Another Knucklehead Build

[petertha]

Be interested in Terry's reply too. I haven't gotten that far yet, but in my simple mind I think an irregular break left untreated is equivalent to using a larger heat set dowel diameter. If you had some means of knowing how much this is, one might be able to guess the net effect relative to Trimble plot. But I would assume stroking it with a perpendicular file would be better than nothing within reason?

 

[mayhugh1]

Jeff,
Here's a photo of a couple gaps. These rings were just now taken out of the furnace and the running gaps haven't yet been filed. I looked at all 24 of them and they were identical to the ones in the photo.

I ground very sharp corners on the HSS blades of my cleaver and then used shims to insure they were perfectly aligned. In use, only the two sharp corners of the blades contact the ring, and I can feel a definite snap when the ring breaks without it twisting. - Terry

 

[mayhugh1]

And here is a reject untreated ring that I just found in the trash and just now

re-cleaved. - Terry

 

[mayhugh1]

The rings were heat treated at 975F for three hours and allowed to cool overnight. In comparison with what I can remember about previous batches, this one came out looking pretty clean. It could very well be that the free machining steels I've been using in the past for the fixtures have been responsible for those annoying surface deposits. Before removing the rings from the mandrel, their collective o.d.'s were burnished with a white Scotchbrite pad. Burnishing didn't remove any metal - it just brighten the oxidized surfaces left behind by the heat treatment.

The extreme heat, along with the pressure exerted by the mandrel on the ring stack, caused the lapped surfaces of some of the rings to stick together. After sliding them off the mandrel, a few had to be separated with a razor blade. Starting at the gaps to avoid damaging important surfaces, they were easily worked apart.

With the rings separated, a .004" running gap was filed into each using a thin diamond file. These inexpensive files leave smooth surfaces on cast iron without magnetizing it. With the ring inside a simple shop-made gage, its gap could be checked with a feeler gage. The exact gap isn't critical since its only purpose is to prevent thermal expansion from binding and breaking a ring inside its cylinder. The gap is an insignificant leak especially since most of it is sealed off by the lower wall of the piston groove. Even at a couple extra thousandths, the leak it presents is equivalent to a circularity error of only micro-inches.

After gapping, the sides of the rings were briefly lapped for a final time using 1000 grit compound. The fixture used in this step gripped the rings around their o.d.'s rather than their i.d.'s as was done during the pre-heat lapping. When completed, the rings were thoroughly cleaned in lacquer thinner.

The final step in the ring making process was the light test. A photo shows the components of the fixture used to adapt a bright (250 lumen) miniature flashlight to the bottom of one of the Knucklehead's cylinders that was used to test the rings. Each ring under test was supported by its own spring force inside the cylinder about 3/8" from its top. At the beginning of each test, a close-fitting shouldered Delrin plug was inserted into the ring through the top of the cylinder in order to close up its center. The o.d. of the plug was .020" under the cylinder's bore to allow light leaking between the ring and cylinder wall to be easily seen from the top of the cylinder. The tests were done in a totally dark room with the flashlight held vertically in a vise.

It's extremely important for the ring to be positioned squarely inside the cylinder during its test. Even a perfect ring tilted inside the cylinder will make contact with the wall as an ellipse and pass enough light to fail the test. To insure the rings are positioned squarely inside the bore, a close-fitting Delrin rod was faced and used to push them up through the bottom of the cylinder and into position for their tests. The diameter of the rod was just .003" under the cylinder's bore to insure its own face was square to the bore.

My criteria for a good ring was for no light to pass between it and the cylinder wall. Out of 25 rings tested, 14 met this criteria. Five rings had obvious leaks that presented as prominent crescents, and these were discarded. Six rings had just a wisp of leakage and in all honesty might have been 'good enough'. These were labeled 'marginal fails' and saved for some future use. I've included representative photos of all three test results. The final post heat-treat yield turned out to be 56% compared with 70% for the Merlin build. - Terry

 

[stragenmitsuko]

truly amazing ! love this thread

 

[propclock]

Very informative, my rings are made the same way but, not light tested.
They will be in the future! I just assumed they will"break in".
And in fact after much running the surface of the un tested rings show
uniform wear on their O.D.'s . But break in did take a while.
Thanks for your wonderful posts.

 

[kuhncw]

Terry, thank you for the detail on your ring making process and especially your light-tight testing setup.

