Another Knucklehead Build

[ddmckee54]

So the engine's not truly dead as a door-nail? It's just relaxing a little bit while recovering from its' injuries?

Don

 

[mayhugh1]

It's about to rise out of the ashes ...

 

[stragenmitsuko]

Like a fenix

 

[ddmckee54]

You gonna be running around yelling "It's alive, it's ALIVE", like Gene Wilder in 'Young Frankenstein'?

 

[IceFyre13th]

 

[mayhugh1]

The entire engine had to be disassembled in order to get access to the crankshaft. Its construction is very similar to that of a full-size crank, which is a pair of flywheels secured with heat treated nuts on either side of a tapered crankpin. The next step in a full-size rebuild would be to pull out the intelligent hammer and start beating on the crank until it was straight again. In my case, I was able to re-use my original alignment sleeve. In actuality, it wasn't as easy as it sounds because I had Loctite'd both nuts to the tapered pin. This was a big mistake because only one broke free greatly complicating dis-assembly. With nothing to work against, I wasn't able to remove the nut from the other end of the pin leaving me later with a nagging concern about whether I had sufficiently tighten both sides. I didn't want to apply heat to the nearly irreplaceable Stressproof tapered pin since weakening and breaking it would have been a real possibility. In any event, I ended up with the crankshaft's TIR back to nearly zero when measured at the crankcase bearings. The TIR at the very end of the crank's long skinny end was now at .0025" which is a thousandth over the originally measured value. I don't anticipate this being a problem since the crank's noodle'y end could always be finger-deflected this much, and a bearing in the gear box cover will stabilize it. The doweled cam box cover easily slipped into position later, and there's no visible flywheel wobble.

Before I could work up enough incentive to begin re-machining the damaged parts, I really wanted to understand what I did wrong the first time so I didn't blindly repeat the same mistake(s). My conclusions weren't very settling, though.

The 1-1/4" long main drive gear on the crankshaft was originally machined from Stressproof and pinned to the crankshaft using a pair of 3/32" diameter dowel pins. These pins were centered in the roots between the gear teeth and oriented 90 degrees to each other. The holes for these pins were originally drilled and reamed with the gear in place on the crankshaft and both pins were 'snugly' fitted. The rear pin was placed so it would be covered by the starter drive gear and the oil pump gear. The front pin was placed so it would be covered by the camshaft gear. My theory was that even if the pins tended to creep out of position a bit, they would be continually re-centered by the teeth of their cover gears. The rear pin, in fact, had two gears to insure it remained centered.

A half dozen seconds after being drill started, the engine locked up. The crankshaft gear, the starter drive gear, and the oil pump gear were all severely damaged. A third pin in the oil pump gear remained in place and undamaged. The outer crankshaft pin was found undamaged and stuck to the magnetic drain plug in the oil sump along with debris from the gears. The rear crankshaft pin had been sheared. A portion of it was still inside the crankshaft, but the other half was also in the sump.

The two dowel pins were the obvious culprits, but the only sequence of events that I can came up with to explain the damage is pretty difficult to swallow. First, in order to escape damage, the front pin had to have fallen out sometime before the catastrophe. This would have occurred sometime before the engine was started and while the pin happened to be vertically oriented. After the engine was started, the rear pin would have to have shifted in order for its shear line to end up where it was. The incredible thing about this shift is that it had to have occurred while running and within the time it took for the crank to rotate just three quarters of one revolution. Otherwise, it would have just been nudged back into position by either the oil pump gear or the starter drive gear. This single shift would also have had to be great enough to jam up the starter drive gear which was presenting no significant load to the crank. As a result of this jam, the momentum of the flywheel sheared the pin, and its free end continued on to create even more damage. I've not made total peace with this explanation, but it's the best I've been able to come up with. Visualizing those pins, especially the rear one, moving around as required to create the resulting failure isn't easy. In any event, I decided it was time to get on with the repairs.

