Quarter Scale Merlin V-12

[mayhugh1]

Kyle,
I agree with you about the spaces. I actually do use them but somewhere in my process they seem to get removed. I normally compose the text on my ipad and then email it to my Windows machine from where it's uploaded. Somewhere along the way the spaces are removed. I just tried to edit my last post and add the spaces back in after the upload and that worked. I'll try to remember to do the same in the future. Glad it's been useful to you. - Terry

 

[dsage]

Terry:
It's not your fault. There is something fundamentally wrong with entering text on this (and most) forums. For instance if you try to use a Tab to indent text - like at the beginning of a paragraph - your cursor disappears outside the text box ending the editing session. There seems to be no way to start a new paragraph with an indent. Nor can you simulate a tab by entering a few spaces. More than one space in a row gets reduced to a single space. For instance:
This is a new paragraph I simulated with a carriage return and a large number of spaces at the beginning, yet you can see the spaces have disappeared before the word "THIS" as soon as I saved the message. Also I tried to enter a bunch of spaces in the middle of the line here between the dashes - - and they disappear leaving only one space when the message is posted. Even if I go back and edit the message the spaces are missing and putting them back does not help.


This line started with three carriage returns so that seems to work. So you may be able to space your paragraphs by using a couple of returns. That would help greatly.
It's all very frustrating and I wish someone would fix it because (like me) I know you try to make your posts clear and understandable and text formatting is very important to that end.

Thanks for all your work. It's amazing to follow along.

Sage

 

[kylenlord]

Great Terry, thank you. I do most of my engine browsing on the phone though it looked the same on the laptop.
Keep it up I'm looking forward to more updates.

Kyle

 

[dsage]

Terry et. al.

I subscribe to the daily digest for this forum i.e I receive a single email with all of the posts for the day, and I just noticed that the spaces that I entered in my post while I was directly on the forum DO show up in the email version. Even though they are NOT on forum post. So it's even more screwed up than I thought.

Sage

 

[mayhugh1]

Since I had machined five sets of mounts and covers and still had plenty Optek sensors left over, I thought I'd make up three more cabled assemblies for spares while the 'art' portion of their construction was still fresh in my mind. To my surprise and disappointment the next three assemblies all had issues with the sensors turning OFF. When I rechecked the first two, I found the one containing the sensor around which the mounts had been optimized was still working reliably in both distributors. A closer look at the second assembly, though, showed it might have been only marginally turning OFF in one of the two distributors.

This is really my first experience with this particular Hall device. The ones I've used in my other engines have long been discontinued, and I'm saving what remaining stock I have of them for replacements. I purchased the Optek parts a couple years ago after reading another builder's comments about his experience with their ruggedness. He claimed to have seen accidental discharges to the sensor's ground lead with no ill effect to the part. This wasn't something I would have necessarily expected the part to survive, and so I ordered a dozen pieces for future projects. And now, here we are.

The Optek data sheet actually warns about potential turn-OFF issues with the OH090U. The 'turn-on' flux densities in Gauss are specified as 0(min), 90(typ), and 180(max). The 'turn-off' densities are listed as -100(min), 65(typ), and 100(max). Turn-off levels which are roughly half the turn-on levels as listed for the typical and maximum specs are reasonable behaviors to design around. The minimum specs aren't at all reasonable, and I originally dismissed them as unlikely anomalies that were probably so many standard deviations away from typical behavior that they would never be seen. After all, the OH090U is sold as a non-latching unipolar device, but the minimum numbers seem to imply that the field may actually have to be reversed in order to turn some sensors OFF.

Still hoping the sensor's internal hysteresis would be proportional to the level of flux that actually turns it ON, I machined two new sets of trigger disks with smaller apertures to reduce the strength of the field applied to the device. I reduced the aperture diameters from the original .187" down to .156" and then again to .125". The disappointing results were basically that the sensors now had difficulty in turning ON as well as turning OFF. I also machined a thicker disk to make sure that saturation wasn't the issue.

Finally, I made up a fixture to carefully compare the turn-on and turn-off distances between a test magnet and the sensors in each of my five mounted assemblies but without a disk between them. The turn-on distances were very similar, but the turn-off distances varied an unacceptable factor of 2.5 among the sensors in my tiny sample size. At this point I had to conclude that either a reputable distributor had sold me floor sweepings, or this particular sensor was not designed to provide the consistency I was looking for.

