Amazing work 
Quarter Scale Merlin V-12
- Thread starter mayhugh1
- Start date
Help Support Home Model Engine Machinist Forum:
The last and most complex wheel case component to be machined was the rear main shaft. A disadvantage of saving it until the very end is that all of its dimensions must be spot on before it can be installed since it affects every moving part inside the wheel case. However, one of the reasons for saving it for last was that with all the other parts finished and in place, I felt I'd have the best chance of coming up with those dimensions.
Two especially critical ones involve the locations of the integral cam drive sprocket and the mounting flange that sets the meshes for the pump bevel and magneto helical gears. The timing chain doesn't flex laterally, and so the drive sprocket must end up aligned with the idler sprockets that center the chain inside its very narrow housing. By design, the bevel and helical gears are assembled as a ring gear set, and so the meshes with their mating gears is accomplished by properly locating their mounting flange. In addition, the shaft splines must end up aligned to the square hole for the spring drive to within a half degree or so.
Because of limited access and visibility inside the wheel case, the dimensions associated with the main shaft were not easily measured even as the wheel case was being built up. They had to be iteratively determined using trial-and-error machined spacers. Since the mesh of the pump gear was hidden from view by the gear itself, its depth had to be inferred from the backlash that it created in the coolant pump shaft when various spacers were placed in front of the bevel gear. Although all the shaft-related measurements taken over the past months seemed reasonably consistent, I felt it safer to assume they were only starting points for the shaft design. So, I changed my original plan and decided to make two shafts: a prototype shaft and a 'production' shaft. The prototype shaft was completely machined and its fit verified before work was started on the production shaft.
I initially set up the order of the machining operations so I could check the fit of the prototype shaft after each operation and verify or modify its design as I went along. I eventually discovered my original dimensions were accurate, but the order I had used to machine the prototype shaft had a number of problems. For example, the forces associated with broaching the square hole for the spring shaft affected the TIR of the prototype shaft. For the production shaft, I moved the broaching operation to the top of the list before any of the shaft's features were machined. A couple interference issues also showed up with the shank of the tool used to spot the front flange mounting holes as well as the end mill used to machine the cam drive sprocket. A different order for these operations solved the spotting tool issue, but I had to grind down the 1/8" shank of the 3/32" cutter used to mill the cam drive sprocket in order to prevent it from rubbing against the front splines.
As with the rest of the shafts in the wheel case, the main shaft was also machined from Stressproof - not so much for its dimensional stability as for its tensile strength and ductility. After initially skimming the outside diameter of the one inch workpiece, the rest of the operations were carried out in a four-jaw chuck to minimize any runout among the shaft's features. For the prototype shaft I mistakenly set up its work sheet using .938" as the finished o.d. for its workpiece instead of .984", and this guaranteed right from the start that it would end up in the scrap box after being completed.
For the production shaft, I faced both ends to bring it to its finished length, and then I drilled and reamed its center from each end so the square drive hole at its rear could be broached. The depth of the spring drive shaft inside the main shaft was carefully set using the i.d. reamer while the workpiece was back in the 4-jaw.
The o.d.'s of the major features were finished before cutting out the material between them. The more complex features, including the front splines, the cam drive sprocket, and the rear groove for the retainer clip were then machined. To reduce gear runout, the locating shoulders for the shaft's three gears were machined for very snug sharp corner fits. The rear of the shaft was also turned for a near-zero fit to its bearing plate bearing.
Two arrays of gear mounting holes had to be drilled/tapped into the two roughed-in flanges. The locations of these holes have been an interesting experience from the very beginning. The six tapped holes used to mount the bevel/helical ring gear set was originally arranged so the mounting screws would come out in the roots of the threads of the bevel gear. Now, the six clearance holes for these screws in the front flange had to be aligned with the twelve tapped holes used to mount the main drive gear on its rear flange. This alignment was needed so a hex wrench could pass through the tapped holes in the rear flange in order to tighten the mounting screws in the front flange. Both of these hole arrays ended up too close to the edges of their mounting flanges in the prototype shaft because of the initial machining error I made on its workpiece.
For the production shaft I also changed the method I had been using to align the workpiece in the mill while machining the shaft splines so they would be properly aligned to the spring shaft. For both the crankshaft adapter and the prototype shaft I had used pin gages as parallels between the fixed jaw of the vise and a tight-fitting square lathe tool blank inserted into the broached hole in the workpieces. I found that this alignment method had created about a two degree error between the adapter and the prototype shaft which I felt was too much. For the production shaft I inserted the spring shaft into the workpiece and then clamped a long horizontal bar to one of its flats which I then indicated to the mill's x-axis. With care, the repeatability of this technique appeared to be about +/-0.2 degrees. When I checked the error between the production shaft and the adapter it was still about two degrees telling me that most of the original alignment error had been a result of the adapter's fixturing. I machined a new adapter using the new alignment method, and found that the error between it and the production shaft had been reduced to essentially zero. My hat is off to anyone able to do this with a splined spring shaft.
A bronze bushing was pressed onto the front nose of the shaft and turned concentric with the rest of the shaft's features. The shaft will only rotate +/-5 degrees inside the crankshaft adapter, and so its real purpose is to keep the shaft centered in the adapter and on axis to the crankshaft.
The prototype shaft appeared to fit perfectly inside the wheel case with the bearing plate installed although, with so many close fits, its installation required a lot of patience. All seventeen meshed gears between the oil pump and the supercharger turned smoothly together as the prop shaft was rotated. I also exercised the gear train by driving it through both starter shafts. I threaded the timing chain down through the idler sprockets and around the cam drive sprocket to verify it had ended up in the correct position. After a good bit of testing I was finally happy that the prototype shaft fit and operated as intended. I then started all over and machined the production shaft to the exact measured dimensions of the prototype shaft.
The next step will be to finally assemble the entire wheel case. - Terry
Two especially critical ones involve the locations of the integral cam drive sprocket and the mounting flange that sets the meshes for the pump bevel and magneto helical gears. The timing chain doesn't flex laterally, and so the drive sprocket must end up aligned with the idler sprockets that center the chain inside its very narrow housing. By design, the bevel and helical gears are assembled as a ring gear set, and so the meshes with their mating gears is accomplished by properly locating their mounting flange. In addition, the shaft splines must end up aligned to the square hole for the spring drive to within a half degree or so.
Because of limited access and visibility inside the wheel case, the dimensions associated with the main shaft were not easily measured even as the wheel case was being built up. They had to be iteratively determined using trial-and-error machined spacers. Since the mesh of the pump gear was hidden from view by the gear itself, its depth had to be inferred from the backlash that it created in the coolant pump shaft when various spacers were placed in front of the bevel gear. Although all the shaft-related measurements taken over the past months seemed reasonably consistent, I felt it safer to assume they were only starting points for the shaft design. So, I changed my original plan and decided to make two shafts: a prototype shaft and a 'production' shaft. The prototype shaft was completely machined and its fit verified before work was started on the production shaft.
I initially set up the order of the machining operations so I could check the fit of the prototype shaft after each operation and verify or modify its design as I went along. I eventually discovered my original dimensions were accurate, but the order I had used to machine the prototype shaft had a number of problems. For example, the forces associated with broaching the square hole for the spring shaft affected the TIR of the prototype shaft. For the production shaft, I moved the broaching operation to the top of the list before any of the shaft's features were machined. A couple interference issues also showed up with the shank of the tool used to spot the front flange mounting holes as well as the end mill used to machine the cam drive sprocket. A different order for these operations solved the spotting tool issue, but I had to grind down the 1/8" shank of the 3/32" cutter used to mill the cam drive sprocket in order to prevent it from rubbing against the front splines.
As with the rest of the shafts in the wheel case, the main shaft was also machined from Stressproof - not so much for its dimensional stability as for its tensile strength and ductility. After initially skimming the outside diameter of the one inch workpiece, the rest of the operations were carried out in a four-jaw chuck to minimize any runout among the shaft's features. For the prototype shaft I mistakenly set up its work sheet using .938" as the finished o.d. for its workpiece instead of .984", and this guaranteed right from the start that it would end up in the scrap box after being completed.
For the production shaft, I faced both ends to bring it to its finished length, and then I drilled and reamed its center from each end so the square drive hole at its rear could be broached. The depth of the spring drive shaft inside the main shaft was carefully set using the i.d. reamer while the workpiece was back in the 4-jaw.
The o.d.'s of the major features were finished before cutting out the material between them. The more complex features, including the front splines, the cam drive sprocket, and the rear groove for the retainer clip were then machined. To reduce gear runout, the locating shoulders for the shaft's three gears were machined for very snug sharp corner fits. The rear of the shaft was also turned for a near-zero fit to its bearing plate bearing.
Two arrays of gear mounting holes had to be drilled/tapped into the two roughed-in flanges. The locations of these holes have been an interesting experience from the very beginning. The six tapped holes used to mount the bevel/helical ring gear set was originally arranged so the mounting screws would come out in the roots of the threads of the bevel gear. Now, the six clearance holes for these screws in the front flange had to be aligned with the twelve tapped holes used to mount the main drive gear on its rear flange. This alignment was needed so a hex wrench could pass through the tapped holes in the rear flange in order to tighten the mounting screws in the front flange. Both of these hole arrays ended up too close to the edges of their mounting flanges in the prototype shaft because of the initial machining error I made on its workpiece.