The first step was to machine a new crankshaft gear. Probably the most critical machining in all the upcoming repair work was to precisely match the dowel holes in the new gear to those already in the crankshaft. I started by using a spindle microscope to measure their locations in the crankshaft. A horizontal rotary was then set up under the mill's spindle to precisely drill and ream the holes through the new gear. After slipping the gear onto the crank and installing a temporary pin in one hole, the other was reamed through for an over-size dowel pin. This pin was then carefully ground down for a tight fit through the gear/crank combination which I augmented with Loctite. With the first pin in place, the process was repeated for the second one. I considered using tapered pins but fitting them in the roots between the gear teeth didn't seem practical, and I wasn't sure a spring pin would handle the torque requirements of the starter motor. In any event, I can't see the new pins going anywhere, although I've been wrong before. - Terry

 

[GRAYHIL]

Hi All
I am thinking of making this model, it looks challenging but I have studied the drawings and drawn a few of the pieces.
Overall the drawings look top class.
I live in the UK and so will probably use the nearest metric fasteners but work to imperial imperial dimensions to produce the parts.

Graham

 

[mayhugh1]

Grahm,
I look forward to hearing your progress. Another builder in South America also started one by making some castings, but he hasn't reported any progress in several months. Keep us advised and preferably start a build thread.
Terry

 

[mayhugh1]

Before closing the crankcase up around the crankshaft, I drilled two 3/8" holes through the gear box and into the right crankcase half. With the ventilation previously added to the gearbox, these holes will help equalize the crankcase pressure throughout the engine. I also reworked the pressure regulator to drop the oil pressure to the top end to about half the 10 psi that I had been running. This required sleeving the housing of the existing regulator and re-boring it for a smaller ball and spring.

I had an aha moment as I was about to reinstall the cylinders over the pistons. I realized that I had mistakenly installed the 6.5 c.r. pistons instead of the 5.3 c.r. set that I had intended to use. I should have caught the mixup when I discovered I had to eyebrow them since I had determined much earlier that the lower compression pistons were interference-free. The load created by the higher compression explains why the starter's cranking speed has been lower than what I had originally estimated based upon my Howell V-twin measurements. Since I had plenty of spare rings, I installed a new set on the lower compression pistons before really installing them this time. The rings on the pistons that were in the engine showed no visible wear. There was only about 15 minutes of idling time on them, but the colored oxide left on their o.d.'s during heat treatment was still uniformly visible around all four rings. The .020" thick teflon head gaskets also looked new and were still at their original thickness, and so they were reused.

The brass starter drive gear was severely damaged and had to be replaced. Its shaft was also bent, and the bearings on either end damaged. This shaft is a complicated part that I wasn't looking forward to re-machining. It contains not only an integral sprocket but a difficult to machine face groove that contains the beveled drive gear for the distributor. I generally have poor success with straightening shafts, but with nothing to lose I chucked this one in the lathe for support and went after it with a tiny hammer and wood drift. I was greatly relieved when its .026" runout was reduced to less than a thousandth.

All the gear box bearings are flanged blind press-ins. When I bored their pockets I left a space behind them for a puller. One of the photos shows an example of one of the simple pullers I made to remove two of the three bearings.

Before realizing my piston mistake, I was planning to machine the replacement starter gear from steel since the loading on the starter system had turned out to be greater than expected. The cam gear had received only minor damage, but since it was nearly identical to the starter gear, I decided to machine both replacements from the same brass blank.

For my own learning, I had been running an experiment involving the camshaft/lifter wear. The large valves required fairly stiff springs, and the tiny contact areas between the lifters and cam lobes created concerns about wear. After their oil quenches, the lifters were tempered at 350F and the cam was tempered at 450F. My experiment was to move all the wear to the cam and away from the custom tipped lifters. Here, the contact areas would be allowed to gradually increase as the surfaces wore into each other. My hope was that the wear would eventually diminish as the lobes moved away from yielding leaving minimal damage on the cam.