Chuck

 

[petertha]

Terry, so you are doing the light test on the same reference liner, right. Not finding a ring that matches a particular liner & keeping them paired?
If so, in your high cylinder count)engines, what kind of +/- bore accuracy were you satisfied with so that rings could be considered interchangeable coming off the light test apparatus?

 

[mayhugh1]

Terry, so you are doing the light test on the same reference liner, right. Not finding a ring that matches a particular liner & keeping them paired?
If so, in your high cylinder count)engines, what kind of +/- bore accuracy were you satisfied with so that rings could be considered interchangeable coming off the light test apparatus?

Petertha,
I used one the of the actual cylinders for all the tests. I lap my cylinders to a tenth of one another, or at least what I believe to be a tenth using the bore gage and micrometer that I use. Realistically, my repeatability is probably only a couple tenths or so. I lap all the cylinders as a group, opening each one up only one or two tenths at a time so they are all at the same diameter as they cross the finish line together. It's time consuming because of all the additional cleanings and measurements, but it avoids one ring from overshooting the target and then having to open up all the others to match. - Terry

 

[mayhugh1]

While machining the pistons last August (post #108), I made two sets: a high and a low compression pair. I've decided to use the low compression pistons since they'll present less of a load to the starter motor. Although it wasn't necessary, I machined 'eyebrows' into them for a bit more valve clearance. Even at their mild 5.3 compression ratio, the edges of the engine's huge valves come uncomfortably close to the hemi-topped pistons. After installing rings on them it was time to begin final assembly.

While re-installing the cylinders, I decided to replace the small pattern 10-32 nuts that I had been using to mount the cylinders to the crankcase and the heads to the cylinders. It was a nit-pick, but the nuts I was using never looked at home on the engine. I spent a day machining a couple sets of flange nuts that are not only more appropriate but provide a few more threads. After bead-blasting, they were painted black with baked-on Gun Kote so they'd blend in with the blued cylinders.

Another last minute but much more significant change was to replace the bolts in the rocker boxes with studs. In the original design, these bolts not only secure the covers to the rocker boxes, but they're also the shafts for the rocker assemblies. I finally realized that because the heads of the bolts are on the outsides of the rocker box covers, a lot of disassembly was going to be required to merely verify the rocker boxes are properly receiving and draining oil. All four of the 4-part rocker arm assemblies were going to have to be pulled in order to remove the covers, and this means the pushrod assemblies would have to be removed as well.

Both ends of the studs were drilled/tapped for 5-40 setscrews that were installed with permanent thread-locker to provide wrench-able hex sockets. The studs not only allow the covers to be removed without disturbing the rocker assemblies, but they also simplify installation of the pushrods and covers.

I wanted to duplicate the signature acorn nuts I've seen on the rocker boxes of the full-size engines. I was about to machine my own when I located some already polished stainless acorns of the correct size in my collection of scrap fasteners. The only problem with them was that they were threaded 1/4-20 instead of 10-32. Since there was no room for a threaded adapter, I bored the nuts out for plugs that were pressed, Loctite'd, and pinned into place. After threading the plugs, I made a set of teflon gaskets to seal them to the covers.

Another loose end involved sealing the four valve box covers. These covers have extremely narrow mounting flanges with lots of mounting holes and aren't well suited to a conventional gasket. I didn't want to use a messy sealer, so instead I used this vinyl sheeting that I found in a local craft store:

https://www.amazon.com/Oracal-Gloss...teway&sprefix=oracal+651+vinyl,aps,183&sr=8-6

It's 2.5 mils thick, adhesive-backed, and intended for making rub-on stencils. Its adhesive probably won't stand up to fuel or to significant heat. But, in this particular application as a gasket completely sandwiched between two machined surfaces, it should be satisfactory. After cleaning the covers with alcohol, they were set down onto the adhesive side of the sheet so the material could be trimmed from around their peripheries using an Xacto knife. Working from the non-adhesive side, the holes were cut remarkably clean using a chucking reamer held in a pin vise.

The gearbox assembly began with the installation of the oil pump and camshaft. The pushrod/cover assemblies were then installed and the valve lash set. I used the doweled plexiglass fixture plate constructed earlier for temporary outer bearing support since the valve train exerts a significant downward force on the camshaft. When it came time to install the starter chain drive assembly, the fixture plate had to be carefully removed to avoid extracting the cam.

With the starting system in place, the gearbox cover could be installed. With the cover engaged on the gearbox dowels, a piece of 3/32" steel welding rod inserted horizontally behind the cover was used to push the outer end of the camshaft upward so the cover could be slid into place. To remove the cover, a thin spatula can be used to prevent the camshaft from being pulled out of place. If this happens, the lifters, the pushrods, and their covers will have to be removed in order to re-install the cam. I practiced this maneuver with a paint scraper to make sure it was feasible.