For the production shaft I also changed the method I had been using to align the workpiece in the mill while machining the shaft splines so they would be properly aligned to the spring shaft. For both the crankshaft adapter and the prototype shaft I had used pin gages as parallels between the fixed jaw of the vise and a tight-fitting square lathe tool blank inserted into the broached hole in the workpieces. I found that this alignment method had created about a two degree error between the adapter and the prototype shaft which I felt was too much. For the production shaft I inserted the spring shaft into the workpiece and then clamped a long horizontal bar to one of its flats which I then indicated to the mill's x-axis. With care, the repeatability of this technique appeared to be about +/-0.2 degrees. When I checked the error between the production shaft and the adapter it was still about two degrees telling me that most of the original alignment error had been a result of the adapter's fixturing. I machined a new adapter using the new alignment method, and found that the error between it and the production shaft had been reduced to essentially zero. My hat is off to anyone able to do this with a splined spring shaft.
A bronze bushing was pressed onto the front nose of the shaft and turned concentric with the rest of the shaft's features. The shaft will only rotate +/-5 degrees inside the crankshaft adapter, and so its real purpose is to keep the shaft centered in the adapter and on axis to the crankshaft.
The prototype shaft appeared to fit perfectly inside the wheel case with the bearing plate installed although, with so many close fits, its installation required a lot of patience. All seventeen meshed gears between the oil pump and the supercharger turned smoothly together as the prop shaft was rotated. I also exercised the gear train by driving it through both starter shafts. I threaded the timing chain down through the idler sprockets and around the cam drive sprocket to verify it had ended up in the correct position. After a good bit of testing I was finally happy that the prototype shaft fit and operated as intended. I then started all over and machined the production shaft to the exact measured dimensions of the prototype shaft.
The next step will be to finally assemble the entire wheel case. - Terry
There were a few loose ends to tie up before finally assembling the wheel case, and I used them also as an opportunity to assemble/disassemble the wheel case several more times to make sure everything went back together consistently. This is probably the most complicated assembly I've made, and I wanted to clear up any glitches while it's construction was still fresh in my mind.
The first task was to permanently install the adapter in the end of the crankshaft. The adapter is bolted to the crankshaft with six 3-48 SHCS's, but it's also intended to be bonded to the interior of the crankshaft. I didn't particularly like the idea of permanently installing it, but I was concerned about an area just behind its mounting flange for which I had questions about its ability to support the starting torque carried by the splines. Hopefully, the Loctite adds some much appreciated margin.
Since the adapter was machined for less than a half thousandth clearance inside the crankshaft, I used Loctite 609 which is a low viscosity retainer designed to augment press fits. With such a close fit, though, I was concerned about the Loctite setting up while I was still scrambling to align the mounting holes which were also reamed for minimum clearance. In order to avoid getting the adapter embarrassingly stuck in the crankshaft, I threaded a pair of temporary studs into the end of the crank in order to guide the wetted adapter into place. I had a plastic mallet on hand, just in case, but it wasn't needed. The steel mounting screws were locked in place with blue (medium strength) threadlocker. Blue threadlocker is overkill for number three screws; but the adapter, for better or worse, was to be a permanent part of the crankshaft.
Green (low strength) thread locker was used on some of the fasteners associated with the various shafts inside the wheel case as insurance against vibration. Before locking the threads on the pump idler bracket, the mesh of its idler gear was set with the oil pumps assembled in the lower crankcase. And, before this was done, I drilled the sump for a drain for which I made up a magnetic drain plug. This feature wasn't part of the documentation, but there was a pocket cast into the bottom of the sump that seemed to scream for a drain. I hope the screaming doesn't turn out to be a warning that I overlooked something more important that was supposed to have gone in there. Since the design of the oil pumps located on either side of the bottom surface of the oil pan made the use of gaskets or sealer risky, I filled the lower portion of the pan with oil and allowed it to sit over night to verify there were no leaks.
Forward thrust of the main shaft is ultimately limited by the depth of the shaft's splines inside the crankshaft adapter. This depth stop doesn't set the gear depths, though, as it was intended only to prevent the main shaft gear from rubbing against the wheel case housing during testing. During running, the shaft is held in position by a pair of spacers that straddle the inner race of the rear bearing located in the bearing plate. It's these spacers that set the engagement depths for the bevel and helical gears discussed previously. The widths of these spacers were determined during the wheel case build-up and verified using the prototype shaft. The front spacer is sandwiched between the rear face of the main gear and the inner race of the rear bearing. The rear spacer is held in place behind the rear bearing with a retainer clip in the groove at the rear of the shaft.
The wheel case is mostly modular which means that it was designed so it could be added to or removed from the crankcase as a completed assembly. However, there are three flange mounting screws located behind the countershaft that are accessible only from inside the disassembled case. These screws were probably added for extra support around the front of the countershaft which is especially stressed. Even though I had some misgivings about being able to align the inner and outer shafts while sliding the unit into place, it wasn't terribly difficult. When the wheel case is assembled to the crankcase for the final time, however, it will have to come apart once more for access to those three screws.
I rechecked the alignment of the timing chain inside the fully assembled wheel case. Some technique is required to get the chain threaded around the drive sprocket because of its restricted view and access. The drive sprocket is located directly over the top of the pump shaft bevel gear, and with the chain in place there is only about .020" clearance between the two. I eventually figured out how to snake the chain into place by pulling it through from the port side of the wheel case on the end of a waxed string. Because of its close proximity to the countershaft gear on the starboard side, it can only be pulled up vertically from that side. The chain isn't yet cut to length which makes working with it a little easier. It won't be shortened to length until final assembly, and threading it into place then with the housing in position is something I'm not looking forward to.
The final task was the fabrication of an oil nozzle for splash lubricating the gears and bearings inside the wheel case. The design of this part for what seemed like a simple task got away from me and grew into something that quickly became complicated. This two-part assembly is designed so oil will drip onto the main gear through a .018" diameter orifice, and windage should care of distributing the oil to the gears and bearings in front of the bearing plate. A second orifice on the end of an extended nozzle gets oil to the gear pair behind the bearing plate. There's an opening at the bottom of the wheel case to return waste oil to the sump. I formed the tiny orifice on the end of the extended nozzle by forming the tip of the soft copper tube around a drill bit.
I've included some assembly photos of the completed wheel case and rear end of the engine. I plan to start working on the magnetos next, and so the wheel case can remain in place until they are completed. Finishing the magnetos will complete all the work involving castings. The complexity of the drawing describing the magnetos is on the order of those for the wheel case, and so we'll see if my print reading skills have improved over the past several months. - Terry
The first task was to permanently install the adapter in the end of the crankshaft. The adapter is bolted to the crankshaft with six 3-48 SHCS's, but it's also intended to be bonded to the interior of the crankshaft. I didn't particularly like the idea of permanently installing it, but I was concerned about an area just behind its mounting flange for which I had questions about its ability to support the starting torque carried by the splines. Hopefully, the Loctite adds some much appreciated margin.
Since the adapter was machined for less than a half thousandth clearance inside the crankshaft, I used Loctite 609 which is a low viscosity retainer designed to augment press fits. With such a close fit, though, I was concerned about the Loctite setting up while I was still scrambling to align the mounting holes which were also reamed for minimum clearance. In order to avoid getting the adapter embarrassingly stuck in the crankshaft, I threaded a pair of temporary studs into the end of the crank in order to guide the wetted adapter into place. I had a plastic mallet on hand, just in case, but it wasn't needed. The steel mounting screws were locked in place with blue (medium strength) threadlocker. Blue threadlocker is overkill for number three screws; but the adapter, for better or worse, was to be a permanent part of the crankshaft.
Green (low strength) thread locker was used on some of the fasteners associated with the various shafts inside the wheel case as insurance against vibration. Before locking the threads on the pump idler bracket, the mesh of its idler gear was set with the oil pumps assembled in the lower crankcase. And, before this was done, I drilled the sump for a drain for which I made up a magnetic drain plug. This feature wasn't part of the documentation, but there was a pocket cast into the bottom of the sump that seemed to scream for a drain. I hope the screaming doesn't turn out to be a warning that I overlooked something more important that was supposed to have gone in there. Since the design of the oil pumps located on either side of the bottom surface of the oil pan made the use of gaskets or sealer risky, I filled the lower portion of the pan with oil and allowed it to sit over night to verify there were no leaks.
Forward thrust of the main shaft is ultimately limited by the depth of the shaft's splines inside the crankshaft adapter. This depth stop doesn't set the gear depths, though, as it was intended only to prevent the main shaft gear from rubbing against the wheel case housing during testing. During running, the shaft is held in position by a pair of spacers that straddle the inner race of the rear bearing located in the bearing plate. It's these spacers that set the engagement depths for the bevel and helical gears discussed previously. The widths of these spacers were determined during the wheel case build-up and verified using the prototype shaft. The front spacer is sandwiched between the rear face of the main gear and the inner race of the rear bearing. The rear spacer is held in place behind the rear bearing with a retainer clip in the groove at the rear of the shaft.
