Swifty
Well-Known Member
They look fantastic Terry, got to get me a CNC mill.
Paul.
Paul.
Last edited:
Another Radial - this time 18 Cylinders
- Thread starter mayhugh1
- Start date
Help Support Home Model Engine Machinist Forum:
I received a pair of Roy Sholl's CDI ignition modules with his 30k spark/sec modification. The pcb is different from the one used in his standard module that I purchased a year ago, and in this model Roy doesn't offer his LED trigger indicator option. Roy told me that the major change in the design of this higher speed module is a reduction in the size of the discharge capacitor from .47uF to .22uF. The output voltage of the dc/dc converter is the same as the standard module and is nominally 300V with a worst-case minimum of 125V. This combination should result in a nominal output of 10 mJ and a minimum of just under 2 mJ. Based on my research and the experience I gained during my 9 cylinder radial build http://www.homemodelenginemachinist.com/f31/my-hodgson-9-radial-final-assembly-20397/index10.html
2 mJ should be fine.
Basically, the CDI ignition is a dc-to-dc converter which steps the 6V battery voltage up to 300Vdc. This voltage is then used to charge a capacitor. The trigger signal from the Hall device is ac coupled to the gate of an SCR that is used to discharge this capacitor through the primary of a step-up transformer. This discharge spike multiplied by the turns ratio of the output transformer generates the voltage used to fire the plug. With a CDI the spark rise time is generally faster and of shorter duration than the spark from a coil type ignition. Very high output voltages and impressively long sparks are often available from CDI ignitions. Even though these fast rise times and high voltages can sometimes be beneficial with oil fouled plugs they can also be a nuisance since they can become a major source of cross fires in whole ignition system. It is the energy in the spark and NOT just the firing voltage that is important when igniting the air/fuel mixture in an engine.
Most of the pcb real estate is taken up by the dc-dc converter components. This includes an additional fairly large transformer and switching transistor that make up the oscillator section of the converter. I noticed the switching transistor used in the standard module has been morphed into a pair of paralleled non-heatsink'd transistors in the 30k sparks/sec module. (I don't think this is good practice for discrete bipolar devices without emitter degeneration.) The critical output transformer which handles the high voltage is a nicely potted component. It's the art involved with making these transformers that keeps me from making my own ignition modules.
The relatively small gate trigger signal combined with the inevitable EMI of the dc-dc converter means that noise is a potential problem especially in my application where I will be using two of these modules. Therefore I decided to individually package the ignitions and to keep them separated from one another with short connections to the engine and extra 6V supply filtering. I also added toggle switches and led trigger indicators so I can separately power only the Hall device and LEDs during timing set-up and verification without dealing with the high voltage. It is important to not allow the ignition to fire without a plug connected to the high voltage output or cumulative damage to the secondary winding of the output transformer can occur. My testing shows that connecting/disconnecting the 6V supply to the module even with the Hall device disconnected can create an output spark.
A trigger led is a really nice indicator that lets the user know the Hall device is alive and well. The hand sketch in one of the photos shows the circuitry I added to the front of each CDI module. I breadboarded these additional components around one of the CDI modules to make a test bed for my completed distributors and to help me visualize the final packaging for the ignition modules. I tested the distributors in a dark room with the transparent caps so I could see the light flashes from the arcs in the rotor-to-HV tower contact gaps and judge their sizes relative to the spark plug gaps. They were reasonably consistent but not perfect indicating I had still managed to accumulate a bit of runout between the rotor and tower inserts inside the caps.
The Hall device in one of my finished distributors didn't trigger the LED indicator at all. I discovered that I hadn't removed enough of the protective shrink tubing around the pig-tailed sensor, and so it wasn't properly seated in its cavity in the Delrin cover plate and ended up too far away from the magnets on the trigger disk. After some re-work it functioned properly.
I used SolidWorks to design a pcb for my front-end circuitry and my CAM program to generate the cut paths for a pair of circuit boards. This was my first attempt at pcb design with non-dedicated pcb software and I soon decided it isn't the best way to create circuit board traces. But, it isn't too unreasonable for a simple board. The additional circuitry I added in combination with my own requirement to be able to easily access the electronics in place on the engine's eventual firewall (I plan to make one similar to the one I did for my H-9) resulted in completed ignition modules larger than I really wanted. I spent a frustrating week massaging the module packaging to minimize its size and to come up with a form factor that wouldn't look like a pair of warts stuck on the engine. I ended up scrapping a set of completed (and populated) pcb's in order to shrink them another factor of two. I went through three complete enclosure designs before I called it quits. I can't tell anymore if I finally ended up with something I'm happy with or if I just decided it was time to move on.
Each machined ABS plastic enclosure contains two trigger LEDs with one each to be visible from either side of the firewall. It also contains the HV enable toggle switch and a bulkhead-mounted mating connector for the distributor Hall device. To reduce the module size I decided to move the 6v toggle switch to a later-to-be determined position on the firewall. An LED indicator was added to each module so I can easily tell when the CDI is energized. I used a tiny breadboard area available on the CDI pcb to mount the LED and its resistor.
The enclosure is designed so a portion of it containing a bulkhead Hall connector will protrude through the firewall. I'm using Robert Sholl's pigtailed sensors which come terminated with a Futaba J series RC type female connector. I replaced the stock connector with the male equivalent since the female is much easier to mount as a bulkhead connector. In addition, I formed a shrink tubing boot around the male pigtailed connector to keep oil, etc. out of the contacts.
All the circuitry will be easily accessible under the enclosure's lid without removing the module from the firewall. In order to improve the appearance a bit, I machined the covers for a look reminiscent of the old style MSD ignition modules. The packaging is very tight, but the electronics is modular and readily maintainable. Final soldering of several components is a process as they protrude off the circuit boards and into recesses in the bulkhead and removable cover. Both distributors were thoroughly tested with both modules and 'burned-in' for some 20 minutes. I'm really glad this portion of the build is finally completed; and, hopefully, it won't have to be re-visited later.
