Discussion in 'A Work In Progress' started by mayhugh1, Mar 22, 2018.
Don't forget the big ones have dry sump & scrapers!
Scavenger pump - on my old Chief, the oil tank was by the gas tank. first thing you would do after kicking it over is pop the oil cap off to make sure the scavenger pump was returning. Otherwise, it would be wet sumping & smoking like hell. Not so easy on my old Knuck, cause the oil tank is under the seat.
Johwen here... One way i would adopt to control oil flow would be to as you have put in an pressure relief valve however I would have made it adjustable by using a screw to increase or decrease the the spring pressure. Alternatively a by pass pipe from the pump outlet to the sump with an adjustable outlet via a needle vale to control maximum bypass Cheers John
There may be an issue with the carburetor and/or intake manifold that's creating the need for a much higher starting speed than I had expected. The engine won't start until the drill starter approaches some 1000 rpm. Before revisiting the carburetion, though, I want to raise the speed of the internal starter.
The motor currently inside the Knucklehead is a 165 rpm planetary gear motor from Servo City:
My original testing (beginning with post #128):
was done using the 437 rpm version of this motor using my Howell V-twin as a makeshift load. This motor was capable of spinning the Howell at 400 rpm and was used to come up with a ballpark spec for the Knucklehead's starter. During the design of a chain drive to connect the motor to the engine's crankshaft, I discovered I'd have to live with a 20% step up between the gear motor's output shaft and the engine's crankshaft meaning that the torque delivered to the crankshaft would be down by 20%. I initially installed the 165 rpm version of this motor in the Knucklehead because of its highest available torque along with a potential 200 rpm cranking speed which at the time was my guess at a usable minimum.
The motor current required for a particular torque can be found from the motors' torque/current curves which are sketched in the first photo. The slopes of these curves, provided by the manufacturer, have been adjusted for the chain drive, and so they reflect the 20% torque drop at the crankshaft. The maximum torque available from any of the motors is limited by their 20 amp stall current.
A starter's cranking speed in a V-twin application is difficult to calculate because of its wildly varying load. Between compression strokes, the motor will attempt to run at its no load speed. A V-twin starter will spend roughly 60% of its time effectively unloaded. When loaded by a cylinder in its compression stroke, the rpm will attempt to fall commensurate with the torque it must deliver. However, inertia, which will be dominated by the flywheel's angular momentum, will attempt to smooth out any changes.
The first scope photo contains the 165 rpm gear motor's current waveform generated during the first second immediately after the starter switch was pressed. There was an initial, but brief, inrush current before the motor began spinning and creating a counter emf. The starter ran into its first compression load while the cranking period was still long compared with the cylinders' leak-down times. By the time it encountered the second one, its speed had increased, and the current peaks were beginning to stabilize at some 11 amps corresponding to torque peaks of 320 oz-in.
The engine's irregular power strokes can also be seen in the waveform. It shows the rear cylinder's power stroke occurring some 250 ms after that of the front cylinder. Inside the 560 ms 4-stroke cycle this corresponds to 321/411 degree firing intervals (compared with the 315/405 degree theoretical values). The small discrepancies are likely caused by the system's inertia.
The cranking speed is slow enough to discern the bifurcated loads presented by the two cylinders' closely spaced TDC's. The highest of the two peaks is created by the load presented by the cylinder in its power stroke whose piston is approaching TDC with both valves are closed. Just 45 degrees earlier, the piston in the other cylinder is also approaching TDC but in preparation for its intake stroke. The load created by its piston moving upward with its intake valve closed and its exhaust valve closing is also significant until the intake valve opens near TDC.
The second scope photo contains a snapshot of the same waveform several seconds later when it has had time to stabilize. The 560 ms 4-stroke cycle includes two crankshaft revolutions and corresponds to a 214 rpm cranking speed which, mysteriously, is 10% higher than should be possible.
If the torque curves are examined with these measurements in mind, it's obvious that the engine's 320 oz-in peak torque requirement should be easily satisfied by the 313 rpm version of the gear motor, while potentially doubling the cranking speed. The 437 rpm motor, on the other hand, might provide even more cranking speed, but it will come up about 65 oz-in shy of producing the required torque. Some of this shortfall may be compensated by the flywheel's additional angular momentum provided by the higher cranking speed. Therefore, I decided to replace the 165 rpm gear motor with the 437 rpm version.
The third scope photo contains the 437 rpm motor's current waveform during the first second after the starter button was pressed. The 20 amp inrush current along with a short stabilization period is still visible, but the bifurcated peaks have disappeared. This motor is spinning fast enough for the flywheel's angular momentum to carry it through the relatively small load variations created by the cylinder approaching its intake stroke. In steady state, the current peaks are just kissing their 20 amp maximums thanks to the flywheel covering the torque shortfall.
The fourth photo shows the current waveform after the cranking has had plenty of time to stabilize. The peak currents created by the rear cylinder in its power stroke are barely reaching 20 amps and those created by the front cylinder are a bit less which probably indicates some minor compression difference between the two cylinders. The 320 ms 4-stroke period corresponds to a 375 rpm cranking speed which is 30% lower than its 524 rpm no-load value.
Out of curiosity, I also temporarily installed the 315 rpm motor for testing. The peak currents were 17 amps as expected, but the cranking speed was only 240 rpm which was 40% lower than its 375 no-load value. The lower cranking speed also uncovered a portion of the bifurcated loads.
In the end, I re-installed the 437 rpm motor which seems most optimum of the three I had available to test. Actual engine starting tests using this motor showed the drill starter was still required to cold start the engine. The internal starter, though, was now able to restart the engine after it had been run for a while. The next step will be to revisit the carburetion. - Terry
Most everything on the engine now seems to be in reasonable working order except for the carburetion. The carburetor, a modified version of the one in the original drawings, has an idle port located downstream from its butterfly throttle. Its adjusting needle, located on the side of the carb body, is semi-accessible through the pushrod covers. It was arbitrarily opened one full turn when final testing began and hasn't been touched since.