The wheel case is mostly modular which means that it was designed so it could be added to or removed from the crankcase as a completed assembly. However, there are three flange mounting screws located behind the countershaft that are accessible only from inside the disassembled case. These screws were probably added for extra support around the front of the countershaft which is especially stressed. Even though I had some misgivings about being able to align the inner and outer shafts while sliding the unit into place, it wasn't terribly difficult. When the wheel case is assembled to the crankcase for the final time, however, it will have to come apart once more for access to those three screws.
I rechecked the alignment of the timing chain inside the fully assembled wheel case. Some technique is required to get the chain threaded around the drive sprocket because of its restricted view and access. The drive sprocket is located directly over the top of the pump shaft bevel gear, and with the chain in place there is only about .020" clearance between the two. I eventually figured out how to snake the chain into place by pulling it through from the port side of the wheel case on the end of a waxed string. Because of its close proximity to the countershaft gear on the starboard side, it can only be pulled up vertically from that side. The chain isn't yet cut to length which makes working with it a little easier. It won't be shortened to length until final assembly, and threading it into place then with the housing in position is something I'm not looking forward to.
The final task was the fabrication of an oil nozzle for splash lubricating the gears and bearings inside the wheel case. The design of this part for what seemed like a simple task got away from me and grew into something that quickly became complicated. This two-part assembly is designed so oil will drip onto the main gear through a .018" diameter orifice, and windage should care of distributing the oil to the gears and bearings in front of the bearing plate. A second orifice on the end of an extended nozzle gets oil to the gear pair behind the bearing plate. There's an opening at the bottom of the wheel case to return waste oil to the sump. I formed the tiny orifice on the end of the extended nozzle by forming the tip of the soft copper tube around a drill bit.
I've included some assembly photos of the completed wheel case and rear end of the engine. I plan to start working on the magnetos next, and so the wheel case can remain in place until they are completed. Finishing the magnetos will complete all the work involving castings. The complexity of the drawing describing the magnetos is on the order of those for the wheel case, and so we'll see if my print reading skills have improved over the past several months. - Terry
- Joined
- Dec 9, 2015
- Messages
- 327
- Reaction score
- 350
Amazing! Everything looks just perfect! Immaculate! How do you do it?
I'm curious about how long something may take to create? Ie your rear shaft assy in post 362? Just to give me an idea on a project like this.
I'm curious about how long something may take to create? Ie your rear shaft assy in post 362? Just to give me an idea on a project like this.
pickleford75
Active Member
- Joined
- Oct 4, 2015
- Messages
- 40
- Reaction score
- 47
Gasp! I cannot wait to see it run..... That is some of the finest workmanship I've ever seen, thank you for sharing.
The full-size Merlin used a pair of magnetos to fire the six pairs of spark plugs in each cylinder head. The port-side magneto fired the outside plugs in both heads, and the starboard-side magneto fired the inside plugs. In case of a magneto failure, this arrangement provided for at least one of the plugs in each cylinder to continue firing. Since there's only one plug per cylinder in the Quarter Scale, the wiring was simplified by arranging for each magneto to fire all the plugs on its particular side of the engine.
The Quarter Scale's magneto housings were cast to look very similar to those on the full-size engine. A design for a distributor, compatible with a conventional spark ignition, was also provided since designing a functional magneto was beyond the scope of the project. Fitting a fairly complex distributor inside the magneto housing created some challenges involving the high voltage that arose from the scaling. The requirements for high voltage isolation didn't scale with the physical sizes of the voltage carrying components, and so the limited space inside the housings even with half the number of plugs created problems for at least one early builder.
I used a translator app to follow Gunnar's online diary of his distributor testing experiences that were plagued with crossfires and internal arcing to ground. His entries seemed to end abruptly during his efforts to modify portions of the original distributor design, and it's not clear from his website that he was ever able to overcome all the problems and get his engine to run:
http://www.123hjemmeside.dk/gunnarsorensen/47386583
The distributor was later redesigned by Dynamotive, and some of Gunnar's improvements may have been incorporated into the new design which included an improved contact block. A photo in Gunnar's diary shows the tower electrodes in his distributor sitting above the contact block similarly to those in the full-scale magneto. In the Quarter Scale, the path lengths between those tower electrodes and the grounded enclosure were marginal insurance against arc-overs. The re-design moved the electrodes deep into the contact block to increase the lengths of the surface paths between them and ground. Other improvements were made to the rotor, and John Ramm may have been the first to implement the new design in a successfully running engine. In the latest design that both John and I received, the original mechanical points used by Gunnar were also replaced with a solid state trigger.
The documentation includes a drawing sheet that clearly shows most of the machining required on the castings. A second sheet contains the distributor design, but several details are left up to the builder. I eventually had to model the distributor components in a 3-d assembly before I could begin to understand and appreciate its operation.
The main shaft in each distributor is driven at 3/2 the crankshaft speed through the Oldham couplers on either side of the wheel case. A timing disk on this shaft opens an aperture between a stationary magnet and a Hall sensor twice per revolution. The rotor is gear reduced to the timing shaft and spins at 1/3 its speed or, equivalently, half the speed of the crankshaft. The result is that for every two crankshaft rotations the rotor has six consecutive angular positions defined by trigger events that are 60 distributor degrees apart. The contact block converts these six angular positions into a 2x3 linear array of tower electrodes - one for each of the six plugs in the distributor's cylinder bank. Internal conductive paths inside the rotor and contact block created using pressed-in metal pins distribute the coil firing voltage to each plug in each bank according to the firing order 1-4-2-6-3-5. A typical firing position, in this case for cylinder number four, is shown in the cross sectional view of the distributor's partial assembly model.
The first step in construction was to machine the castings according to the drawing I had so I could have some parts in my hands to help me visualize the design. While boring the housings for the timing shaft I discovered the BOM hadn't specified a flanged bearing for the inner main bearing as shown on the drawing. While reordering these I decided to convert the outer bearing to a flanged type also since the distributor drawing seemed to describe an overly complicated mounting scheme for a flangeless bearing.
The castings for the top covers that I received seem to be less desirable early versions which are the same as those used by Gunnar but different from those in John's engine. John's covers include integral troughs for the plug wires that look similar to those in the photos of the full-size Merlins I've seen. The trough-less castings I received would be awkward to adapt, and so I'll likely machine a pair of replacements. In the full-size Merlins the plug wires were routed between the magnetos and the spark plugs through metal conduits mounted to the sides of the cylinder blocks. These tubes nicely protected the wires, but their real purpose was most likely to shield the aircraft's onboard radio receiver from the RF interference they created.
The rotor spins on a two-piece shaft in a housing which is an integral part of the contact block. The outboard portion of the shaft is metal, and since it forms part of the contact to the coil's secondary, it is at full coil potential. The inboard portion, including the rotor's driven gear, is Delrin so the 'hot' half of the shaft is kept well isolated from ground. In the stock design the rear rotor bearing is a simple Delrin sleeve. John, instead, used ball bearings on both ends of his rotors to reduce the mechanical load carried by the acetal gear and shaft. I plan to do something similar since precision bearings may allow me to reduce the size of the rotor's air gaps that are in series with each spark plug. I've worked out a reasonable solution for the isolated inboard bearing, but the outboard bearing which will be 'hot' is still a work in progress. I machined the retainer blocks for the inboard bearings and bolted them into place in the main castings. They will be line-bored later along with the rotor housing.
Every air gap in the high voltage path between the coil and the spark plug dissipates coil energy proportional to the width of the gap. This dissipation, over time, erodes the ends of the conductors on both sides of the gap causing them to grow even wider. The tiny plug gaps associated with the Mini-Viper plugs used in the Quarter Scale make these air gaps especially significant. Anything that can be done to reduce their size and number will improve the reliability and reduce the maintenance of the ignition system. - Terry
The Quarter Scale's magneto housings were cast to look very similar to those on the full-size engine. A design for a distributor, compatible with a conventional spark ignition, was also provided since designing a functional magneto was beyond the scope of the project. Fitting a fairly complex distributor inside the magneto housing created some challenges involving the high voltage that arose from the scaling. The requirements for high voltage isolation didn't scale with the physical sizes of the voltage carrying components, and so the limited space inside the housings even with half the number of plugs created problems for at least one early builder.
I used a translator app to follow Gunnar's online diary of his distributor testing experiences that were plagued with crossfires and internal arcing to ground. His entries seemed to end abruptly during his efforts to modify portions of the original distributor design, and it's not clear from his website that he was ever able to overcome all the problems and get his engine to run:
http://www.123hjemmeside.dk/gunnarsorensen/47386583
The distributor was later redesigned by Dynamotive, and some of Gunnar's improvements may have been incorporated into the new design which included an improved contact block. A photo in Gunnar's diary shows the tower electrodes in his distributor sitting above the contact block similarly to those in the full-scale magneto. In the Quarter Scale, the path lengths between those tower electrodes and the grounded enclosure were marginal insurance against arc-overs. The re-design moved the electrodes deep into the contact block to increase the lengths of the surface paths between them and ground. Other improvements were made to the rotor, and John Ramm may have been the first to implement the new design in a successfully running engine. In the latest design that both John and I received, the original mechanical points used by Gunnar were also replaced with a solid state trigger.