The next step in this build is to start on the cylinders and heads. The CAD time will likely take several weeks as I don't yet know how they'll look, but I'm hoping I'll know them when I see them. Since long stretches of design time without making chips tends to bore me, I plan to break up this CAD phase with the monotonous task of machining some twenty pistons. I plan to use my H9 pistons in this engine and so I already have their design. If the cylinder and head designs aren't completed by the time the pistons are finished, I'll also start on some of the final display components. -Terry
2 mJ should be fine.
Basically, the CDI ignition is a dc-to-dc converter which steps the 6V battery voltage up to 300Vdc. This voltage is then used to charge a capacitor. The trigger signal from the Hall device is ac coupled to the gate of an SCR that is used to discharge this capacitor through the primary of a step-up transformer. This discharge spike multiplied by the turns ratio of the output transformer generates the voltage used to fire the plug. With a CDI the spark rise time is generally faster and of shorter duration than the spark from a coil type ignition. Very high output voltages and impressively long sparks are often available from CDI ignitions. Even though these fast rise times and high voltages can sometimes be beneficial with oil fouled plugs they can also be a nuisance since they can become a major source of cross fires in whole ignition system. It is the energy in the spark and NOT just the firing voltage that is important when igniting the air/fuel mixture in an engine.
Most of the pcb real estate is taken up by the dc-dc converter components. This includes an additional fairly large transformer and switching transistor that make up the oscillator section of the converter. I noticed the switching transistor used in the standard module has been morphed into a pair of paralleled non-heatsink'd transistors in the 30k sparks/sec module. (I don't think this is good practice for discrete bipolar devices without emitter degeneration.) The critical output transformer which handles the high voltage is a nicely potted component. It's the art involved with making these transformers that keeps me from making my own ignition modules.
The relatively small gate trigger signal combined with the inevitable EMI of the dc-dc converter means that noise is a potential problem especially in my application where I will be using two of these modules. Therefore I decided to individually package the ignitions and to keep them separated from one another with short connections to the engine and extra 6V supply filtering. I also added toggle switches and led trigger indicators so I can separately power only the Hall device and LEDs during timing set-up and verification without dealing with the high voltage. It is important to not allow the ignition to fire without a plug connected to the high voltage output or cumulative damage to the secondary winding of the output transformer can occur. My testing shows that connecting/disconnecting the 6V supply to the module even with the Hall device disconnected can create an output spark.
A trigger led is a really nice indicator that lets the user know the Hall device is alive and well. The hand sketch in one of the photos shows the circuitry I added to the front of each CDI module. I breadboarded these additional components around one of the CDI modules to make a test bed for my completed distributors and to help me visualize the final packaging for the ignition modules. I tested the distributors in a dark room with the transparent caps so I could see the light flashes from the arcs in the rotor-to-HV tower contact gaps and judge their sizes relative to the spark plug gaps. They were reasonably consistent but not perfect indicating I had still managed to accumulate a bit of runout between the rotor and tower inserts inside the caps.
The Hall device in one of my finished distributors didn't trigger the LED indicator at all. I discovered that I hadn't removed enough of the protective shrink tubing around the pig-tailed sensor, and so it wasn't properly seated in its cavity in the Delrin cover plate and ended up too far away from the magnets on the trigger disk. After some re-work it functioned properly.
I used SolidWorks to design a pcb for my front-end circuitry and my CAM program to generate the cut paths for a pair of circuit boards. This was my first attempt at pcb design with non-dedicated pcb software and I soon decided it isn't the best way to create circuit board traces. But, it isn't too unreasonable for a simple board. The additional circuitry I added in combination with my own requirement to be able to easily access the electronics in place on the engine's eventual firewall (I plan to make one similar to the one I did for my H-9) resulted in completed ignition modules larger than I really wanted. I spent a frustrating week massaging the module packaging to minimize its size and to come up with a form factor that wouldn't look like a pair of warts stuck on the engine. I ended up scrapping a set of completed (and populated) pcb's in order to shrink them another factor of two. I went through three complete enclosure designs before I called it quits. I can't tell anymore if I finally ended up with something I'm happy with or if I just decided it was time to move on.
Each machined ABS plastic enclosure contains two trigger LEDs with one each to be visible from either side of the firewall. It also contains the HV enable toggle switch and a bulkhead-mounted mating connector for the distributor Hall device. To reduce the module size I decided to move the 6v toggle switch to a later-to-be determined position on the firewall. An LED indicator was added to each module so I can easily tell when the CDI is energized. I used a tiny breadboard area available on the CDI pcb to mount the LED and its resistor.
The enclosure is designed so a portion of it containing a bulkhead Hall connector will protrude through the firewall. I'm using Robert Sholl's pigtailed sensors which come terminated with a Futaba J series RC type female connector. I replaced the stock connector with the male equivalent since the female is much easier to mount as a bulkhead connector. In addition, I formed a shrink tubing boot around the male pigtailed connector to keep oil, etc. out of the contacts.
All the circuitry will be easily accessible under the enclosure's lid without removing the module from the firewall. In order to improve the appearance a bit, I machined the covers for a look reminiscent of the old style MSD ignition modules. The packaging is very tight, but the electronics is modular and readily maintainable. Final soldering of several components is a process as they protrude off the circuit boards and into recesses in the bulkhead and removable cover. Both distributors were thoroughly tested with both modules and 'burned-in' for some 20 minutes. I'm really glad this portion of the build is finally completed; and, hopefully, it won't have to be re-visited later.
The next step in this build is to start on the cylinders and heads. The CAD time will likely take several weeks as I don't yet know how they'll look, but I'm hoping I'll know them when I see them. Since long stretches of design time without making chips tends to bore me, I plan to break up this CAD phase with the monotonous task of machining some twenty pistons. I plan to use my H9 pistons in this engine and so I already have their design. If the cylinder and head designs aren't completed by the time the pistons are finished, I'll also start on some of the final display components. -Terry
- Joined
- Jun 24, 2010
- Messages
- 2,445
- Reaction score
- 968
Wow, you are a man of many talents!
One thing I've wondered about on the big multi-cylinder model engines are the ignition wires themselves. What did you end up using on your 9-cyl? I see that braided metal sheathed wire used on RC gassers, maybe has more to do with shielding RF interference to radio gear. Do all those wires congregating together at your distributer present any special issues? Interested to hear what you came up with & why, obviously its working well. Is there such a thing as small gauge solid/graphite core wire like cars use these days?