The lifters showed no wear as expected, but the wear on the cam lobes seemed excessive for the relatively small amount of running time. I decided to switch to my backup plan which was to make everything as hard as the back of Superman's head. When I made the camshafts, I machined two identical copies, and so I re-heat treated the unused spare. This time I used a 3 hour temper at 325F which left it as hard (and as brittle) as I felt comfortable with. Draw tests with a file showed a very obvious increase in hardness compared with the first cam. The spare was then assembled with its newly machined gear and timed to crankshaft.

During disassembly of the engine after the crash, I noticed a significant amount of oil (about 3 ml) laying in the outside corners of both rocker boxes. In the rear rocker box, the nearest pushrod drain hole is the one for the intake valve. This hole is far enough away so the level of oil that accumulates in the corner will continually flood the o-ring seal between the rocker box and the valve box. This has been the source of the mysterious leak that invariably continued well after the engine was shut down. The front rocker box isn't as much of a problem since its nearest drain, the hole for the exhaust pushrod, is closer and the accumulated oil level isn't as high.

With the drain holes where they are, it isn't possible to eliminate the problem by adding something inside the rocker boxes. However, I did find and JB Weld a small machining defect in the rear valve box's o-ring sealing surface that should make a difference. I also redirected the outside corner oil spray nozzles away from the rear surfaces of the rocker boxes and toward their covers to help hasten oil drainage during running.

In addition to straightening two shafts, the final damage tally was four gears and three bearings. After installing the distributor and setting the valve lash, the engine is finally ready to go once more. I can't say the past couple weeks have been any fun, but a number of small improvements made to the engine have definitely made it productive.

I've been studying the jet design in the drawing for the air bleed carburetor that George Britnell posted to the downloads section years ago. I now think the Knucklehead's main jet design is backwards causing it to provide too little control over too much fuel. Since the current carburetor looks like it really belongs on the engine, I'm going to try to re-work its jet assembly before totally abandoning it. - Terry

 

[kvom]

That took less time than you likely expected.

 

[RonC9876]

Looking good Terry. Sometimes it isn’t too difficult to straighten a crank. Other times things just get worse as you go along and you have to scrap the part. I have had them go both ways. Not that I make a habit of bending cranks, but things happen. Usually when transporting them to a show and they fall off the cart. You know; trying to cram them on the cart to make the fewest trips to the car as possible. Happened twice in the past. I never learn. Looking forward to some videos of this creation running. Best of luck as always.

 

[mayhugh1]

With the engine finally back together, I made another attempt at modifying the main jet. The cross-sectional drawings in the first photo show these most recent changes. In the top (original) design, the needle was expected to regulate the amount of fuel entering the jet through the twelve radially-drilled inlet holes at its lower end. In operation, this fuel will be drawn out of its top end by the low pressure area inside the Venturi where it's mixed with the air flowing through the Venturi on its way into the engine.

Based on George Britnell's air bleed carburetor for an engine with similar size pistons:

https://www.homemodelenginemachinist.com/threads/air-bleed-carburetor.29019/

one or two entry holes would have been sufficient. With so many more to deal with, a needle with a pretty challenging taper is required for usable control. So far, I've ground three different needles with essentially no success. The needle that I was testing at the time of the crash was supplemented by a bushing that was pressed over the jet to cover up several of the holes.

For this latest trial, I bushed the i.d. of the jet's output end and lengthened the needle so it has only to meter fuel entering the bushing which now protrudes into the Venturi. The needle's taper is still important but much less demanding. Surprisingly, its performance in the engine was pretty much the same as before. The engine had to be drill-started, and it idled overly rich with neither the throttle nor the needle having any noticeable effect. Something was obviously overshadowing the fuel contributed by the main jet.

For the first time, I began playing with the idle mixture screw and found it too had little to no effect. This was a surprise because the components making up the idle circuit are straightforward, and their machining was easily verified. I removed the carb and plugged the idle pickup tube in order to completely disable the idle circuit. Now, with only the main jet metering fuel, the engine started immediately with the electric starter and for the first time could be made to idle with a clean exhaust. There was still no reaction to the throttle, but control of the air/fuel mixture by the main jet's needle was now obvious, and it functioned as expected. Up until this point, the idle circuit has been supplying an excessive and uncontrolled amount of fuel to the engine that has been masking the contribution from the high speed jet.