Many of the specs for the Infineon TLE4905 are similar to those of the Optek device, and its data sheet lists minimums and maximums but no typicals. The turn-on flux densities converted to Gauss are listed as 70(min), and 180(max), and the turn-off densities are 50(min), and 160(max). The lack of a typical spec implies the parts probably come out of manufacturing uniformly distributed between the two extremes, and this is likely the case for the Optek parts as well. These ON/OFF ratios are very attractive for a trigger disk, and I eventually ordered the 4905's to replace the OH090U's. One of the reasons for selecting this particular part from a bewildering number of available possibilities was that its package width is similar to that of the Optek part, and so I could reuse my mounts. I had to shave off another .010", however, in order to optimize them for the new parts. I didn't modify the one Optek assembly that continues to work well with both distributors, but the sensors in the other four assemblies were all replaced. All four assemblies finally worked as expected. My distance measurements for the new assemblies showed the turn-on and turn-off distances to all be tightly clustered around similar values, and so hopefully the sensor stuff is finally behind me.

With some fifty fairly unforgiving machined parts plus the shafts, bearings, and gears, each of these distributors has ended up requiring as much effort as an entire single cylinder engine. Instead of a pair of running engines, though, so far I only have a pair of blinking LED's to show for more than a month's work. And since I have to manually spin their shafts to get the LED's to flash, my wife isn't very impressed. I think the end is in sight, but I still have three sets of housing covers to design and machine before I can finally check the distributors' operation under high voltage.

Both distributors are located at the rear of the engine behind the exhausts. This means the electrically sensitive portions of the distributors should be shielded from the oil, water, and gasoline they'll likely see during cold start-ups. The full-size Merlin's magnetos had the advantage of not being vulnerable to start-up grunge since the exhaust tips were located outside the plane's engine compartment. The Quarter Scale's distributors have an advantage of their own, though, they're not vulnerable to enemy gunfire.

The easiest covers to make were the ones shielding the timing adjusters. It seems like a shame to cover up all the painstakingly machined hardware associated with the trigger disks and sensors, but hiding the neat stuff has become common practice in this build. Other than some intricate profiling to clear the components of the timing adjuster, the machining of these black Delrin covers was fairly straightforward. Hex nuts secure them to a pair of 1-72 studs threaded into the adjusters.

The covers for the ends of the rotor housings were more complex because they contain and protect the critical high voltage coil connections to the spinning rotors. To make things more interesting, I machined the exteriors of the covers from black Delrin for appearance sake, but the interiors surrounding the contacts were machined from white Delrin for its better dielectric properties. Before machining the covers I made up their workpieces by pressing a pair of tight-fitting white Delrin inserts into the black Delrin blanks. The shapes of the covers' peripheries complicated their work-holding in the lathe, and so I machined an expandable mandrel to grip the shallow interior of its end. The Delrin's slipperiness greatly limited the depth of cut. For the rotor button, I decided on a spring-loaded phosphor bronze contact instead of the carbon button I had been considering in order to avoid carbon dust inside the relatively small volume of the cover.

The last pair of covers will be the most complex of the three. They will be designed to replace the contact block cover castings in order to provide a more elegant exit for the plug wires. I plan to design them to be similar to the magneto covers used on one of the versions of the full-scale Merlin. - Terry

 

[dsage]

Hi Terry:

I think you made the right move to steer away from the Optek sensors. Who knows how their already variable specs might have changed with time and temperature etc getting you in trouble. I too have many of these sensors but they are triggered by spinning magnets. It might be interesting for me to have a look on a scope to see how the dwell time is varying with each magnet as it passes.
Thanks for the heads up. I'll order some of the Infineon sensors for future consideration. Keep us posted on their operation.

Thanks

Sage

 

[apointofview]

Terry,
I have been amazed at your work and want to thank you for taking the time to document all of this for everyone to enjoy !!

I am late with this suggestion but maybe future designs could use this or maybe not... Have you ever looked into an optical sensor. I ran one in a nitro r/c helicopter that sees crankshaft speeds in the neighborhood of 17000 rpm. This sensor - http://www.digikey.com/catalog/en/p...opb703wz-opb704wz-opb705wz-and-opb70awz/17829 - was triggered by reflections from a very small piece of aluminum tape on a black clutch bell and it gave rpm feedback to a governor that maintained rpm as the load of the rotor changed. It supplies its own light and was very reliable.

Pete

 

[mayhugh1]

Pete,
Back in the early 70's I designed a capacitive discharge ignition for my brand new 6-cylinder Ford Maverick. To trigger it, I made up an optical sensor assembly using a six hole metal disk that slid down over the distributor's cam and interrupted a light beam between an led and a photosensor that were attached to the distributor's advance plate. I temperature tested the electronics to 125C in my employer's environmental chamber during the evenings, because at the time we lived in the middle of the Mohave Desert in California and trek'd across it nearly every weekend. I wasn't trying to solve any particular problem except maybe to avoid adjusting points, but in those days if it wasn't broke I fixed it anyway. Although I kept the original parts in the glovebox as a back-up, the system worked flawlessly for over three years until we left the desert to move east. During our cross-country drive I spent about an hour one evening somewhere along Route 66 in New Mexico, with my wife holding a flashlight in one hand and and a baby in her other arm, putting the points back in so we could be on our way. That experience helped me to appreciate the auto industry's move to the much more reliable variable reluctor. - Terry