Peter,
The wire I used on my 9 cylinder is 20KV wire purchased from S/S Engineering (Robert Sholl). I plan to use it in this engine also. It is just a tinned braided core with a thick silicone insulator. It is black and about 1/8" in diameter. The white stuff in the above test bed photo is quite a bit thicker and it rated at 25KV and is just some stuff I had purchased in a surplus sale many years ago. I was planning to use it on this engine but I'm going to have so many wires and that wire is so thick that I don't think it will look right. And, I didn't have enough of it anyway.
The self inductance of the plug wires and their stray capacitances form a resonant circuit which can be excited by the fast edges of the voltages in the secondary of the output transformer in both a CDI or a coil-type ignition. This can be a source of EMI especially in the servo electronics in RC applications. Besides shielding, it can be advantageous to use wires with a resistive core as the added resistance will spoil the quality of the resonant circuit and reduce the magnitude of this EMI. I notice that these CDI modules have a small resistor in series with the output wire and this is doing approximately the same thing. I'm pretty sure that resistor was placed there because of the RC market. The faster rise times of the CDI ignitions make the EMI situation worse. Their advantage over a coil-type ignition is smaller size and less weight.
The resistance doesn't significantly affect plug firing voltage because even at a few K ohm it is much smaller that the plug gap resistance. - Terry
The wire I used on my 9 cylinder is 20KV wire purchased from S/S Engineering (Robert Sholl). I plan to use it in this engine also. It is just a tinned braided core with a thick silicone insulator. It is black and about 1/8" in diameter. The white stuff in the above test bed photo is quite a bit thicker and it rated at 25KV and is just some stuff I had purchased in a surplus sale many years ago. I was planning to use it on this engine but I'm going to have so many wires and that wire is so thick that I don't think it will look right. And, I didn't have enough of it anyway.
The self inductance of the plug wires and their stray capacitances form a resonant circuit which can be excited by the fast edges of the voltages in the secondary of the output transformer in both a CDI or a coil-type ignition. This can be a source of EMI especially in the servo electronics in RC applications. Besides shielding, it can be advantageous to use wires with a resistive core as the added resistance will spoil the quality of the resonant circuit and reduce the magnitude of this EMI. I notice that these CDI modules have a small resistor in series with the output wire and this is doing approximately the same thing. I'm pretty sure that resistor was placed there because of the RC market. The faster rise times of the CDI ignitions make the EMI situation worse. Their advantage over a coil-type ignition is smaller size and less weight.
The resistance doesn't significantly affect plug firing voltage because even at a few K ohm it is much smaller that the plug gap resistance. - Terry
The design of the heads for my twin-18 has turned out to be every bit as difficult as I figured it would be. The cylinder design wasn't a problem because I started out with a vision of how I want them to look. Instead of the vanilla straight-walls of the H-9 cylinders I wanted a tapered walled design for a more integrated fit with the head. This will complicate the machining, but I think the improved appearance will make the extra work worthwhile. In addition, I decided to double the widths of the fins and grooves and to deepen them by 50%. I think this improves the look and maybe also the cooling since air might flow more easily through the wider grooves.
The head and rocker support design, though, has been an entirely different story. Part of the reason for this is that I'm still not sure how I want these to look. My goal was to come up with something that would look in place on a full size engine and, hopefully, require fewer machining set-ups than the H-9 heads. I was also hoping to duplicate the proven valve train geometry of my H-9 and re-use as many of its already-constructed tooling fixtures as possible.
I very much like the appearance of the Chaos heads, and they ended up greatly influencing my design. I designed and inserted four different heads into my over-all CAD model before ending up with my current result. My favorite head design had a really pleasing elliptical shape that looked great in my model. But, I eventually realized the entire exterior would have to be inefficiently machined on my mill with a lot of complex under-cutting that I'm not sure that I or my CAM is capable of handling. Morphing the elliptical geometry into circular so I can do most of the machining on a lathe required me to modify the stock valve train geometry in order to get the proper pushrod clearance. Another clearance issue that arose with all my large bulbous designs involved the front row intake pipes and the rear cylinder row. The heads in the rear row will have to be notched to clear the intake pipes that feed the front row of cylinders. A close inspection of the Chaos photos shows they ran into the same issue.
The sparkplugs in both rows of cylinders are located in the front halves of the heads. However, the pushrods for the front row are located toward the front of the engine while the pushrods for the rear row are at the rear of the engine. This also complicates any head design for this engine since the rocker arm supports have to be capable of being mounted on the heads while facing in either direction. This, along with the intake pipe clearance, will require two flavors of heads. This really isn't a big deal except for the fact it will double the number of spares that I'll have to make.
The first two illustrations show the original SolidWorks model for my original H-9 cylinder/head assembly. That engine uses the stock Hodgson cylinders and heads but my own rocker arm supports and intake/exhaust flanges. It also uses valve cages instead of the stock pressed-in valve seats.
The rest of drawings show the current models for my twin-18 components. The new valve train geometry is such that the pushrods will be perpendicular to the crankshaft axis instead of being canted 4 deg toward the head. The 16 deg port/aft cant of the stock H-9 head assembly, though, is maintained. My H-9 valve cages will also be carried over to the new head design. I'm slowly becoming fond of the new heads and cylinders, but I've not yet made peace with the rocker arm supports. I had the same misgivings with those I designed for the H-9. I started out leaning toward rocker boxes, but soon felt they would overly complicate the valve lash adjustments on the lower cylinders.
I plan to fine tune the models over the next week and make sure, as best I can, that I can actually fabricate them on the equipment I have and insure they will actually fit into the available space. I will also build and verify separate front and rear models. I've ordered the 12L14 rod for the cylinders as I don't have much left in my scrap collection, although I have plenty of 6061 for the heads. - Terry
The head and rocker support design, though, has been an entirely different story. Part of the reason for this is that I'm still not sure how I want these to look. My goal was to come up with something that would look in place on a full size engine and, hopefully, require fewer machining set-ups than the H-9 heads. I was also hoping to duplicate the proven valve train geometry of my H-9 and re-use as many of its already-constructed tooling fixtures as possible.