With the gearbox cover bolted in place, a test could finally be made of the fully loaded starter inside the engine. Spark plugs were temporarily installed and the flywheel manually turned over a few times. Plenty of compression was evident - much more than in my Howell V-twin. Neither of my shop-made model engine compression gages can fit inside the plug recess in these heads, and so I couldn't make an actual compression measurement. The good news was that the starter had no difficulty in cranking the engine. The starter motor that's currently in the engine is the 165 rpm gear motor (post #148) which spins the engine at 200 rpm.

The distributor was then installed and the timing set to about 15 degrees BTDC using the timing light feature that I designed into the ignition circuitry (post #167). Looking down on top of the distributor, the rotor spins clockwise which is opposite to the direction of the distributor in the original drawings.

Before installing the carburetor, I added 30 ml of oil to the sump and then spent the next few days chasing oil leaks. The first one, which I should have anticipated, occurred between the bottoms of the valve boxes and the heads. This issue concerned only the two outside valve boxes since they're mounted at steep angles. The oil that collects inside them can't drain quickly enough through the tubes intended to return the oil to the crankcase, and so some oil seeps out between the bottom of the valve box and the top of the head.

The leak was easily solved without disassembling everything that had just been assembled. After dropper'ing acetone into the spaces between the valve boxes and heads to flush out all traces of oil, I dropper'd in a small quantity of Loctite 290. This product is a wicking thread-locker that will easily and permanently seal gaps up to .005" after an overnight cure. I should have removed the drain tubes before applying it because I managed to also seal one of them up and eventually had to fabricate a replacement. - Terry

 

[gbritnell]

Terry,
I really enjoy following your builds. The in-depth explanations teach me something every time I red them.
Here's some food for thought. When I built my V-twin I used 90 degrees for the cylinder angle. (Better balance?) l used aluminum for my intake runner with just .015 gaskets at the head joint. The engine started and ran fine for just a short time then it needed the carb richened, more and more until it would finally stall. At that point it wouldn't restart. Upon letting ir cool the same scenario would play out again. As I was running the engine my friend noticed the fuel in the clear supply line was oscillating, back and forth. This was due to the heat migrating from the head, up the manifold and into the carb. This would boil the gas (vapor lock) and if the needle valve was opened it would draw extra fuel to cool it for a short while until eventually stalling. I removed the manifold and machined the head joint enough to insert a Corian spacer. This insulator solved the problem.
With your cylinder angle at 45degrees I'm thinking you're going to transmit the heat much quicker so when you run the engine be aware of the manifold temperature.
gbritnell

 

[mayhugh1]

George,
Thanks for the nice comments and advice. Although it's actually in the above photos, it's not at all obvious, and I should have made mention of it in the above text. The square flange on the end of the intake manifold is a pressed-in 1/8" thick Delrin spacer to block heat from the manifold to the carb just as you say. The black-looking flange end that's visible in two of the above photos is the Delrin and not just typical bad photo glare. - Terry

 

[Rozlo]

While machining the pistons last August (post #108), I made two sets: a high and a low compression pair. I've decided to use the low compression pistons since they'll present less of a load to the starter motor. Although it wasn't necessary, I machined 'eyebrows' into them for a bit more valve clearance. Even at their mild 5.3 compression ratio, the edges of the engine's huge valves come uncomfortably close to the hemi-topped pistons. After installing rings on them it was time to begin final assembly.

While re-installing the cylinders, I decided to replace the small pattern 10-32 nuts that I had been using to mount the cylinders to the crankcase and the heads to the cylinders. It was a nit-pick, but the nuts I was using never looked at home on the engine. I spent a day machining a couple sets of flange nuts that are not only more appropriate but provide a few more threads. After bead-blasting, they were painted black with baked-on Gun Kote so they'd blend in with the blued cylinders.

Another last minute but much more significant change was to replace the bolts in the rocker boxes with studs. In the original design, these bolts not only secure the covers to the rocker boxes, but they're also the shafts for the rocker assemblies. I finally realized that because the heads of the bolts are on the outsides of the rocker box covers, a lot of disassembly was going to be required to merely verify the rocker boxes are properly receiving and draining oil. All four of the 4-part rocker arm assemblies were going to have to be pulled in order to remove the covers, and this means the pushrod assemblies would have to be removed as well.

