The next task was to fabricate an output line for the oil pump. I was able to use one of the panel mount compression fittings machined earlier for a portion of a feedthrough that vertically exits the cam box near the dipstick. A new fitting had to be machined and soldered to the output side of the oil pump, however. Although there wasn't much space available for a formed 1/8" oil line with compression fittings on each end, I was eventually able to come up with something that looked at home inside the cam box. A piece of 1/8" aluminum tig rod was used for all the trial-and-error bends before the final shape was transferred to a length of soft drawn copper tubing. Copper tubing work hardens when bent, and over-bends can be difficult to correct because the tube will tend to bend back in a slightly different (softer) location and leave an unsightly wrinkle. I typically use a deep grooved oak dowel rod as a mandrel for bending small diameter copper tubing.

With 90% of the cam box machining completed, it was time to bead blast its exterior to match the crankcase. Already drilled holes were plugged, and the interior was sealed up by clamping the box between a couple pieces of wood. After finishing its surface, the cam box was thoroughly scrubbed with hot water and dish detergent to remove any remaining glass particles.

Finally, recesses for the flanged ball bearings for the camshaft and distributor gear reducer were machined into the rear wall of the cam box and its cover so the bearing pairs could be pressed into place. Test shafts were machined to verify the bearings in the box properly aligned with those in the cover. Since the cam box wall thickness exactly matches that of the cam bearing, the recess for its flange was machined a few thousandths shallow to maintain clearance for its inner race against the crankcase. The bearing for the distributor gear reducer was much thicker, and so per the cam box drawing its flange wasn't recessed at all. Additional bearings related to the starter drivetrain will be added later once they have been determined. Since the bearings are open style ball bearings, any further cam box machining will require protective tape over them.

While preparing to bore a hole through the top of the cam box for the distributor, I realized I was going to have an interference issue between the distributor and the crankcase oil return fitting located near the front corner of the front cylinder. This area is very busy, and the drawings show what appears to be an unrealistic tubing bend underneath the fins of the front cylinder that would have been required to route an oil line to this fitting. Rather than notch the cylinder for clearance, I earlier moved this fitting out from under the fins. Unfortunately, that change has placed it under the body of the distributor which would again require some heroic tubing work. To avoid this, I relocated the fitting to the rear of the cam box which turns out to be the same location used by Draw-Tech for the original oil pump's output line. As discussed earlier, my oil pump output line exits the other end of the engine.

After trial installing the heads, I've noticed some issues coming up related to the oil return lines in the vee between the cylinders. - Terry

 

[mayhugh1]

The intake manifold and the oil return lines for the inner valve boxes compete for space between the cylinders. There's potential for interference between them that I wanted to clear up before burying them behind the pushrod assemblies. My original plan was to use the threaded compression fittings that I machined much earlier for the other more accessible oil lines. However, they may be a bit large for the space available between the heads and, without having the actual intake manifold to play with, it's hard to tell if there really is a problem.

The manifold is yet another challenging part of this engine. It's a four-piece stainless steel tube assembly that requires fixturing for a precise fit up during soldering. It attaches, unforgivingly, to the flanged intake ports located on the flywheel side of the heads. It also supports a cantilevered carburetor that hangs over the cam box. One of my concerns is that the carburetor/manifold assembly might be plagued with a mechanical resonance that's triggered by the engine's vibration. If so, the strengths of the solder joints will be extra important. Since the manifold is in a prominent location on the engine, it also needs to look good and not leak.

The relative orientations of the heads' port flanges were established long ago by several machining operations including the tricky compound-angled ports that were drilled through the heads. My modifications to these ports which were Loctite'd in place included floating flanges with o-rings for positive seals to the manifold. I also reduced the i.d.'s of the plenum runners from .375" to .312" for increased air/fuel velocity. This allowed the use of thicker wall tubing in the manifold for additional solder strength. With the accumulation of so many tolerance'd machining steps, though, I wasn't at all confident that my CAD model would be reliable enough for anything but a starting point for the manifold's design.

I didn't feel it would be wise to use the engine itself as a soldering fixture because of the limited access around the heads and the high temperatures (1200F) involved with silver soldering. Although my approach may seem backwards, the first task was to create a soldering fixture.

The top-end of the engine was temporarily assembled with all gaskets in place so I could compare the measured angle between the port flanges on the heads with the angle in my model. I did this by trial-and-error bending a piece of 16 gauge sheet metal until it lay perfectly flat against the faces of both flanges. Remarkably, the angle measured within a degree or two of the one in my CAD model. Based upon the model, I designed and machined first passes of the three components that make up the manifold's plenum. The two flanged tubes were then bolted up to the heads so I could trial fit the critical third part, a short connecting tube with matching mitered ends. My model resulted in that part being several thousandths too short, but I was still able to J-B Weld the three pieces together in situ on the engine. A combination of epoxy and 'rebar' inside the joints resulted in a non-functioning mock-up that fit the heads perfectly.

To create the soldering fixture, I trial-and-error bent a piece of 1/8" hot rolled steel plate until the flanges on my mocked-up manifold lay perfectly flat against it. The mock-up's flange mounting holes were carefully transferred to the plate, and a hole was drilled through its center for clearance around the carburetor connecting tube. The mock-up was once more bolted to the engine so a dummy carb tube could be J-B Welded to it at the exact angle needed to insure the carburetor will be aligned to the engine.

At this point I had a mock-up of the manifold that fit the engine and a critical part of the soldering fixture that fit the mock-up. The fixture was then bolted to a sturdy baseplate to which an angled support for the carb connecting tube was added. A spring loaded ram, added to hold the assembly together during soldering, completed the fixture.

From my updated sketches, the components of the actual manifold were then machined and assembled in the fixture. After fluxing their surfaces, I slipped thin brazing strips (cadmium bearing) designed for carbide tool inserts (McMaster-Carr #7759A11) into the joints between the four parts. The brand name of the flux used was Ultra-Flux, but it was just the common white silver soldering flux available from many sources. The assembly was heated with a MAPP gas torch, and the throw-away fixture that I spent nearly a week developing had its five seconds of fame when it pushed and held the assembly together while the solder flowed out and cooled.

The cooled assembly was pickled in sulfuric acid (drain cleaner) to remove the residual flux and then neutralized in a baking soda/water solution. After bead blasting, I noticed one the joints was slightly misaligned, but it was filled with solder and looked very acceptable. I've run into this misalignment problem before. With the ribbon sticking out from around the joint, it's difficult to judge when the parts on either side of it are precisely aligned. The soldered manifold fit the engine perfectly, though, and the carb connecting tube wound up horizontal as I had hoped. Finally, a flange for the carburetor was machined and soft soldered to the end of the carb tube.

For a better looking finish and a bit more strength, I re-heated the assembly and 'buttered' the joints with low temperature (450F) Sn/Ag solder (TM Technologies ABS-0065). After a few hours of filing and sanding, the manifold was finally finished. The imperfections that I covered up were minor, but with all the hours put into the fixture, I felt like the manifold deserved a few of its own. - Terry