I very much like the appearance of the Chaos heads, and they ended up greatly influencing my design. I designed and inserted four different heads into my over-all CAD model before ending up with my current result. My favorite head design had a really pleasing elliptical shape that looked great in my model. But, I eventually realized the entire exterior would have to be inefficiently machined on my mill with a lot of complex under-cutting that I'm not sure that I or my CAM is capable of handling. Morphing the elliptical geometry into circular so I can do most of the machining on a lathe required me to modify the stock valve train geometry in order to get the proper pushrod clearance. Another clearance issue that arose with all my large bulbous designs involved the front row intake pipes and the rear cylinder row. The heads in the rear row will have to be notched to clear the intake pipes that feed the front row of cylinders. A close inspection of the Chaos photos shows they ran into the same issue.
The sparkplugs in both rows of cylinders are located in the front halves of the heads. However, the pushrods for the front row are located toward the front of the engine while the pushrods for the rear row are at the rear of the engine. This also complicates any head design for this engine since the rocker arm supports have to be capable of being mounted on the heads while facing in either direction. This, along with the intake pipe clearance, will require two flavors of heads. This really isn't a big deal except for the fact it will double the number of spares that I'll have to make.
The first two illustrations show the original SolidWorks model for my original H-9 cylinder/head assembly. That engine uses the stock Hodgson cylinders and heads but my own rocker arm supports and intake/exhaust flanges. It also uses valve cages instead of the stock pressed-in valve seats.
The rest of drawings show the current models for my twin-18 components. The new valve train geometry is such that the pushrods will be perpendicular to the crankshaft axis instead of being canted 4 deg toward the head. The 16 deg port/aft cant of the stock H-9 head assembly, though, is maintained. My H-9 valve cages will also be carried over to the new head design. I'm slowly becoming fond of the new heads and cylinders, but I've not yet made peace with the rocker arm supports. I had the same misgivings with those I designed for the H-9. I started out leaning toward rocker boxes, but soon felt they would overly complicate the valve lash adjustments on the lower cylinders.
I plan to fine tune the models over the next week and make sure, as best I can, that I can actually fabricate them on the equipment I have and insure they will actually fit into the available space. I will also build and verify separate front and rear models. I've ordered the 12L14 rod for the cylinders as I don't have much left in my scrap collection, although I have plenty of 6061 for the heads. - Terry
![DreamPlan Home Design and Landscaping Software Free for Windows [PC Download]](https://m.media-amazon.com/images/I/51kvZH2dVLL._SL500_.jpg)
dave-in-england
Junior Member
- Joined
- Aug 30, 2012
- Messages
- 88
- Reaction score
- 20
Nice Solidworks modelling !!!!
- Joined
- Jun 24, 2010
- Messages
- 2,445
- Reaction score
- 968
Verrry nice Terry. I think I understand the machining / setup logic behind the original head that incorporates those kind of cylindrical, finned valve towers. But your rendition looks much nicer IMO. Its amazing how complex 1:1 scale head castings on radials really are when up close. And how thin the fins are. Tough to miniaturize this, but yours look great.
- do the inlet & exhaust ports go perpendicular from that flat segment & axially intersect the centerline of valve/cage? Or at an angle or penetrate the cage on either side for example?
- I see just one screw hole for securing the flange piece to the head? Probably tough to get more & work around the I/O pipes? Are you relying on a gasket or is there any issue with the hot exhaust & cool inlet pipes on the same flange plate? Do the pipes themselves get soldered/welded on those little stubs?
- I guess these heads screw onto the cylinders?
- do the inlet & exhaust ports go perpendicular from that flat segment & axially intersect the centerline of valve/cage? Or at an angle or penetrate the cage on either side for example?
- I see just one screw hole for securing the flange piece to the head? Probably tough to get more & work around the I/O pipes? Are you relying on a gasket or is there any issue with the hot exhaust & cool inlet pipes on the same flange plate? Do the pipes themselves get soldered/welded on those little stubs?
- I guess these heads screw onto the cylinders?
Peter,
The pipes will perpendicularly intersect the axis of the axis of the valve cage. There will be a gasket between the flange and the head, and the flange will be machined from stainless steel. The pipes will be silver soldered into the flange and they will extend about .030" past the gasket and into the head. This will be identical to what I did on my nine cylinder. After silver soldering I had a jig that held the assembly in the mill and then milled the flange surface flat and put a slight taper on the portions of the tubes that protruded past the gasket. This gave a nice snug fit of the assembly into the head that was then held with the single screw. The intake/exhaust assembly is held in place at three points in the final engine and so it really isn't going to loosen up with vibration. The real issues will be getting them seated in place. This is one of the possible problems I've been concerned about since day one of this build. I will need a near perfect jig to hold the assembly in place while the stainless two-into-one pipe stainless pipe assembly is being welded together and then the pipe assembly silver soldered to a pair of flanges. Then it will need to be snaked into position in a very crowded portion of the engine with three ends ending up in precise positions. And then I'll need to do it 9 times. This will likely be the very last fabrication step and so I still have a long time to think on it. And yes, the heads will screw onto the cylinders. I just don't do threads well in SolidWorks. -Terry
The pipes will perpendicularly intersect the axis of the axis of the valve cage. There will be a gasket between the flange and the head, and the flange will be machined from stainless steel. The pipes will be silver soldered into the flange and they will extend about .030" past the gasket and into the head. This will be identical to what I did on my nine cylinder. After silver soldering I had a jig that held the assembly in the mill and then milled the flange surface flat and put a slight taper on the portions of the tubes that protruded past the gasket. This gave a nice snug fit of the assembly into the head that was then held with the single screw. The intake/exhaust assembly is held in place at three points in the final engine and so it really isn't going to loosen up with vibration. The real issues will be getting them seated in place. This is one of the possible problems I've been concerned about since day one of this build. I will need a near perfect jig to hold the assembly in place while the stainless two-into-one pipe stainless pipe assembly is being welded together and then the pipe assembly silver soldered to a pair of flanges. Then it will need to be snaked into position in a very crowded portion of the engine with three ends ending up in precise positions. And then I'll need to do it 9 times. This will likely be the very last fabrication step and so I still have a long time to think on it. And yes, the heads will screw onto the cylinders. I just don't do threads well in SolidWorks. -Terry
dave-in-england
Junior Member
- Joined
- Aug 30, 2012
- Messages
- 88
- Reaction score
- 20
.