The documentation includes a drawing sheet that clearly shows most of the machining required on the castings. A second sheet contains the distributor design, but several details are left up to the builder. I eventually had to model the distributor components in a 3-d assembly before I could begin to understand and appreciate its operation.
The main shaft in each distributor is driven at 3/2 the crankshaft speed through the Oldham couplers on either side of the wheel case. A timing disk on this shaft opens an aperture between a stationary magnet and a Hall sensor twice per revolution. The rotor is gear reduced to the timing shaft and spins at 1/3 its speed or, equivalently, half the speed of the crankshaft. The result is that for every two crankshaft rotations the rotor has six consecutive angular positions defined by trigger events that are 60 distributor degrees apart. The contact block converts these six angular positions into a 2x3 linear array of tower electrodes - one for each of the six plugs in the distributor's cylinder bank. Internal conductive paths inside the rotor and contact block created using pressed-in metal pins distribute the coil firing voltage to each plug in each bank according to the firing order 1-4-2-6-3-5. A typical firing position, in this case for cylinder number four, is shown in the cross sectional view of the distributor's partial assembly model.
The first step in construction was to machine the castings according to the drawing I had so I could have some parts in my hands to help me visualize the design. While boring the housings for the timing shaft I discovered the BOM hadn't specified a flanged bearing for the inner main bearing as shown on the drawing. While reordering these I decided to convert the outer bearing to a flanged type also since the distributor drawing seemed to describe an overly complicated mounting scheme for a flangeless bearing.
The castings for the top covers that I received seem to be less desirable early versions which are the same as those used by Gunnar but different from those in John's engine. John's covers include integral troughs for the plug wires that look similar to those in the photos of the full-size Merlins I've seen. The trough-less castings I received would be awkward to adapt, and so I'll likely machine a pair of replacements. In the full-size Merlins the plug wires were routed between the magnetos and the spark plugs through metal conduits mounted to the sides of the cylinder blocks. These tubes nicely protected the wires, but their real purpose was most likely to shield the aircraft's onboard radio receiver from the RF interference they created.
The rotor spins on a two-piece shaft in a housing which is an integral part of the contact block. The outboard portion of the shaft is metal, and since it forms part of the contact to the coil's secondary, it is at full coil potential. The inboard portion, including the rotor's driven gear, is Delrin so the 'hot' half of the shaft is kept well isolated from ground. In the stock design the rear rotor bearing is a simple Delrin sleeve. John, instead, used ball bearings on both ends of his rotors to reduce the mechanical load carried by the acetal gear and shaft. I plan to do something similar since precision bearings may allow me to reduce the size of the rotor's air gaps that are in series with each spark plug. I've worked out a reasonable solution for the isolated inboard bearing, but the outboard bearing which will be 'hot' is still a work in progress. I machined the retainer blocks for the inboard bearings and bolted them into place in the main castings. They will be line-bored later along with the rotor housing.
Every air gap in the high voltage path between the coil and the spark plug dissipates coil energy proportional to the width of the gap. This dissipation, over time, erodes the ends of the conductors on both sides of the gap causing them to grow even wider. The tiny plug gaps associated with the Mini-Viper plugs used in the Quarter Scale make these air gaps especially significant. Anything that can be done to reduce their size and number will improve the reliability and reduce the maintenance of the ignition system. - Terry
Last edited:
- Joined
- Dec 31, 2010
- Messages
- 813
- Reaction score
- 215
Hi Terry:
All I can say (as usual) is Wow!. Fantastic work. I admire your ability to work through problems (and your ability to figure out you have one ahead of time).
I'm wondering what your plan is for coils and drivers? Perhaps a couple of CDI units?
I'm looking forward to seeing how the ignition system works out. This is always a problem for model builders because the management of high voltage is of those things that doesn't scale.
Great work.
Sage
All I can say (as usual) is Wow!. Fantastic work. I admire your ability to work through problems (and your ability to figure out you have one ahead of time).
I'm wondering what your plan is for coils and drivers? Perhaps a couple of CDI units?
I'm looking forward to seeing how the ignition system works out. This is always a problem for model builders because the management of high voltage is of those things that doesn't scale.
Great work.
Sage
Dave,
I'll probably use a pair of CDIs. I bought a pair plus a spare of Roy Sholl's original CDI units for this build last year after hearing that that he was losing his supplier for the boards he was selling. I visited his website last night since it's probably time to bite the bullet and order the dozen ZR2 plugs I need and noticed he is now carrying the Rcxcel units.
Thanks for the kind comment. - Terry
I'll probably use a pair of CDIs. I bought a pair plus a spare of Roy Sholl's original CDI units for this build last year after hearing that that he was losing his supplier for the boards he was selling. I visited his website last night since it's probably time to bite the bullet and order the dozen ZR2 plugs I need and noticed he is now carrying the Rcxcel units.
Thanks for the kind comment. - Terry
I started work on the internals of the distributors with the most difficult parts - the contact blocks with their integral rotor housings. Since they looked like they would probably require more thought and setup time than machining time, I launched a pair of spares as well. I used white Delrin because of its excellent machinability and resistance to high voltage breakdown (500V/mil). I'd rather have been working with black Delrin, as it looks more at home on an engine, but the carbon used to color it adds so much uncertainty to its electrical properties that DuPont doesn't even specify them for the black material. Even though it makes great looking distributor caps, black Delrin really isn't a good choice around high voltage.
I had some white round drops in my scrap collection, but they required their first machining step to be a 4-jaw offset turning operation in order to get the blocks started in the centers of their workpieces. The circular ends of the rotor housings were lathe turned before milling the rectangular portions of the contact blocks. The blocks will eventually be bolted inside the housings, but I also machined them for snug fits inside my particular castings so they'd be consistently re-mounted after the boring operations for the rotors and bearings.
After drilling the contact blocks for the tower electrodes, I machined some trial electrodes from 360 brass. In particular, I was looking for the right amount of interference to keep the 1/16" diameter pins in position during the boring operation for the rotor housing but not bend them during the pressing operations. I eventually settled on .004". Due to workpiece deflection, turning the long skinny pins without tailstock support created a taper going in the wrong direction, but it was correctable with a simple compensation program run during the finishing pass on my 9x20 CNC lathe.
I plan to use 1/8" diameter black 20 kV silicone wire for the plug wires; but I wasn't comfortable with, or maybe I didn't understand, what was being shown in distributor drawing to secure them within the contact blocks. John held his wires in place with individual Delrin sleeves, and I did something that probably ended up being very similar. I machined slotted Delrin collets for the wires. Testing showed that pressing a wire-filled collet into the block with just the right width slot compressed the silicone insulation enough to retain the wire so it would pass my 'tug' test. The collets were straight forward enough to machine, but cutting the slot was much more difficult than I expected. The i.d., o.d., and the slot width of the collet all interact to affect the amount of force that's required to press a wire-filled collet into the contact block. The collets won't be pressed into position until after the distributors are completed and the contact blocks have been weakened by the boring operation for their rotors. It's important to not get carried away with the amount of seating force since a hard pressing operation could distort the rotor housing and upset the rotor air gap. After spending an evening scrapping dozens of collets, I finally came up a reasonably consistent slotting tool made by bolting two single-edge razor blades together on either side of a 'goldilocks' spacer. With the collet resting in a wood v-block and a tiny wood filler dowel through its center, rapping the tool with a small mallet usually punched out the right slot. Somehow, I managed to get through all the parts I made without spilling blood.
The plug wires will eventually have their ends stripped and the strands wound into a circular 'bird nest' below the bottom end of the collet. After the collet is pressed into place, the strands will be captured between the bottom end of the collet and the flat head of the pin. In addition to strain relief, this should also provide a reliable electrical connection between the two.
After pressing in the tower electrodes, the contact blocks and the rear bearing blocks were assembled in their housings so a shaft alignment hole could be drilled/reamed through them. This hole was used to pick up the centers for the boring operations for the rotor bearings as well as the opening for the rotor itself. Unfortunately, it wasn't possible to do these operations in a single setup with the components installed in the housings. The rear bearing blocks were bored for their flanged bearings and tapped for 0-80 retaining screws. A fixture was then machined to support the contact blocks so the rotor housings could be bored without distortion.
After assembly, the angular positions of the rotors will be hidden from view inside their housings, but these positions will be needed when timing the engine. An index hole, shown drilled through the outer end of each rotor housing, will align with a similar hole in the rotor to determine the rotors position. The use of these holes weren't documented, and my CAD model showed that with the coordinates provided alignment would actually occur 7-1/2 degrees before full alignment with plug number one. I altered the location of the hole in the housing to set the alignment at dead center, and I reduced its diameter from 1/8" to 1/16" since the hole was uncomfortably close to my high voltage coil contact. Also not mentioned on the drawing is that fact that the locations of the housing holes must be mirrored between the port-side and starboard-side distributors since they rotate in opposite directions. I didn't realize this until I had already drilled mine, and so I had press in filler plugs and re-drill. The plug sequences in the tower arrays must also be mirrored.