With the idle circuit still disabled, I added an atmospheric vent to the carb bowl. So far, the pressure inside the bowl (the pressure that actually 'pushes' fuel into the Venturi) has been defined by the recirculating loop's input/output pressure differential. Adding this vent changed the main jet's operating point, and its needle had to be closed an additional 1-1/2 turns to return to the engine to its peak idling speed. Evidently the pressure inside the bowl had somehow been a bit less than atmospheric. Even though it had no effect on the throttle issue, it's now clear that a bowl vent is required to ensure the carb settings are independent of the pump's speed.

A clue to the throttle issue may lie in the transition space between the Venturi and the intake manifold. This space is different in a model engine application than it is for this style carburetor's more common usage in full-size engines. In this application, the diameter of the intake runner (I need to quit calling it a manifold) is smaller that the diameter of the carb body feeding it. The second photo shows a cross-sectional view of the various areas involved. In a full-size engine the carb feeds a voluminous plenum whose diameter is at least greater than that of the carb's throat. Visually, this space screams for a smooth transition between the carb body and the intake runner. I'm beginning to think, though, that an unfortunate combination of geometries may be reducing the clearance allowed around the butterfly to a difficult-to-achieve level. It would certainly be a disappointment although not totally surprising to discover that the engine has actually been running at its maximum speed all along due to a leaky butterfly. In any event, this is where I plan to focus my attention.

The grandkids are coming in for a week, and since they aren't shop types I'll have some time to think about this. In the meantime, I made a video that captures the project's progress so far, just in case another mishap occurs before the carburetor problems are sorted out:




As an aside, the rebuild has eliminated all the annoying oil leaks. The oil that's blown out through the crankcase vent has increased, probably due to the venting added between the crankcase and gear box. This particular 'leak' isn't of concern since it's typical for vintage engines and almost a requirement for old Harleys. There's probably an oil maelstrom going on inside the engine while it's running that's keeping everything nicely oiled, and that's a good thing. Now, if I can just get it to rev up (or down) ... - Terry

 

[RonC9876]

Terry: When I first tried adapting a diaphragm carb to my V-8 Challenger engine, I got very little response from the throttle settings. This carb had a 5/8” diameter butterfly valve. I bushed the diameter down to 1/4” to where the butterfly valve diameter was now about the same diameter as the head of the screw holding it to the actuating shaft. The engine would now respond to the throttle settings and main and idle jet settings as well. It was tricky drilling holes in this bushing to line up to the jet holes in the original 5/8” bore. It has worked well ever since. Just my experience with a butterfly valve carb and sizing it to a model engine. Most of my carbs use throttle barrel valves with about the same small bores.

 

[mayhugh1]

The stock throttle has so little effect on engine speed that it defies logic and creates an interesting problem to solve. Although not a perfect design, a few calculations show that it should certainly be functional. When fully closed, the .005" clearance around the disk is equivalent to a leak of only 5% when compared with the area of the intake runner. The theoretical difference between being fully closed and fully open (100% intake area) is about 35 degrees of rotation. Its effectiveness, however, is getting lost somewhere - either in the transitional space labeled D in the cross-sectional drawing of the induction system or within the long complex intake runner itself.

In the full-scale carburetor applications that I'm familiar with, the carb typically feeds an induction system whose diameter is at least as large as the carb's i.d. below its throttle. In this case, though, the .312" intake diameter is significantly smaller than the .500" i.d. of the carb. For air-only flow, the throttle would probably work as expected, but once fuel is added, the density and momentum of the resulting mixture are changed. It's difficult to appreciate just how much they're changed until you see a typical air/fuel flow inside an actual carburetor. I found a Youtube video demonstration:



of a clear plastic carburetor attached to a small Briggs & Stratton engine. The carburetor in the video is relatively simple and very similar to the Knucklehead's, but its throat matches the engine's intake. A two minute segment beginning at 1:20 is pretty eye-opening and shows a much wetter mixture than I had been picturing. I had been visualizing the flow through the carburetor as something from a air hose when I should have been thinking more in terms of a garden hose. This made me suspicious of the stepped down area looking into the intake, and had me wondering whether fuel bouncing off it might be creating a bottleneck into the engine.