Odd question ....but why don't you make the crank-case longer to give more space between the cylinders so that
the exhaust pipes will be easier to fit in ?
.
Odd question ....but why don't you make the crank-case longer to give more space between the cylinders so that
the exhaust pipes will be easier to fit in ?
.
Dave,
I think it's mainly a problem of radial spacing. There just isn't much space between the cylinders in a two row 9 cylinder radial. Essentially there is a cylinder every 20 deg and each cylinder has two pushrods and a pair of intake and exhaust pipes going to it. Lengthening the crankcase would lengthen the pipes but I'd still be left with the same space between the cylinders to route them. Or maybe you're thinking the extra length would help with angling them into place? Anyway, the Chaos guys did it and so I should be able also. The thing that worries me though is that I noticed they removed the skirts from their cylinders and that was a BIG deal thing to have done. They may have found during assembly they needed to do it in order to get the pipes assembled to the cylinders. The stroke of this engine is such that the pistons go down deep into the cylinder skirt and so there had to have been a really good reason for them to have removed theirs. -Terry
I think it's mainly a problem of radial spacing. There just isn't much space between the cylinders in a two row 9 cylinder radial. Essentially there is a cylinder every 20 deg and each cylinder has two pushrods and a pair of intake and exhaust pipes going to it. Lengthening the crankcase would lengthen the pipes but I'd still be left with the same space between the cylinders to route them. Or maybe you're thinking the extra length would help with angling them into place? Anyway, the Chaos guys did it and so I should be able also. The thing that worries me though is that I noticed they removed the skirts from their cylinders and that was a BIG deal thing to have done. They may have found during assembly they needed to do it in order to get the pipes assembled to the cylinders. The stroke of this engine is such that the pistons go down deep into the cylinder skirt and so there had to have been a really good reason for them to have removed theirs. -Terry
I've been anxious to get back into the shop, and so I decided to take a break from the head design and make a 'small' trial run of cylinders. I'm pretty satisfied with the current cylinder design, and after all the time I've spent thinking about them, it's probably safe to find out if I can build them. I decided to make an initial lot of six parts since I was able to round up enough scrap in my shop for that many, and I need at least four cylinders to verify the critical clearances around the oil sump on the bottom of my crankcase. My approach to making a large batch of identical parts is to make a 'prototype' run and to develop a step-by-step process for making them as efficiently as I possibly and then carefully document it. Due to the large number of parts I'll eventually be making I want to have the machines do as much of the finishing as possible in order to minimize manual clean-up. I'll use this process later to make the remainder of the parts and also any replacements I may need in the future.
I started by sawing six identical lengths of 1-3/4" diameter 12L14 rod. These were chucked in my Enco 12x36 lathe where all the heavy cutting was done to prepare them as blanks for my cylinders. The parts were faced to .050" over the cylinder's finished length, and then the centers were drilled out in two steps to 15/16". The OD's were finally turned to .025" over the cylinder's maximum finished OD. The manual preparation time for each of these blanks turned out to be about 45 minutes.
The blanks were then chucked into my 9x20 CNC converted lathe where the bores were completed. I wrote two simple programs to bore the IDs. The first one roughed the ID to 0.986", and the second one finished the bore out at 0.998". The tool offset was corrected to match the actual measured ID of each roughed part before the final finishing program was run. This resulted in all the finished bores being within 0.0002" of each other. I used high rake Korloy inserts designed for aluminum to get a mirror finish even though the cylinders will be honed later. My own personal experience with honing is that trying to remove more than a half thousandth is a long, messy, and, in general, miserable experience with sometimes inconsistent results. Therefore, I try to get the cylinders as close as possible to their finished state before honing. If I can hold the .0002" over my complete run of cylinders, I may lap them instead of honing. My lathe left a 0.0005" taper over the length of the blank, and so I was careful to make sure the small end wound up at the top of the combustion chamber. The important part of the taper which is that seen by the combustion rings is less than half of the total; and nearly all of that will, hopefully, be honed or lapped away later.
I was only able to get four finished bores per insert edge which is pretty extravagant, but the worn inserts still have lots of aluminum roughing time left on them for later projects.
For the remainder of the lathe work the blanks must be supported by their IDs using an expandable mandrel that I turned for this purpose during my H-9 build.
The aluminum heads will eventually be screwed onto the cylinders. The first operation on the mandrel was to prepare the top of the cylinder and to turn the threads, thread relief, and a mating surface for a soft aluminum head gasket.
The tapered body with its fully radiused fins and filleted cooling grooves is a complex feature of my cylinder design, and it has to be carefully machined to avoid a lot of tedious and inconsistent manual clean-up. My plan involved using a full radius grooving tool and a CAM program to fully machine the complex surfaces with negligible scallops. Unfortunately, the grooving operation available in my CAM software didn't seem to like the tapered body of my cylinders. Grooving and parting on my lathe, especially in steel, is always a sobering experience because of the machine's lack of rigidity. Every new project is a new experience requiring lots of experiments and sacrificial inserts to find just the right combination of feeds and speeds that not only just machine the groove but also give a nice surface finish. I generally have to take small bites with aggressive feeds and plenty of chip breaks. Because of the shape of my cylinder, the CAM software wanted to generate tool paths that spent most of the machining time cutting air and using up the chip breaks before plunging the tool into the workpiece the full depth of the groove. Adding insult to injury, the simulator estimated the machining time to well over an hour. I spent some 10 hours convincing it to behave rationally and eventually ended up with a 10 minute operation that did a beautiful job. As is sometimes the case, I had to lie to the software about the shapes of the workpiece and the part and then fiddle, in trial and error fashion, with several of the operation's parameters. If the real truth were known, though, I wouldn't be surprised if most of my problems were caused by my incomplete understanding of the software.
Another grooving program was written to clear out the material above the mounting flange where no other turning tool would fit. This grooving operation was relatively straightforward after my experience with the first one. Two final programs had to be created. One removed OD material from below the mounting flange and the other cut a short ID taper in the bottom of the bore to provide clearance for the connecting rod. This taper will also help ease the insertion of the ringed pistons during assembly.