Finally, the recessed pocket for the outer rotor bearing was bored, and this finally completed the machining on the contact blocks.
In order to verify the accuracy of the boring operations, I turned a trial rotor blank for a .003" theoretical rotor air gap. I pressed in a test shaft and spun the combination in all four rotor housings while mounted in their corresponding magneto housings. The rotor spun freely in all cases, and there didn't appear to be any interference with any of the housings including the spares. Since I didn't expect to be able to reliably do so, I didn't try to reduce the clearance any further even though a .003" rotor air gap is a whopping 25% of what the plug gap will probably end up being.
The rotor-side design of the high voltage coil contact that I'm currently working with is somewhat different from what is shown in the distributor drawing, and it's shown in the exploded rotor assembly diagram. The stationary side of the coil contact will be designed later after the rotor is machined and in place.
In order to maximize the path length to ground along the outer surface of the rotor housing, the rotor housing was set back away from the metal enclosure, and the front bearing was recessed into the inboard surface of the rotor housing. Only its inner race is visible from the outboard side. Even though this bearing is at full coil potential, damaging current should flow only through the inner race and not through the balls themselves. An insulated end cap will be machined later to cover the coil connection and seal against the outboard side of the rotor housing. - Terry
I had some white round drops in my scrap collection, but they required their first machining step to be a 4-jaw offset turning operation in order to get the blocks started in the centers of their workpieces. The circular ends of the rotor housings were lathe turned before milling the rectangular portions of the contact blocks. The blocks will eventually be bolted inside the housings, but I also machined them for snug fits inside my particular castings so they'd be consistently re-mounted after the boring operations for the rotors and bearings.
After drilling the contact blocks for the tower electrodes, I machined some trial electrodes from 360 brass. In particular, I was looking for the right amount of interference to keep the 1/16" diameter pins in position during the boring operation for the rotor housing but not bend them during the pressing operations. I eventually settled on .004". Due to workpiece deflection, turning the long skinny pins without tailstock support created a taper going in the wrong direction, but it was correctable with a simple compensation program run during the finishing pass on my 9x20 CNC lathe.
I plan to use 1/8" diameter black 20 kV silicone wire for the plug wires; but I wasn't comfortable with, or maybe I didn't understand, what was being shown in distributor drawing to secure them within the contact blocks. John held his wires in place with individual Delrin sleeves, and I did something that probably ended up being very similar. I machined slotted Delrin collets for the wires. Testing showed that pressing a wire-filled collet into the block with just the right width slot compressed the silicone insulation enough to retain the wire so it would pass my 'tug' test. The collets were straight forward enough to machine, but cutting the slot was much more difficult than I expected. The i.d., o.d., and the slot width of the collet all interact to affect the amount of force that's required to press a wire-filled collet into the contact block. The collets won't be pressed into position until after the distributors are completed and the contact blocks have been weakened by the boring operation for their rotors. It's important to not get carried away with the amount of seating force since a hard pressing operation could distort the rotor housing and upset the rotor air gap. After spending an evening scrapping dozens of collets, I finally came up a reasonably consistent slotting tool made by bolting two single-edge razor blades together on either side of a 'goldilocks' spacer. With the collet resting in a wood v-block and a tiny wood filler dowel through its center, rapping the tool with a small mallet usually punched out the right slot. Somehow, I managed to get through all the parts I made without spilling blood.
The plug wires will eventually have their ends stripped and the strands wound into a circular 'bird nest' below the bottom end of the collet. After the collet is pressed into place, the strands will be captured between the bottom end of the collet and the flat head of the pin. In addition to strain relief, this should also provide a reliable electrical connection between the two.
After pressing in the tower electrodes, the contact blocks and the rear bearing blocks were assembled in their housings so a shaft alignment hole could be drilled/reamed through them. This hole was used to pick up the centers for the boring operations for the rotor bearings as well as the opening for the rotor itself. Unfortunately, it wasn't possible to do these operations in a single setup with the components installed in the housings. The rear bearing blocks were bored for their flanged bearings and tapped for 0-80 retaining screws. A fixture was then machined to support the contact blocks so the rotor housings could be bored without distortion.
After assembly, the angular positions of the rotors will be hidden from view inside their housings, but these positions will be needed when timing the engine. An index hole, shown drilled through the outer end of each rotor housing, will align with a similar hole in the rotor to determine the rotors position. The use of these holes weren't documented, and my CAD model showed that with the coordinates provided alignment would actually occur 7-1/2 degrees before full alignment with plug number one. I altered the location of the hole in the housing to set the alignment at dead center, and I reduced its diameter from 1/8" to 1/16" since the hole was uncomfortably close to my high voltage coil contact. Also not mentioned on the drawing is that fact that the locations of the housing holes must be mirrored between the port-side and starboard-side distributors since they rotate in opposite directions. I didn't realize this until I had already drilled mine, and so I had press in filler plugs and re-drill. The plug sequences in the tower arrays must also be mirrored.
Finally, the recessed pocket for the outer rotor bearing was bored, and this finally completed the machining on the contact blocks.
In order to verify the accuracy of the boring operations, I turned a trial rotor blank for a .003" theoretical rotor air gap. I pressed in a test shaft and spun the combination in all four rotor housings while mounted in their corresponding magneto housings. The rotor spun freely in all cases, and there didn't appear to be any interference with any of the housings including the spares. Since I didn't expect to be able to reliably do so, I didn't try to reduce the clearance any further even though a .003" rotor air gap is a whopping 25% of what the plug gap will probably end up being.
The rotor-side design of the high voltage coil contact that I'm currently working with is somewhat different from what is shown in the distributor drawing, and it's shown in the exploded rotor assembly diagram. The stationary side of the coil contact will be designed later after the rotor is machined and in place.
In order to maximize the path length to ground along the outer surface of the rotor housing, the rotor housing was set back away from the metal enclosure, and the front bearing was recessed into the inboard surface of the rotor housing. Only its inner race is visible from the outboard side. Even though this bearing is at full coil potential, damaging current should flow only through the inner race and not through the balls themselves. An insulated end cap will be machined later to cover the coil connection and seal against the outboard side of the rotor housing. - Terry
I didn't spend much time in the shop during the past week while two of our grandchildren were visiting, but I did manage enough late night hours to finish up the rotors.
If my brain had been plugged in before starting, I'd have realized that the large driven rotor gear could easily have been machined as an integral part of the rotor, itself. Since it wasn't, though, the first thing I did was to machine a Delrin prototype gear from which I sliced off the two gears that I needed. I'd never before used a gear cutter on plastic, and even though I thought the tooth cutting operation went smoothly, the tedious de-burring required afterward on both sides of every tiny tooth did not. My zero rake gear cutter didn't play well with Delrin, and I eventually had to lap the faces of the gears in order to clean them up. If I had it to do it over I might try grinding a single point cutter with some rake.
Since I'd already made two spare rotor housings, I decided to also start a couple extra rotors since there was some unforgiving drilling ahead related to the high voltage conductors. For minimum TIR, I planned for all the finishing operations to be completed in the same lathe setup. My goal was to at least duplicate the .003" air gap I had been able to achieve earlier with the test rotor.
I faced an end of each rotor blank and then rough-turned the diameters to leave .020" excess stock for finishing. The blanks were then moved over to my mill's fourth axis where a pair of radial holes were drilled for the rotor electrodes and a pair of flats were machined on their cylindrical sides. These flats serve two purposes. First, they open up the volume around the rotor and reduce the amount of unneeded/unwanted dielectric material rotating very close to the housing. More importantly, though, they provide a convenient way to hold the cylindrical parts in a vise at a precise and consistent angular orientation while their ends are machined. After pressing short lengths of 1/16" diameter brass rod into the holes for the rotor electrodes, the parts were returned to the lathe where their o.d.'s were finished and their centers bored for press-fit shafts.
The ends of each rotor were then machined. The inside end was simply drilled and tapped for four 0-80 flat head screws for mounting the driven gear. The outer end machining was a little more complex because of the need to electrically connect the rotor electrodes to the coil contact. I found some 20 AWG dead soft silver wire in a local craft store that I used to connect the rotor electrodes to a contact disk machined on the stainless steel rotor shaft. A pair of .032" diameter axial holes were drilled through the outer end of each rotor so they would intersect the sides of the radial electrodes and penetrate about half their diameters. A short piece of the wire was inserted into each of these holes, and the tops were bent over into a shallow slot milled across the outer end of the rotor. The wires were cut slightly long so the pressure from the contact disk forced them into the drilled spots in the rotor electrodes. The folded-over wires in the slots were sandwiched tightly against the contact disk and held with a pair flat head screws for a reliable connection. An ohmmeter verified the final continuities. Although silver is very close to both brass and stainless steel on the galvanic reaction chart, I thoroughly cleaned and removed all traces of cutting fluids from the wire and wire holes before final assembly to insure I didn't leave an electrolyte behind that might corrode the electrical connections.