I was curious to see if I could improve the throttle's effectiveness with a simple transitional filler machined to smooth the abrupt step between the carb and the intake runner. Although a different angle of attack could have been selected, the one I used is shown in the second CAD drawing (the violet-colored filler). Although the filler changed the operating point of the main jet (it had to be leaned out compared with using no filler), and it raised the engine's steady state rpm, the throttle still had zero effect on engine speed.

The next attempt was to machine a plug for the carb body behind the Venturi that was bored to match the i.d. of the intake. A new .312" diameter throttle disk and shaft were machined to replace the original .490" disk. For a first test, I didn't extend the passage for the original idle jet through the plug which meant that all air and fuel would have to come through the Venturi. For now, I was only interested getting the engine to respond in some manner to the throttle.

Testing showed eliminating the step transition between the carb and intake dramatically solved the throttle mystery. With the throttle half open, the engine immediately started, and it finally revs up and down under throttle control as expected. The speed appears to change smoothly over the entire range of the throttle, although others' experiences would say there might be room for improvement at one end or the other with a second jet. The next (and hopefully final) step is to determine whether a second jet is actually needed and, if so, whether it should control additional air or fuel. - Terry

 

[V22]

Thanks for taking the time to post the R&D and associated CAD models Terry. Very interesting work and a great outcome. The plastic carb video was an eye opener in illustrating how the wet mixture is.

 

[mu38&Bg#]

Scale carbs are by far too large for the resulting displacement.

 

[mayhugh1]

I planned to drill a radial hole through the throttle-reducing plug in order to make the idle jet functional again although I wasn't sure if I'd be using it to add additional air or fuel. After playing with three different main jet needles, I finally decided to not make anymore carb modifications and to continue using it as a single jet device. Upon adjusting the needle for running at w.o.t., the carb seems to maintain a usable air/fuel ratio at idle. Surprisingly, the idle speed or quality doesn't seem to be very sensitive to the jet's setting. There's still some mysterious goings on inside that carburetor, but at least this time they're working in my favor.

With a fixed, and what feels like a barely open main jet, the engine will start and rev up to 4500 rpm and idle down to 1200 rpm. The exhaust noise is deceptive, and the engine sounds like it's running at less than half those speeds. I measured the flywheel rpm using both a mechanical and an optical tach and got essentially identical results. The engine was run at both these extremes in the second video. It wasn't totally unexpected, but starting is much easier when hot. The main jet setting is stable and once set didn't require further adjustment even during multiple runs over several days.

With the engine finally responding to its throttle, I took another look at ignition timing. I had been running earlier with only five degrees advance in order to eliminate kickbacks during cold startup. I may be overly sensitive to this issue, but an old Virago of mine cold started similarly, and I went through two starter clutches during the time I owned it. The cold startup procedure that I eventually adopted allows 10 degrees advance, but the kickbacks return if I try to add an additional 5 degrees. Cold startup begins with a few seconds of full choke cranking at part throttle and the ignition turned OFF. This preliminary step wets the induction system enough so plug firing is delayed until the flywheel's momentum has a chance to build while re-cranking with the throttle at idle and the ignition ON.

Being more accustomed to the heat build up in engines having more cylinders, the knucklehead's operating temperatures were a pleasant surprise. During typical three minute variable speed runs in 80F ambient air, the outside head and cylinder temperatures never exceeded 145F. The temperatures of the heads at the end of the second video measured less than 120F. The temperatures of the exhaust pipes are essentially identical, and they typically run some 15 degrees cooler than the heads.