The final results are shown in the photos. The surface finish is as the parts came off the lathe and will require no polishing before being blued. The only manual clean-up required is the removal of a 'whisker' where the last thread at the top of the cylinder terminates in the thread relief groove. The cylinder clearances around the oil sump were checked, and I also verified that I could actually insert the cylinders into the crankcase bores around the sump with the studs temporarily installed. One of the photos also shows a side-by-side comparison of my twin-18 cylinder with a stock Hodgson cylinder from my H-9 build. All six blanks made it through to good finished parts even though I had expected to ruin a couple along the way.
The machining of these cylinders is an example of a fairly complex project done on hobby-level CNC equipment but which could also have been done manually. I think it's interesting to recall what was required to create the first articles.
Approximately 45 minutes of manual prep was required to create each blank that my CNC lathe could handle. Ten individual g-code programs were created to complete the machining of the cylinder. Seven of these were rather trivial, and were quickly done using the conversational wizards within the Mach-3 control program. The other three required CAM software for their creation. I spent a total of 12 hours creating these programs, but the learning experience with the particularly troublesome grooving program consumed 10 of those hours. The total CNC machining time for all ten programs turned out to be about one hour per cylinder. This included the part set-ups, machine referencing, and verification of the parts. Therefore, amortized over an eventual 24 cylinders, my CAM time per part will be about 1/2 hour. The total fabrication time per cylinder added up to be 1-3/4 hours for a grand total 2-1/4 hours of my actual time per cylinder.
Hodgson estimates 8 hours of machining time by a typical builder for one of his H-9 cylinders, and they are comparable in complexity to my design. That's probably close to the time it would take me to make one assuming I could create a form tool for the fins that my lathe can handle. Unfortunately, though, after the third or fourth cylinder I'd probably set the project up on a high shelf.- Terry
I started by sawing six identical lengths of 1-3/4" diameter 12L14 rod. These were chucked in my Enco 12x36 lathe where all the heavy cutting was done to prepare them as blanks for my cylinders. The parts were faced to .050" over the cylinder's finished length, and then the centers were drilled out in two steps to 15/16". The OD's were finally turned to .025" over the cylinder's maximum finished OD. The manual preparation time for each of these blanks turned out to be about 45 minutes.
The blanks were then chucked into my 9x20 CNC converted lathe where the bores were completed. I wrote two simple programs to bore the IDs. The first one roughed the ID to 0.986", and the second one finished the bore out at 0.998". The tool offset was corrected to match the actual measured ID of each roughed part before the final finishing program was run. This resulted in all the finished bores being within 0.0002" of each other. I used high rake Korloy inserts designed for aluminum to get a mirror finish even though the cylinders will be honed later. My own personal experience with honing is that trying to remove more than a half thousandth is a long, messy, and, in general, miserable experience with sometimes inconsistent results. Therefore, I try to get the cylinders as close as possible to their finished state before honing. If I can hold the .0002" over my complete run of cylinders, I may lap them instead of honing. My lathe left a 0.0005" taper over the length of the blank, and so I was careful to make sure the small end wound up at the top of the combustion chamber. The important part of the taper which is that seen by the combustion rings is less than half of the total; and nearly all of that will, hopefully, be honed or lapped away later.
I was only able to get four finished bores per insert edge which is pretty extravagant, but the worn inserts still have lots of aluminum roughing time left on them for later projects.
For the remainder of the lathe work the blanks must be supported by their IDs using an expandable mandrel that I turned for this purpose during my H-9 build.
The aluminum heads will eventually be screwed onto the cylinders. The first operation on the mandrel was to prepare the top of the cylinder and to turn the threads, thread relief, and a mating surface for a soft aluminum head gasket.
The tapered body with its fully radiused fins and filleted cooling grooves is a complex feature of my cylinder design, and it has to be carefully machined to avoid a lot of tedious and inconsistent manual clean-up. My plan involved using a full radius grooving tool and a CAM program to fully machine the complex surfaces with negligible scallops. Unfortunately, the grooving operation available in my CAM software didn't seem to like the tapered body of my cylinders. Grooving and parting on my lathe, especially in steel, is always a sobering experience because of the machine's lack of rigidity. Every new project is a new experience requiring lots of experiments and sacrificial inserts to find just the right combination of feeds and speeds that not only just machine the groove but also give a nice surface finish. I generally have to take small bites with aggressive feeds and plenty of chip breaks. Because of the shape of my cylinder, the CAM software wanted to generate tool paths that spent most of the machining time cutting air and using up the chip breaks before plunging the tool into the workpiece the full depth of the groove. Adding insult to injury, the simulator estimated the machining time to well over an hour. I spent some 10 hours convincing it to behave rationally and eventually ended up with a 10 minute operation that did a beautiful job. As is sometimes the case, I had to lie to the software about the shapes of the workpiece and the part and then fiddle, in trial and error fashion, with several of the operation's parameters. If the real truth were known, though, I wouldn't be surprised if most of my problems were caused by my incomplete understanding of the software.
Another grooving program was written to clear out the material above the mounting flange where no other turning tool would fit. This grooving operation was relatively straightforward after my experience with the first one. Two final programs had to be created. One removed OD material from below the mounting flange and the other cut a short ID taper in the bottom of the bore to provide clearance for the connecting rod. This taper will also help ease the insertion of the ringed pistons during assembly.
The final results are shown in the photos. The surface finish is as the parts came off the lathe and will require no polishing before being blued. The only manual clean-up required is the removal of a 'whisker' where the last thread at the top of the cylinder terminates in the thread relief groove. The cylinder clearances around the oil sump were checked, and I also verified that I could actually insert the cylinders into the crankcase bores around the sump with the studs temporarily installed. One of the photos also shows a side-by-side comparison of my twin-18 cylinder with a stock Hodgson cylinder from my H-9 build. All six blanks made it through to good finished parts even though I had expected to ruin a couple along the way.
The machining of these cylinders is an example of a fairly complex project done on hobby-level CNC equipment but which could also have been done manually. I think it's interesting to recall what was required to create the first articles.