Both ends of the studs were drilled/tapped for 5-40 setscrews that were installed with permanent thread-locker to provide wrench-able hex sockets. The studs not only allow the covers to be removed without disturbing the rocker assemblies, but they also simplify installation of the pushrods and covers.

I wanted to duplicate the signature acorn nuts I've seen on the rocker boxes of the full-size engines. I was about to machine my own when I located some already polished stainless acorns of the correct size in my collection of scrap fasteners. The only problem with them was that they were threaded 1/4-20 instead of 10-32. Since there was no room for a threaded adapter, I bored the nuts out for plugs that were pressed, Loctite'd, and pinned into place. After threading the plugs, I made a set of teflon gaskets to seal them to the covers.

Another loose end involved sealing the four valve box covers. These covers have extremely narrow mounting flanges with lots of mounting holes and aren't well suited to a conventional gasket. I didn't want to use a messy sealer, so instead I used this vinyl sheeting that I found in a local craft store:

https://www.amazon.com/Oracal-Glossy-Permanent-Vinyl-Inch/dp/B01N42YR3P/ref=sr_1_6?crid=2HA1C5QA6S92E&keywords=oracal+651+vinyl&qid=1558133549&s=gateway&sprefix=oracal+651+vinyl,aps,183&sr=8-6

It's 2.5 mils thick, adhesive-backed, and intended for making rub-on stencils. Its adhesive probably won't stand up to fuel or to significant heat. But, in this particular application as a gasket completely sandwiched between two machined surfaces, it should be satisfactory. After cleaning the covers with alcohol, they were set down onto the adhesive side of the sheet so the material could be trimmed from around their peripheries using an Xacto knife. Working from the non-adhesive side, the holes were cut remarkably clean using a chucking reamer held in a pin vise.

The gearbox assembly began with the installation of the oil pump and camshaft. The pushrod/cover assemblies were then installed and the valve lash set. I used the doweled plexiglass fixture plate constructed earlier for temporary outer bearing support since the valve train exerts a significant downward force on the camshaft. When it came time to install the starter chain drive assembly, the fixture plate had to be carefully removed to avoid extracting the cam.

With the starting system in place, the gearbox cover could be installed. With the cover engaged on the gearbox dowels, a piece of 3/32" steel welding rod inserted horizontally behind the cover was used to push the outer end of the camshaft upward so the cover could be slid into place. To remove the cover, a thin spatula can be used to prevent the camshaft from being pulled out of place. If this happens, the lifters, the pushrods, and their covers will have to be removed in order to re-install the cam. I practiced this maneuver with a paint scraper to make sure it was feasible.

With the gearbox cover bolted in place, a test could finally be made of the fully loaded starter inside the engine. Spark plugs were temporarily installed and the flywheel manually turned over a few times. Plenty of compression was evident - much more than in my Howell V-twin. Neither of my shop-made model engine compression gages can fit inside the plug recess in these heads, and so I couldn't make an actual compression measurement. The good news was that the starter had no difficulty in cranking the engine. The starter motor that's currently in the engine is the 165 rpm gear motor (post #148) which spins the engine at 200 rpm.

The distributor was then installed and the timing set to about 15 degrees BTDC using the timing light feature that I designed into the ignition circuitry (post #167). Looking down on top of the distributor, the rotor spins clockwise which is opposite to the direction of the distributor in the original drawings.

Before installing the carburetor, I added 30 ml of oil to the sump and then spent the next few days chasing oil leaks. The first one, which I should have anticipated, occurred between the bottoms of the valve boxes and the heads. This issue concerned only the two outside valve boxes since they're mounted at steep angles. The oil that collects inside them can't drain quickly enough through the tubes intended to return the oil to the crankcase, and so some oil seeps out between the bottom of the valve box and the top of the head.

The leak was easily solved without disassembling everything that had just been assembled. After dropper'ing acetone into the spaces between the valve boxes and heads to flush out all traces of oil, I dropper'd in a small quantity of Loctite 290. This product is a wicking thread-locker that will easily and permanently seal gaps up to .005" after an overnight cure. I should have removed the drain tubes before applying it because I managed to also seal one of them up and eventually had to fabricate a replacement. - Terry



View attachment 109394 View attachment 109395 View attachment 109396 View attachment 109397 View attachment 109398 View attachment 109399 View attachment 109400 View attachment 109401 View attachment 109402 View attachment 109403

You have done an awesome job on this so far, a work of art.