The rotor shaft is actually two separate shafts. The outer shaft was machined from stainless steel and will be at coil potential when a plug fires. I modified the shaft design provided in the distributor drawing to accommodate the ball bearings that I added to the rotor housing. I also integrated the contact disk with the shaft and added the screws to secure it to the rotor. The inner race of the outside bearing is captured between the end of this shaft and a custom phosphor bronze contact screwed into its end. The bronze contact was machined with hex flats for a nut driver, and a screwdriver slot was cut into the rear of the shaft so the pair can be easily tightened. The contact pair on the port side is left-hand threaded, and the starboard-side contact pair is right-hand threaded.
The inner shaft was machined from Delrin to maintain isolation of the 'hot' outer shaft from chassis ground. Since the rotor is held in place in its housing by the outer bearing, the rear shaft merely supports the inner end of the rotor in its bearing to maintain concentricity with its housing. The end of this shaft was threaded for a puller since after assembly it and its bearing must be pulled before the rotor can be disassembled.
Both completed rotor assemblies spun freely in their housings with no detectable rubbing. Except for the stationary coil-side contact, the high voltage portions of the distributors should now be completed. A potential problem with the distributor's basic design, though, is created by its tightly enclosed rotor. Ozone will be continually produced by the arc at the rotor's air gap, and if not adequately vented it can fill the space between the rotor and its housing after only a brief period of operation. If the ozone is allowed to build up in this space it will ionize and crossfires to unwanted tower electrodes can occur due to the lower plug firing voltage requirements of the cylinders that are not under compression. An unshielded/unprotected Hall device inside the distributor can also be affected. These gas discharges typically have a softer appearance but are only visible through a transparent distributor cap. It this does become an issue for me the discharges may be visible through the thin white Delrin wall of the rotor housing if tested in a room with subdued lighting.
The next step is to add the low voltage triggering mechanisms so the high voltage sections can be tested. - Terry
If my brain had been plugged in before starting, I'd have realized that the large driven rotor gear could easily have been machined as an integral part of the rotor, itself. Since it wasn't, though, the first thing I did was to machine a Delrin prototype gear from which I sliced off the two gears that I needed. I'd never before used a gear cutter on plastic, and even though I thought the tooth cutting operation went smoothly, the tedious de-burring required afterward on both sides of every tiny tooth did not. My zero rake gear cutter didn't play well with Delrin, and I eventually had to lap the faces of the gears in order to clean them up. If I had it to do it over I might try grinding a single point cutter with some rake.
Since I'd already made two spare rotor housings, I decided to also start a couple extra rotors since there was some unforgiving drilling ahead related to the high voltage conductors. For minimum TIR, I planned for all the finishing operations to be completed in the same lathe setup. My goal was to at least duplicate the .003" air gap I had been able to achieve earlier with the test rotor.
I faced an end of each rotor blank and then rough-turned the diameters to leave .020" excess stock for finishing. The blanks were then moved over to my mill's fourth axis where a pair of radial holes were drilled for the rotor electrodes and a pair of flats were machined on their cylindrical sides. These flats serve two purposes. First, they open up the volume around the rotor and reduce the amount of unneeded/unwanted dielectric material rotating very close to the housing. More importantly, though, they provide a convenient way to hold the cylindrical parts in a vise at a precise and consistent angular orientation while their ends are machined. After pressing short lengths of 1/16" diameter brass rod into the holes for the rotor electrodes, the parts were returned to the lathe where their o.d.'s were finished and their centers bored for press-fit shafts.
The ends of each rotor were then machined. The inside end was simply drilled and tapped for four 0-80 flat head screws for mounting the driven gear. The outer end machining was a little more complex because of the need to electrically connect the rotor electrodes to the coil contact. I found some 20 AWG dead soft silver wire in a local craft store that I used to connect the rotor electrodes to a contact disk machined on the stainless steel rotor shaft. A pair of .032" diameter axial holes were drilled through the outer end of each rotor so they would intersect the sides of the radial electrodes and penetrate about half their diameters. A short piece of the wire was inserted into each of these holes, and the tops were bent over into a shallow slot milled across the outer end of the rotor. The wires were cut slightly long so the pressure from the contact disk forced them into the drilled spots in the rotor electrodes. The folded-over wires in the slots were sandwiched tightly against the contact disk and held with a pair flat head screws for a reliable connection. An ohmmeter verified the final continuities. Although silver is very close to both brass and stainless steel on the galvanic reaction chart, I thoroughly cleaned and removed all traces of cutting fluids from the wire and wire holes before final assembly to insure I didn't leave an electrolyte behind that might corrode the electrical connections.
The rotor shaft is actually two separate shafts. The outer shaft was machined from stainless steel and will be at coil potential when a plug fires. I modified the shaft design provided in the distributor drawing to accommodate the ball bearings that I added to the rotor housing. I also integrated the contact disk with the shaft and added the screws to secure it to the rotor. The inner race of the outside bearing is captured between the end of this shaft and a custom phosphor bronze contact screwed into its end. The bronze contact was machined with hex flats for a nut driver, and a screwdriver slot was cut into the rear of the shaft so the pair can be easily tightened. The contact pair on the port side is left-hand threaded, and the starboard-side contact pair is right-hand threaded.
The inner shaft was machined from Delrin to maintain isolation of the 'hot' outer shaft from chassis ground. Since the rotor is held in place in its housing by the outer bearing, the rear shaft merely supports the inner end of the rotor in its bearing to maintain concentricity with its housing. The end of this shaft was threaded for a puller since after assembly it and its bearing must be pulled before the rotor can be disassembled.
Both completed rotor assemblies spun freely in their housings with no detectable rubbing. Except for the stationary coil-side contact, the high voltage portions of the distributors should now be completed. A potential problem with the distributor's basic design, though, is created by its tightly enclosed rotor. Ozone will be continually produced by the arc at the rotor's air gap, and if not adequately vented it can fill the space between the rotor and its housing after only a brief period of operation. If the ozone is allowed to build up in this space it will ionize and crossfires to unwanted tower electrodes can occur due to the lower plug firing voltage requirements of the cylinders that are not under compression. An unshielded/unprotected Hall device inside the distributor can also be affected. These gas discharges typically have a softer appearance but are only visible through a transparent distributor cap. It this does become an issue for me the discharges may be visible through the thin white Delrin wall of the rotor housing if tested in a room with subdued lighting.
The next step is to add the low voltage triggering mechanisms so the high voltage sections can be tested. - Terry
Last edited:
Terry: Most excellent work as usual. I read with interest your statements concerning the ionization of ozone gasses being produced inside of a distributor and it being the cause of possible cross-firing. I had never given this any thought and have never experienced a problem that I am aware of on any of my many engines. Maybe that's because most of my designs are not machined to the precision you are so capable of and plenty of ventilation takes place? Makes perfect sense that this would occur and could be a source of trouble. Very interesting and something else to consider with a poorly performing engine. Thanks for enlightening an old friend. Ron Colonna
Ron,
You're very welcome. If you're a GM nut (I'm not) you might remember they had to ventilate even the huge distributor caps they used for their HEI ignitions back in the 70's. Terry
You're very welcome. If you're a GM nut (I'm not) you might remember they had to ventilate even the huge distributor caps they used for their HEI ignitions back in the 70's. Terry
- Joined
- Dec 31, 2010
- Messages
- 813
- Reaction score
- 215
Hmm. Thanks for pointing that out Terry. It might explain some of the issues I have with the Howell V8 ignition. I made the (relatively small) distributor cap with a pretty tight fit over the base plate. Since it needs work anyway for other reasons I'll keep the vent idea in mind.
I also picked up on your earlier comment about the black delrin not being as good an insulator (as the white). Another item to consider.
I have an HEI ignition on my old Corvette. How did they ventilate it. When they went to HEI there was no longer the sloppy slide up side plate to adjust the points. Which would have been a good leak for fresh air in the old distributors.
Thanks
Sage
I also picked up on your earlier comment about the black delrin not being as good an insulator (as the white). Another item to consider.
I have an HEI ignition on my old Corvette. How did they ventilate it. When they went to HEI there was no longer the sloppy slide up side plate to adjust the points. Which would have been a good leak for fresh air in the old distributors.