Approximately 45 minutes of manual prep was required to create each blank that my CNC lathe could handle. Ten individual g-code programs were created to complete the machining of the cylinder. Seven of these were rather trivial, and were quickly done using the conversational wizards within the Mach-3 control program. The other three required CAM software for their creation. I spent a total of 12 hours creating these programs, but the learning experience with the particularly troublesome grooving program consumed 10 of those hours. The total CNC machining time for all ten programs turned out to be about one hour per cylinder. This included the part set-ups, machine referencing, and verification of the parts. Therefore, amortized over an eventual 24 cylinders, my CAM time per part will be about 1/2 hour. The total fabrication time per cylinder added up to be 1-3/4 hours for a grand total 2-1/4 hours of my actual time per cylinder.
Hodgson estimates 8 hours of machining time by a typical builder for one of his H-9 cylinders, and they are comparable in complexity to my design. That's probably close to the time it would take me to make one assuming I could create a form tool for the fins that my lathe can handle. Unfortunately, though, after the third or fourth cylinder I'd probably set the project up on a high shelf.
- Terry
- Joined
- Jun 24, 2010
- Messages
- 2,445
- Reaction score
- 968
Terry, I think you mentioned you jobbed out the blueing/blackening operation to gunsmith or similar service on your last engine. Is that step done before final bore finishing or is it ok to run the rings against that surface? Are the threads or any other areas somehow 'masked off' or the whole jug gets the dipper-oo? I've heard different variations of surface prep treatments from mild acid to even light blasting. But looks like your fins & profiles are 'final shape' so assume they cant be doing anything too harsh? Do you have to oil or treat the blackening some how initially to seal it in?
Terry
It is coming along nicely ! Very nice work. What cam package are you using ? I have to lie to mine all the time. ( SprutCam )
Peter
Bluing is just a fancy form of rust. Commonly know as black oxide. Gunsmiths use a refined form of salts for a deeper more lustrous color. But it is still just a "Stain" on the metal. For surface prep all it really needs is to be clean and oil free. Acid etching and blasting or polishing are all for the final appearance. If you want a bright finish you polish it if you want a matte finish you either etch it or blast it.
There would really be no need to mask the bore or hone it later. The rings will clean it up with a few strokes of the piston.
I don't mask the inside of barrels when bluing.
Scott
It is coming along nicely ! Very nice work. What cam package are you using ? I have to lie to mine all the time. ( SprutCam )
Peter
Bluing is just a fancy form of rust. Commonly know as black oxide. Gunsmiths use a refined form of salts for a deeper more lustrous color. But it is still just a "Stain" on the metal. For surface prep all it really needs is to be clean and oil free. Acid etching and blasting or polishing are all for the final appearance. If you want a bright finish you polish it if you want a matte finish you either etch it or blast it.
There would really be no need to mask the bore or hone it later. The rings will clean it up with a few strokes of the piston.
I don't mask the inside of barrels when bluing.
Scott
Walltoddj
Well-Known Member
- Joined
- Aug 5, 2013
- Messages
- 145
- Reaction score
- 57
I like what you did with the cyl it looks more realistic if you look at the original engines, we had one at GM Powertrain in Buffalo, NY they built them there. The test rooms were cool two foot thick concrete walls and they had holes in them. Are you going to have some prints of your changes? How can I get some or the changes dimensions? I feel it would be worth changing it looks much better. Did you see chaos changes he did the heads and rocker boxes too.
Todd
Todd
Peter,
The bluing will be done after the honing plus everything Scott said.
Scott,
I'm using Sprutcam also. I think I will try to make a movie of the fin grooving on my next run of cylinders. It's a lot of fun to watch especially if you've ever cut similar fins manually.
Todd,
I haven't created any dimensioned drawings of anything I've done as I go directly from my SolidWorks model into my CAM software and no drawings are involved. If there is enough interest, though, I might create a drawing package of the cylinders and heads when I'm done. Anyone building a Hodgson nine cylinder could drop these in place of the stock parts.
Terry
The bluing will be done after the honing plus everything Scott said.
Scott,
I'm using Sprutcam also. I think I will try to make a movie of the fin grooving on my next run of cylinders. It's a lot of fun to watch especially if you've ever cut similar fins manually.
Todd,
I haven't created any dimensioned drawings of anything I've done as I go directly from my SolidWorks model into my CAM software and no drawings are involved. If there is enough interest, though, I might create a drawing package of the cylinders and heads when I'm done. Anyone building a Hodgson nine cylinder could drop these in place of the stock parts.
Terry
SolidWorks uses something called a hierarchical structure to keep track of a part design. This means that a user designs his part by sequentially adding features such as cuts, extrusions, holes, fillets, etc.; and SolidWorks stores the design in a sequential hierarchy where each added feature depends upon the current state of the part created by all the previously added features. To the user this also means that even small changes to a design may not be possible without major re-work (read sometimes as must start over) to the part. So, a user is warned to design his part with potential design changes in mind so they can be easily made even after the design is finished. This is probably one of the hardest aspects of a CAM tool to master, and I'm still learning even after several years of hobbyist experience.
I designed my heads with this in mind since I was constantly making changes to the design while searching for something I liked. When it came time to start thinking about how I was going to build the part, though, I found that I had ended up with a poor hierarchy. For example, I left the fin profiling until the end of the design after the valve towers, sparkplug port, and exhaust manifold were completed. I did this to avoid continually un-doing and re-doing the fillets every time a fin was affected by one of my changes. However, when I actually machine the head, I plan to turn the finished profile including the fins during one of the very early operations.
As soon as I tried to drag the filleting features from the bottom of the design hierarchy to the top my entire CAD model fell apart since several of the features that I was trying to fillet didn't yet exist at that point in the model. It's important that I have access to legitimate intermediate states in the hierarchy, because they represent the current workpiece that my CAM will need to generate the tool paths for the various machining operations. Also, I want an accurate step-by-step rendering of my part so I can decide whether the effort to machine certain minor features such as the fin radii in and around the spark plug recess is really justified. Therefore I'm starting over and re-creating a SolidWorks model driven by my machining sequence. This time, though, I have a visual model to work toward. A pro would probably have been able to do both the first time around. I also need to come up with lathe and mill work-holding fixtures with proper geometry to not only hold the workpieces while they are being machined but to also provide machine reference points that can be easily indicated.