 

[mayhugh1]

While chasing oil leaks, I got some up close experience with the performance of the Knucklehead's oiling system and then realized I had a major issue. The gear pump, for what little is required of it, is over capacity by an order of magnitude. I eventually came to realize that cranking the engine with the starter will likely flood the top end of the engine with nearly all the oil in the sump before it has a chance to start.

Some calculations that I should have performed earlier show the pump's capacity at, say, 50% efficiency is about .032 cubic inches per pump revolution. This was easily determined by computing the volume of all the little gear pockets that move oil around the inside of the pump housing during one revolution of its driven gear. Since the pump is driven at 2/3 the speed of the crankshaft, the flow rate will be about 4.2 c.i./min. at a 200 rpm cranking speed and about 21 c.i./min at 1000 rpm.

An reasonable sump capacity for this engine is about 60 ml or 3.7 c.i. This level insures the pump's intake is covered and that the crankshaft's flywheels are in enough oil to whip up windage to lubricate the engine's bottom end. Calculations, however, show the sump will be entirely drained after only 50 seconds of starter cranking.

I made a few measurements to back up the calculations. Using a syringe, I calibrated the engine's dipstick to the amount of 5W20 in the sump. A 15 second cranking test dropped the amount of oil in the sump by 20ml or 1.2 c.i. corresponding to an effective flow rate of 4.8 c.i./min.

Unfortunately, easy fixes don't seem to abound. The return rate for the top end oil is on the order of 1 c.i./min and is too low to be of much help during what will likely be a typical run. The intake probably can't be safely restricted to be of much help since the pump will just react by sucking harder. Since the pump is self-priming, a common solution in dry sump model engines is to starve the pump with a slipping two phase mixture using a drip feed input from the storage tank. This won't work for this engine since its crankshaft uses a sump.

The only potential solution that I've been able to come up with is a relief valve on the output side of the pump to drop its head pressure. This is messy because it will have to be located inside the gear box and some heroic tubing work will be required to fit it place. Fine tuning the setting will have to be done by trial and error, which means it will have to be accessible from outside the engine.

If anyone has any other suggestions, I'd be happy to consider them. - Terry

 

[Ghosty]

Terry, can you fit a regulator valve in the case pipe, a simple "T" with a needle valve would work to re-direct a % of the oil flow straight into the sump

 

[stragenmitsuko]

A relief valve at the discharge of the pump to limit and stabilise the oil pressure . Thinking of a T-kind of piece at the discharge end of the pump , with a small ball and a spring . Pressure can be adjusted on the bench before assembly .
Combine this with a restriction , like a needle valve , somewhere in the feed tube . Possably where it exits the crank case . A small grub screw could be used to adjust it . Or a calibrated orifice in the union would work to .

Feed is directly proportional to rpm , you could slow down the pump rpm by extra gearing .
Hard to do and the problem will still exist at higher rpm .

Reduce the pump's efficiency by altering the bore around the gears or make the gears a bit smaller .
This could cause problems with priming if overdone and it's a "bad engineering" solution .


I'm sure you'll figure out a way ...

pat

 

[minh-thanh]

Hi mayhugh1!
just an idea :

The pump will pump both air and oil -> reduce the amount of oil pumped up
It does not change any part of the engine, just add a small tube inside the oil pipe for air

 

[RonC9876]

Terry: Wasn’t the Knucklehead a dry sump engine? I know the Panhead was and mine has an external oil tank with a dual pumping system. The crankcases on these engines can’t hold enough oil in the sump due to the flywheels taking up so much space. I seem to remember reading somewhere that the Knucklehead was the first Harley to use an external oil tank and to actually filter the oil for reuse.

 

[mayhugh1]

Ron,
You're right. The Knuckleheads were dry sump engines. In fact, I think they were the first Harley-Davidson engines to even have a recirculating oil system. Earlier engines were total loss systems using oil that the rider had to add before every ride. The model I'm working on was designed to use a wet sump, though. - Terry

 

[RonC9876]

The air bleed idea sounds plausible. It would be easy to implement and would work like a scavenge pump, pulling in a quantity of air along with the oil. Might not be reliable to prime every time though. The one on my Offy works flawlessly but the one on my Novi needs to be primed manually before a run. Easy to do since the pump is external. Has to do with the distance of lift required. Then again, it might be fine as is. It always amazes me how little oil is actually needed in these model engines. Once the internals are wet with oil, the engine will run happily for some time until the oil gets back to the sump. Not the best situation but this is a model and not required to do any work. Beautiful work by the way. Not that you needed me to tell you that. You already knew!