Thanks
Sage
Dave,
The one HEI distributor I had experience with had an opening in the base of the distributor housing. There was a wire mesh over it, probably to keep out insects. Lots of drag racers who added high energy after market ignitions drilled a hole or two in the tops of their caps. Many for whom this solved problems thought they were venting carbon dust, but it was really ozone they were venting. Some Ford street racers created issues when they added after-market ignition systems and after market distributor caps because of their typical smaller sizes and smaller internal volumes. Drilling vent holes in the distributor caps on street cars with stock ignitions wasn't a good idea because of possible condensation problems during start-up. Most of my experience was with Ford engines, and I never understood why some of the hot rod parts manufacturers, (MSD, for example, if I remember correctly) made their caps smaller with smaller volumes. It might have been for larger aftermarket air cleaner clearance since the Ford engines had front mounted distributors. Of course, the time frame I'm talking about was during the 60's and 70's. Today, with COP, ozone inside a distributor is no longer an issue and a good thing, too, with the extremely high energy ignitions used today. - Terry
The one HEI distributor I had experience with had an opening in the base of the distributor housing. There was a wire mesh over it, probably to keep out insects. Lots of drag racers who added high energy after market ignitions drilled a hole or two in the tops of their caps. Many for whom this solved problems thought they were venting carbon dust, but it was really ozone they were venting. Some Ford street racers created issues when they added after-market ignition systems and after market distributor caps because of their typical smaller sizes and smaller internal volumes. Drilling vent holes in the distributor caps on street cars with stock ignitions wasn't a good idea because of possible condensation problems during start-up. Most of my experience was with Ford engines, and I never understood why some of the hot rod parts manufacturers, (MSD, for example, if I remember correctly) made their caps smaller with smaller volumes. It might have been for larger aftermarket air cleaner clearance since the Ford engines had front mounted distributors. Of course, the time frame I'm talking about was during the 60's and 70's. Today, with COP, ozone inside a distributor is no longer an issue and a good thing, too, with the extremely high energy ignitions used today. - Terry
Last edited:
Terry,
A while ago I built a "mad scientist" Jacob's ladder for my grandson - one of those things where an arc propels itself upward between two electrodes - blows itself out at the top and starts again at the bottom.
What came as a surprise was that it generated copious amounts of Ozone, Nitric Oxide and Nitric Acid - the whole thing was built in a perspex tube (for Safety) - the atmosphere inside rapidly turned brown (Nitric Acid and Nitric Oxide gas).
When I was fooling around with the exposed electrodes at the prototype stage you could smell it and it rapidly became unbearable / downright dangerous and everything in the vicinity rusted.
So now I only run it outside on Halloween (it has a ventilating fan) or flush it with Argon if I want to run it indoors (without the fan)..
I would imagine an H.T. ignition is going to do much the same thing on a smaller scale because of the very short arc length.
Just a thought.
Regards,
Ken
A while ago I built a "mad scientist" Jacob's ladder for my grandson - one of those things where an arc propels itself upward between two electrodes - blows itself out at the top and starts again at the bottom.
What came as a surprise was that it generated copious amounts of Ozone, Nitric Oxide and Nitric Acid - the whole thing was built in a perspex tube (for Safety) - the atmosphere inside rapidly turned brown (Nitric Acid and Nitric Oxide gas).
When I was fooling around with the exposed electrodes at the prototype stage you could smell it and it rapidly became unbearable / downright dangerous and everything in the vicinity rusted.
So now I only run it outside on Halloween (it has a ventilating fan) or flush it with Argon if I want to run it indoors (without the fan)..
I would imagine an H.T. ignition is going to do much the same thing on a smaller scale because of the very short arc length.
Just a thought.
Regards,
Ken
Ken,
That corrosion you saw was also an anomaly seen inside a lot of after market high energy distributors by irate street racers who thought the 'high performance' distributor caps weren't sealing properly and allowing moisture to rust the internals. After a few thousand miles some distributor baseplates looked like they had been sitting open to the weather in a junkyard for a decade. The manufacturers told them to drill holes in the tops of their caps, but this sounded so counterintuitive that many ignored the device and went on to a different manufacturer's system. In fairness to the racers, the manufacturers were late in figuring out what was really going on because they hadn't properly tested the longitivtiy of their equipment for the street market that they were going after. Most users would have been much better off to have stuck with the OEM equipment. - Terry
That corrosion you saw was also an anomaly seen inside a lot of after market high energy distributors by irate street racers who thought the 'high performance' distributor caps weren't sealing properly and allowing moisture to rust the internals. After a few thousand miles some distributor baseplates looked like they had been sitting open to the weather in a junkyard for a decade. The manufacturers told them to drill holes in the tops of their caps, but this sounded so counterintuitive that many ignored the device and went on to a different manufacturer's system. In fairness to the racers, the manufacturers were late in figuring out what was really going on because they hadn't properly tested the longitivtiy of their equipment for the street market that they were going after. Most users would have been much better off to have stuck with the OEM equipment. - Terry
The trigger components for the distributor are located on the distributor driveshaft below the rotor. The distributor driveshafts are driven by the crankshaft through Oldham couplers on either side of the wheel case. Since the triggers are magnetic, the driveshafts were machined from aluminum to avoid influencing the magnetic fields at the Hall sensors mounted near their ends.
The trigger disks in the distributors of the three engines that I've previously built have all used multiple magnets mounted on nonferrous disks rotating in front of single Hall sensors. The sensors were located inside the housings of the distributors but were protected from ionized discharges by spark shields covering them. All three distributors worked reliably, and the only ignition-related issue I ever encountered was with the Howell V-4 distributor whose neodymium magnets slowly weakened over time and reduced the dwell for its transistorized ignition. The magnets' loss of strength was most likely a result of the extremely close proximity of the opposing fields of the distributor's irregularly placed magnet pairs.
The Merlin's magneto housing has room only for a tiny trigger disk, and so multiple magnets weren't an option. Instead, the Quarter Scale uses a single stationary magnet with a ferrous disk rotating between it and the sensor. A pair of holes drilled through the disk exposes the magnet to the sensor as it rotates. As a bonus, the sensor ends up mounted outside the rotor housing and safely away from the thunderstorm at the tip of the rotor.
The trigger disk is driven at 1-1/2 times the speed of the crankshaft, but a gear on the driveshaft drives the rotor at half the speed of the crankshaft. The phasing of the rotor with respect to the crankshaft is easily adjusted thanks to a pair of spacers that grip the drive gear after the assembly is tightened together by a bolt on the end of the driveshaft. Previously drilled index holes in the rotor and rotor housing align when the rotor points to tower electrode number one, and the rotor can be temporarily pinned in this position during alignment. The trigger disk is simultaneously gripped by the same spacers, and this allows the timing of plug number one's firing to also be adjusted.
The ideal material for this type of trigger disk has a high magnetic permeability so it will easily conduct and shunt the flux of the source magnet away from the sensor when the trigger is OFF. It also has a low remanence which means that over time it will acquire little magnetism of its own from the nearby source magnet since this would change the threshold of the sensor. Silicon or 'magnetic' steel is best used in these applications and is found in the laminated cores of transformers and motors. Magnetic steel is typically laminated for ac applications in order to reduce power losses, but laminating provides little benefit in a dc application. I considered salvaging the steel from an old transformer, but its laminations were too thin, and I didn't want to deal with the extra complexity involved with stacking them. A good second choice would probably have been wrought iron, but it can be gummy to machine, and I didn't have any on hand. My third choice was hot-rolled steel. Quality hot-rolled mild steel typically has half the remanence of cold-rolled steel, and even less remanence than most hardenable alloys.
The trigger disks in the Merlin's distributors require a pair of diametrically placed aperture holes to handle the two rows of tower electrodes. In order to reduce timing jitter, the holes must be matched and carefully placed. I made my disks by turning a rod to final o.d., plunge-milling the apertures from one end using an end mill, parting off a pair of disks, and then surface grinding their faces. I verified there were no visible asymmetries by stacking the aligned and anti-aligned pair back-to-back under magnification on a snug fitting rod.
There's probably a good reason why the apertures may appear to be excessively wide to someone who has built a more conventional distributor. Although no ignition details were provided in the documentation, the disk was likely designed to provide dwell for a transistorized ignition. With the disk rotating at 1-1/2 times the speed of the crankshaft instead of at the more familiar half speed of the rotor shaft, the apertures had to be widened accordingly. Since I plan to use CDI modules, the wide apertures probably weren't necessary.
The distributor drawing included a timing adjuster which is a two part Delrin assembly that supports the magnet and sensor while the trigger disk spins between them. Neither a particular magnet nor sensor were specified, but the timing adjuster drawing showed a 1/8" diameter by 1/8" deep flat bottom hole for a magnet. I trial tested an 1/8" long neodymium magnet with an Optek OH090U sensor that I had on hand and found that the sensor would fire through the disk apertures using the .125" magnet/sensor separation shown in the drawing. Details about how a sensor was to be consistently supported in the mount provided was unclear in the drawing. Since I had several Optek sensors on hand, I designed a mount around them with a tight fitting pocket for the sensor and a strain relief for a cable.
I've standardized on Futaba J male servo connector cables for the sensors in all my engines. These are readily available from RC hobby shops, and I've found the connectors to be reliable and easy to work with. Using a standard connector allows the use of common test fixtures which in the past have helped speed testing and troubleshooting. A minor problem is that the order of the three sensor leads doesn't match the order of the color-coded wires in the Futaba 3-wire flat cable. When soldering a sensor to the cable, one of the leads has to cross over the other two, and so I included a milled trough in the sensor mount to allow this to be cleanly done. After soldering the cable to the sensor in its mount, a few drops of silicone windshield sealer stabilizes and insulates the connections. A cover bears down on the wired assembly inside the sensor mount and the pair are held together with the two mounting screws and a short length of shrink tubing on the adjuster arm.
After machining the first part of the timing adjuster assembly according to the drawing, I discovered a significant interference between it and the magneto housing. I modified its fit for use in testing but changed the design before machining the final parts. The test part ended up being put to good use, though.