That aside, the 12L14 material that I ordered online arrived, and so I actually spent most of the week making the rest of my cylinders. I don't buy much metal since I prefer to use, whenever possible, the pennies/pound scrap that I've accumulated over the years. The shipping costs for this piece of steel turned out to be comparable to the cost of the material. Most of my scrap originated as drops from machine shops and is usually quality stuff although the exact alloy can sometimes be tricky to identify.
Anyway, while drilling the 15/16" hole through this new rod to prepare more cylinder blanks I noticed things felt very 'different" from when I made the first six. It wasn't more or less difficult, but it was just somehow different. All went well until some 30 hours into the run when I got to the cylinder grooving operation. The grooving code that had run so beautifully on the first six pieces would not run on this new material without really serious chatter. I spent several hours with test pieces of this new rod trying to the find a new feed/speed sweet spot, but there just wasn't any. I eventually had to abandon the 5C expandable mandrel I had been using and turned a new one for my 3-jaw chuck. This only slightly increased the rigidity of my set-up, but it was enough along with an additional reduction in peck depth to get me going again. Something was very different about the stiffness of this lot of 12L14. My grooving tool seemed to be exciting a new resonance with this new material in my original work holding fixture and causing a violent chatter. Changing the fixture a bit evidently damped or moved the resonance and reducing the peck depth reduced the amount of time the cutter had to excite it. I think the 12L14 that I used for the first six parts came from at least four different lots, and so I think the real culprit was the expensive stuff I had just purchased. I tried to make a video of the fin profiling process because once it was running again it was neat to watch, but getting a viewable close-up image from my flip video camera was not possible.
I made 18 more cylinders for a total of 24. I'm not sure why I made so many. Every blank made it through to a useable finished part. The cylinder bores ranged from 0.9972" to 0.9979" on 23 of the finished parts with the 24th coming in at 0.9997" due to a start-up tool calibration error. I'll finish them later with a combination of honing and lapping as I did on my H-9 before having them hot-blued. At this point it looks like the finish of the parts as they came off the lathe will be adequate preparation for the bluing process. -Terry
I designed my heads with this in mind since I was constantly making changes to the design while searching for something I liked. When it came time to start thinking about how I was going to build the part, though, I found that I had ended up with a poor hierarchy. For example, I left the fin profiling until the end of the design after the valve towers, sparkplug port, and exhaust manifold were completed. I did this to avoid continually un-doing and re-doing the fillets every time a fin was affected by one of my changes. However, when I actually machine the head, I plan to turn the finished profile including the fins during one of the very early operations.
As soon as I tried to drag the filleting features from the bottom of the design hierarchy to the top my entire CAD model fell apart since several of the features that I was trying to fillet didn't yet exist at that point in the model. It's important that I have access to legitimate intermediate states in the hierarchy, because they represent the current workpiece that my CAM will need to generate the tool paths for the various machining operations. Also, I want an accurate step-by-step rendering of my part so I can decide whether the effort to machine certain minor features such as the fin radii in and around the spark plug recess is really justified. Therefore I'm starting over and re-creating a SolidWorks model driven by my machining sequence. This time, though, I have a visual model to work toward. A pro would probably have been able to do both the first time around. I also need to come up with lathe and mill work-holding fixtures with proper geometry to not only hold the workpieces while they are being machined but to also provide machine reference points that can be easily indicated.
That aside, the 12L14 material that I ordered online arrived, and so I actually spent most of the week making the rest of my cylinders. I don't buy much metal since I prefer to use, whenever possible, the pennies/pound scrap that I've accumulated over the years. The shipping costs for this piece of steel turned out to be comparable to the cost of the material. Most of my scrap originated as drops from machine shops and is usually quality stuff although the exact alloy can sometimes be tricky to identify.
Anyway, while drilling the 15/16" hole through this new rod to prepare more cylinder blanks I noticed things felt very 'different" from when I made the first six. It wasn't more or less difficult, but it was just somehow different. All went well until some 30 hours into the run when I got to the cylinder grooving operation. The grooving code that had run so beautifully on the first six pieces would not run on this new material without really serious chatter. I spent several hours with test pieces of this new rod trying to the find a new feed/speed sweet spot, but there just wasn't any. I eventually had to abandon the 5C expandable mandrel I had been using and turned a new one for my 3-jaw chuck. This only slightly increased the rigidity of my set-up, but it was enough along with an additional reduction in peck depth to get me going again. Something was very different about the stiffness of this lot of 12L14. My grooving tool seemed to be exciting a new resonance with this new material in my original work holding fixture and causing a violent chatter. Changing the fixture a bit evidently damped or moved the resonance and reducing the peck depth reduced the amount of time the cutter had to excite it. I think the 12L14 that I used for the first six parts came from at least four different lots, and so I think the real culprit was the expensive stuff I had just purchased. I tried to make a video of the fin profiling process because once it was running again it was neat to watch, but getting a viewable close-up image from my flip video camera was not possible.
I made 18 more cylinders for a total of 24. I'm not sure why I made so many. Every blank made it through to a useable finished part. The cylinder bores ranged from 0.9972" to 0.9979" on 23 of the finished parts with the 24th coming in at 0.9997" due to a start-up tool calibration error. I'll finish them later with a combination of honing and lapping as I did on my H-9 before having them hot-blued. At this point it looks like the finish of the parts as they came off the lathe will be adequate preparation for the bluing process. -Terry
Walltoddj
Well-Known Member
- Joined
- Aug 5, 2013
- Messages
- 145
- Reaction score
- 57
Nice job how about doing another 27 so I can do my engines, no I'll be doing mine by hand don't have CNC. They look great as I said I like what you've done to improve the design it looks a whole lot better to me!!
Todd
Todd
barnesrickw
Well-Known Member
- Joined
- Sep 8, 2012
- Messages
- 645
- Reaction score
- 121
Nice job on the cylinders, and great photo log of the build.
Sent from my iPad using Model Engines
Sent from my iPad using Model Engines