This will likely be my last post on this build. I'd like to thank Draw-Tech for the drawings that were made freely available to us in the downloads section of the forum. The quality of those drawings are among the best we're likely to come across in the model engine community.

Of the readily available V-twin model engine designs, Draw-Tech's Knucklehead is the most realistic looking, but this realism comes at the expense of complexity. It's not a beginner's project, and it felt as challenging as any other engine I've built. Although a majority of its parts were made on my Tormach mill and Wabeco lathe, CNC is not a requirement. The most demanding machining steps included several drilling/boring operations that required careful compound angle setups, and those operations were best done manually.

The component with which I struggled most was the tapered crankpin. The crankshaft design closely follows that of the full-size engine, and it works well. Its machining would have been much simpler if I had fabricated the crankshaft before the crankcase. This would have allowed the widths of the crankcase halves to be adjusted around the finished crankshaft for a proper thrust clearance. The alignment sleeve described in the build should be considered a must-have for assembling the crankshaft.

Issues with any new design can be expected to turn up during early builds. On this engine, these included the oiling system and, of course, the carburetor. The oil system's capacity is excessive, and the pump can easily be reduced in size. I don't recommend trying to solve the problem with sloppy machining, since the addition of a pressure regulator seems to be an effective solution.

Some may have been put off from building this engine by the 'big bang' camshaft provided for it in the original drawings. After all, the 'Harley sound' is a big reason for building this engine. The replacement cam that I designed works well, and the documentation included in the build log should be sufficient for others to duplicate it. Coupled with the drag pipes, it creates a loud exhaust with the infamous Harley lope.

A starter motor is a nice addition to any model engine, and the gear motor that I chose works well and looks at home on the engine. The engine's starting torque required additional gearing beyond that in the original design as well as a one-way clutch. I used a system of sprockets and chains for the starter drive, but the cogged pulleys and kevlar belts used in the original design may work as well.

If I were going to build a second engine, I'd make three changes. The first is that I'd lengthen the cylinders about three quarters of an inch. I received a comment during the engine's initial assembly saying the engine looked a bit 'squatty' compared with a full-size Knucklehead. And now, with the engine completed, I have to agree.

The second change would be the addition of a crankcase vent tube to route puke oil to the engine's underbelly and onto the floor of the stand. The internal oil scrubber that I designed for the crankcase vent added to the dipstick doesn't work, and every run leaves some oil on the top of the crankcase around the dipstick. This requires installing an appropriate fitting in the side of the gearbox , but I'll likely wait until the gearbox has to be opened up for some other reason before it's added to this engine.

A third change would be a modification to the distributor to manually advance the ignition timing from 0 to 20 degrees once the engine is started and running. I believe the early Knuckleheads with their kick starters were provided with this as well. An ideal solution would be independent of the distributor's installation in the crankcase. The current 10 degree static advance doesn't leave significant performance on the table, but the engine's starting procedure would become more reminiscent of the early Harleys. - Terry

 

[Scott_M]

Outstanding !! It sounds as good as it looks.
I cannot thank you enough for taking us along with you on this build. It is as impressive as any of your previous builds and does indeed set the bar. I have to admit though that I am a little saddened that it has concluded I guess I will just have to wait for your next project.

Thanks Terry !! and congratulations.


Scott

 

[napoleonb]

Well done!

 

[stragenmitsuko]

It's been a wonderfull journey , thx for letting us tag along .

Something I'v been wondering about with the way harley davidson designed their crankshaft .
Maybe a little OT , altough the model expiranced thesame issue when the gears locked up .

Imagine a knucklehead bike , with a typical harley-davidson rider and maybe a matching girfriend on the duo
going up hill in the highest gear at relatively low speed with the throttle wide open .... Worst case senario .

That would put a huge torque on the crank pin and the output side of the crankshaft tht drives the rear wheel .
With only that taper to hold it . I can imagine the taper breaking loose given harsh conditions
with all kinds of problems . If I was a designer , I would have used mabe a splined shaft with a taper
or something similar . Just brainstorming