A 1/4" Venturi, immediately behind the butterfly, receives fuel from the main (high speed) jet. Above idle, the engine's speed should be controlled by the butterfly by regulating the volume of air/fuel allowed into the engine. Fuel is drawn from the main jet and mixed with high speed air flowing through the Venturi and is controlled by a second needle that's accessible from below the carburetor bowl. It was also opened one turn at the beginning of testing and an additional turn sometime later.
Fuel from both the low and high speed circuits is often required to start a cold engine because of the fuel's temporarily low volatility. By restricting the amount of air allowed into the carb using a mechanical choke (this carb has one), or a finger over the carb's inlet, the Venturi increases the velocity of a smaller volume of air flowing through it, and this causes additional fuel to be drawn from the main jet. The resulting overly rich mixture is easier to ignite in a cold engine.
The current symptoms are that an unreasonably high cranking speed is required to start the engine, and once it does start, the throttle has no effect on speed. The engine idles indicating the idle circuitry is functional, but the symptoms are those of a non-functioning high speed circuit.
To begin troubleshooting, the carburetor was removed and a vacuum gage attached to the intake manifold. The gage indicated pulses 10 psi below atmospheric while cranking the engine with the internal starter. Although I don't have any comparative measurements from other engines, these results seemed reasonable for this engine's displacement and manifold volume. Admittedly, the 1/4" Venturi could be smaller, but it shouldn't be a show stopper.
After disassembling the carb and taking a close look at the main jet assembly, it seemed to me that with the needle's current taper, I should more likely be dealing with a rich and overly sensitive high speed jet instead of one that's not working at all.
I eventually sealed the bowl with a plexiglass cover using a bolt running through its center in order to simulate conditions inside the bowl with the pump running. At first, everything appeared normal with the recirculating loop maintaining a full level of fuel in the bowl. However, after adding a bit of dye to the fuel and playing with the pump voltage, I soon discovered a vortex beneath the surface encircling the bolt. This vortex was uncovering fuel from around the bolt's mid-section in the exact area where the inlet for the high speed jet would otherwise be trying to draw fuel. The inside corners of the bowl were actually full of fuel, and it was from one of these that the idle pickup tube had been happily feeding.
Although, with the variable DC converter powering it, I have plenty of control over the constant displacement fuel pump, the vortex was unaffected by speed. I spent a couple days playing with the diameters of the inlet and outlet hoses as well as dozens of screen and baffle designs with no success. I eventually realized my bowl design had a fatal flaw.
All the other fuel loops I've made have been three hose designs with remotely located carb bowls. In those loops, fuel was pumped into the bowl through an inlet located on its bottom. The return to-tank-line was also on the bottom, but an internal standpipe connected to it regulated the fuel level. The carb drew its fuel from a third tube located on the bottom of the bowl where turbulences were relatively minor.
The Knucklehead's carburetor with its integral bowl requires only two hoses. The inlet was installed mid-level in the bowl, but the return-to-tank line exited the bottom of the bowl so unused fuel would drain back into the tank. The bowl's only venting is through the return-to-tank hose which allows some latitude in setting the pump voltage. Excess volume is merely returned to the tank. Testing at the time of completion of the carburetor showed the fuel level inside the bowl was consistently controlled with the pump voltage. However, for those tests, the top of the bowl was left open while its bottom was sealed with a short screw and o-ring. What happened is that when the bowl was sealed to the carb, a vortex was created around the main jet that prevented fuel from entering it. Sealing the bowl to the plexiglass cover allowed the vortex to be seen spinning around the central screw although it was hidden beneath the surface of the fuel.
I hoped to salvage the bowl, so I removed and sealed up the original mid-bowl inlet so I could attach the pump to the existing bottom tube. A new outlet was then machined and installed as high up on the side of the bowl as possible. Testing showed the vortex was finally gone.
Testing on the engine showed I'm finally getting fuel - too much fuel - from the main jet. The internal starter was able to start the engine a couple times, but it's now running way too rich with lots of black smoke. The throttle is now having some effect, but a new main jet assembly is now going to be required. - Terry
Over the years I have experimented with, modified and built all types of carbs for my engines. Like the one on your engine I have made, or tried to make miniatures of the full sized versions. This would include air bleeds and accelerator pumps. No matter how I experimented I just couldn't get the desired overall running performance that I thought I should. There was a very gifted builder, Lee Root, who created some amazing I.C. engines and on one of them he used a very complex carb, somewhat replicating a Stromberg 97. Strictly I.C. magazine published a build article on it so I made one. Here again it would run my V-8 engine but not like I would have hoped for. There is a group of fellows that belong to the BAEM out in California and for some of their engines they use modified shall we call it, weed eater carbs. The small 2 cycle Walbro type. To use these carbs even in the modified state they require a fuel pump or delivery system. I played around with these, again not garnering the results I wanted.
I ended up building carbs based on the simple air bleed types found on RC airplanes. For each engine configuration I would build a carb that represented the full sized version but the internals were just basically an air bleed. I must say that I have achieved very good performance with these carbs, not perfect but very good.
Knowing that a lot of the properties of a running full sized engine don't scale well, air flow, centrifugal force etc. I have always wondered how the vacuum signals react when making ultra tiny porting in highly complex model carbs.
The one thing I have discovered over the years is that small is better when it comes to venturi size. My engines all have bores in the .875 to 1.25 range and my largest venturi size is about .210.
I wish I could help you but when it comes to small carbs it just takes a lot of experimentation.
Best of luck,
Separate names with a comma.