With all the parts for the timing adjusters finally completed, I assembled the first set to check its functionality but without high voltage. The sensor fired through the aperture holes but would not turn off. The sensor's hysteresis combined with the huge flux level flowing through the wide apertures and around the edge of the disk prevented the sensor from turning OFF. The OH090U sensor is the most sensitive part in an Optek Hall sensor line which also includes the OH180U and OH360U. (The number in the part number is actually the typical flux density, in Gauss, required to turn the sensor ON.)
A shorter 1/10" long magnet gave about the same result, and an even shorter 1/16" long magnet appeared to work but with little margin. I decided to machine a new set of thicker sensor mounts so I could experimentally reduce their thicknesses and find an optimum magnet-sensor separation. After some testing I found that .175" was near optimum using the 1/10" long magnet and the OH090U sensor, and so the whole batch of new mounts were modified accordingly.
During final engine assembly when the rotor and timing disk are initially phased to the crankshaft and locked into position with the spacers, the timing can be adjusted over a limited range by rotating the arm of the sensor mount. The adjuster rotates the magnet/sensor pair with respect to the timing disk apertures causing the plug firing angle to change. After adjustment, it's locked in place with its own screw. The adjustment range is ultimately limited by the widths of the rotor and tower electrodes. It's important for a portion of the rotor electrode to remain overlapped with the corresponding tower electrode over the entire usable range. For the electrode and rotor diameters used, the available timing range works out to a maximum of +/-9 distributor degrees or +/-18 crankshaft degrees around the electrodes' center. Since timing changes are made with respect to the trigger disk which rotates 1.5 times faster than the crankshaft, one degree of crankshaft timing change will require 1.5 degrees of timing adjuster rotation. The total adjustment range of the arm is 38 degrees. - Terry
The trigger disks in the distributors of the three engines that I've previously built have all used multiple magnets mounted on nonferrous disks rotating in front of single Hall sensors. The sensors were located inside the housings of the distributors but were protected from ionized discharges by spark shields covering them. All three distributors worked reliably, and the only ignition-related issue I ever encountered was with the Howell V-4 distributor whose neodymium magnets slowly weakened over time and reduced the dwell for its transistorized ignition. The magnets' loss of strength was most likely a result of the extremely close proximity of the opposing fields of the distributor's irregularly placed magnet pairs.
The Merlin's magneto housing has room only for a tiny trigger disk, and so multiple magnets weren't an option. Instead, the Quarter Scale uses a single stationary magnet with a ferrous disk rotating between it and the sensor. A pair of holes drilled through the disk exposes the magnet to the sensor as it rotates. As a bonus, the sensor ends up mounted outside the rotor housing and safely away from the thunderstorm at the tip of the rotor.
The trigger disk is driven at 1-1/2 times the speed of the crankshaft, but a gear on the driveshaft drives the rotor at half the speed of the crankshaft. The phasing of the rotor with respect to the crankshaft is easily adjusted thanks to a pair of spacers that grip the drive gear after the assembly is tightened together by a bolt on the end of the driveshaft. Previously drilled index holes in the rotor and rotor housing align when the rotor points to tower electrode number one, and the rotor can be temporarily pinned in this position during alignment. The trigger disk is simultaneously gripped by the same spacers, and this allows the timing of plug number one's firing to also be adjusted.
The ideal material for this type of trigger disk has a high magnetic permeability so it will easily conduct and shunt the flux of the source magnet away from the sensor when the trigger is OFF. It also has a low remanence which means that over time it will acquire little magnetism of its own from the nearby source magnet since this would change the threshold of the sensor. Silicon or 'magnetic' steel is best used in these applications and is found in the laminated cores of transformers and motors. Magnetic steel is typically laminated for ac applications in order to reduce power losses, but laminating provides little benefit in a dc application. I considered salvaging the steel from an old transformer, but its laminations were too thin, and I didn't want to deal with the extra complexity involved with stacking them. A good second choice would probably have been wrought iron, but it can be gummy to machine, and I didn't have any on hand. My third choice was hot-rolled steel. Quality hot-rolled mild steel typically has half the remanence of cold-rolled steel, and even less remanence than most hardenable alloys.
The trigger disks in the Merlin's distributors require a pair of diametrically placed aperture holes to handle the two rows of tower electrodes. In order to reduce timing jitter, the holes must be matched and carefully placed. I made my disks by turning a rod to final o.d., plunge-milling the apertures from one end using an end mill, parting off a pair of disks, and then surface grinding their faces. I verified there were no visible asymmetries by stacking the aligned and anti-aligned pair back-to-back under magnification on a snug fitting rod.
There's probably a good reason why the apertures may appear to be excessively wide to someone who has built a more conventional distributor. Although no ignition details were provided in the documentation, the disk was likely designed to provide dwell for a transistorized ignition. With the disk rotating at 1-1/2 times the speed of the crankshaft instead of at the more familiar half speed of the rotor shaft, the apertures had to be widened accordingly. Since I plan to use CDI modules, the wide apertures probably weren't necessary.
The distributor drawing included a timing adjuster which is a two part Delrin assembly that supports the magnet and sensor while the trigger disk spins between them. Neither a particular magnet nor sensor were specified, but the timing adjuster drawing showed a 1/8" diameter by 1/8" deep flat bottom hole for a magnet. I trial tested an 1/8" long neodymium magnet with an Optek OH090U sensor that I had on hand and found that the sensor would fire through the disk apertures using the .125" magnet/sensor separation shown in the drawing. Details about how a sensor was to be consistently supported in the mount provided was unclear in the drawing. Since I had several Optek sensors on hand, I designed a mount around them with a tight fitting pocket for the sensor and a strain relief for a cable.
I've standardized on Futaba J male servo connector cables for the sensors in all my engines. These are readily available from RC hobby shops, and I've found the connectors to be reliable and easy to work with. Using a standard connector allows the use of common test fixtures which in the past have helped speed testing and troubleshooting. A minor problem is that the order of the three sensor leads doesn't match the order of the color-coded wires in the Futaba 3-wire flat cable. When soldering a sensor to the cable, one of the leads has to cross over the other two, and so I included a milled trough in the sensor mount to allow this to be cleanly done. After soldering the cable to the sensor in its mount, a few drops of silicone windshield sealer stabilizes and insulates the connections. A cover bears down on the wired assembly inside the sensor mount and the pair are held together with the two mounting screws and a short length of shrink tubing on the adjuster arm.
After machining the first part of the timing adjuster assembly according to the drawing, I discovered a significant interference between it and the magneto housing. I modified its fit for use in testing but changed the design before machining the final parts. The test part ended up being put to good use, though.
With all the parts for the timing adjusters finally completed, I assembled the first set to check its functionality but without high voltage. The sensor fired through the aperture holes but would not turn off. The sensor's hysteresis combined with the huge flux level flowing through the wide apertures and around the edge of the disk prevented the sensor from turning OFF. The OH090U sensor is the most sensitive part in an Optek Hall sensor line which also includes the OH180U and OH360U. (The number in the part number is actually the typical flux density, in Gauss, required to turn the sensor ON.)
A shorter 1/10" long magnet gave about the same result, and an even shorter 1/16" long magnet appeared to work but with little margin. I decided to machine a new set of thicker sensor mounts so I could experimentally reduce their thicknesses and find an optimum magnet-sensor separation. After some testing I found that .175" was near optimum using the 1/10" long magnet and the OH090U sensor, and so the whole batch of new mounts were modified accordingly.
During final engine assembly when the rotor and timing disk are initially phased to the crankshaft and locked into position with the spacers, the timing can be adjusted over a limited range by rotating the arm of the sensor mount. The adjuster rotates the magnet/sensor pair with respect to the timing disk apertures causing the plug firing angle to change. After adjustment, it's locked in place with its own screw. The adjustment range is ultimately limited by the widths of the rotor and tower electrodes. It's important for a portion of the rotor electrode to remain overlapped with the corresponding tower electrode over the entire usable range. For the electrode and rotor diameters used, the available timing range works out to a maximum of +/-9 distributor degrees or +/-18 crankshaft degrees around the electrodes' center. Since timing changes are made with respect to the trigger disk which rotates 1.5 times faster than the crankshaft, one degree of crankshaft timing change will require 1.5 degrees of timing adjuster rotation. The total adjustment range of the arm is 38 degrees. - Terry
Last edited:
Hi Terry,
I'd like to echo everyone else saying that your 12
Cyl build is just unreal, something to aspire to. Your radial builds as well I'll be referring to when I start on the Hodgson 14 cyl.
I have a small request though, can you put spaces into your threads? I struggle to keep my place reading your post.
There's a ton of great information, a lot Im still learning too so I'm trying to take it all in.
Either way keep it up, I love seeing the progress you make.
Kyle
I'd like to echo everyone else saying that your 12
Cyl build is just unreal, something to aspire to. Your radial builds as well I'll be referring to when I start on the Hodgson 14 cyl.
I have a small request though, can you put spaces into your threads? I struggle to keep my place reading your post.
There's a ton of great information, a lot Im still learning too so I'm trying to take it all in.
Either way keep it up, I love seeing the progress you make.
Kyle