 

[mayhugh1]

The distributors have top covers that protect the high voltage plug wire connections to the contact blocks. The covers I received were most likely early castings designed for the original Quarter Scale distributors that used a single row contact block. In addition to the plug wires, the coil connections to these first generation distributors, as well as the battery connections to the full-scale magnetos, were also under these covers. A pair of sketchy holes had to be drilled in the sides of the Quarter Scale's covers to provide exits for the wires. I found an online photo of the completed covers on Gunnar's engine that shows what had to be done. Similar covers used on some of the full-size Merlins included integrally cast conduits that were angled inward for the wires. Space at the fronts of the full-scale magnetos was limited, and several versions of the housings appeared over time to accommodate engine design changes as well as differences among the various Merlin models.

When the Quarter Scale's distributors were redesigned for a double-row contact block, the coil connections were moved under their own covers at the ends of the rotor housings. The top covers were redesigned with more elegant front elbow exits for the plug wires. A photo of John Ramm's engine shows an example of these later castings. They aren't much different from the magneto covers found on some of the later versions of the Merlin.

I liked the looks and functionality of the later covers much more than the castings I received, and so I designed a pair of replacements. My particular castings were usable although the 20kV wires I wanted to use overfilled them. Since the covers would end up prominently mounted at the top of the engine I really wanted a tidier solution for the wire exits. My goal was to design a set of covers that were similar to those on the late model Merlins and with a little more interior volume for the 1/8" diameter plug wires and retaining collets I was using. The space available for a front conduit is limited even more so in the Quarter Scale than it was in the full-size engines because of the addition of the timing chain cover.

I spent a few days with SolidWorks designing and printing out actual size cross-sectional patterns as a cover design slowly evolved. Since I didn't have CAD models for any of the castings, I tested my paper models against a temporary assembly of the engine's potentially interfering parts so I could estimate the clearances. The timing chain cover was the most difficult obstacle to deal with, and the magneto cover was essentially designed around it.

I initially considered splitting the cover design lengthwise and machining it in black ABS as two halves to be glued together. This would have made it possible to use a cylindrical conduit to completely enclose the exiting wires. Although not a 'must have', I was hoping to be able to install and remove the covers from the installed distributors with the timing chain cover installed, and such a conduit would have made this very difficult. The design eventually evolved into one with a three-sided exit trough that I hoped would simplify maneuvering the covers into and out of place. The open rear allowed the covers to be machined from single blocks of aluminum without compromising the shielding of the contact block from the start-up grunge.

I was able to machine each cover in just two set-ups despite all the filleting added to make them look like castings. The total machining time on my Tormach was about three hours per cover and, when finished, their shiny surfaces were bead blasted to try to match the surface texture of the castings. The color match isn't very good, but they may end up later painted black. The resulting minimum clearance to the timing chain cover tubes came out to only .020" which was half of what I had estimated from my paper pattern checks. Another close feature was the final internal volume. The six collet'd 1/8" diameter 20kV wires fit under the covers, but they have to be carefully laid out. In the end it wasn't possible to remove or install the covers on the already installed distributors after the timing chain cover was installed. The 20kV wires drove the minimum dimensions of the trough, and the final result was just too close to the timing chain cover. This only means the wires and covers will have to be installed on the distributors before the distributor is installed on the engine. Both covers were checked on both distributors, and so hopefully there won't be any unpleasant surprises during final assembly. - Terry

 

[apointofview]

Thanks Terry that explains why your not using an optical setup really well, once something bites you its hard to consider using it again. Mad Dog 20/20 comes to mind for me
Pete

 

[nel2lar]

Terry
I just run on to your build and what a job, very nice. I am sure you are satisfied as you advance closer to completion.
Very very sweet
Nelson

 

[mayhugh1]

With the distributors completed (I hope), I could finally check their operation under high voltage. I inserted six lengths of 20 kV plug wire into the contact block on the starboard distributor using the retaining collets made earlier. A shop-made insertion tool helped with the collet installations in the limited area at the top of the assembled distributor. My custom machined covers have an internal shoulder that wiil bear down slightly against the tops of the collets for back-up retention against engine vibration, and so it's now important that the collets be fully inserted.

Temporary terminals were soldered on the other ends of the wires for connections to the CM-6 spark plugs that I chose for testing instead of the more expensive Mini Viper plugs that will actually be used in the engine. I expected the CM-6's wider gaps would make a better worst-case test, but mainly I didn't want to risk damaging the Vipers or deforming their washers in my test fixture. I practiced synchronizing the rotor and trigger disk to a couple arbitrary crankshaft positions. A pin in the rotor index hole proved invaluable for locating and securing the position of the rotor so the trigger wheel shaft screw could be tightened.

Although it's much too early to attempt valve adjustments, I thought I would make up a degree wheel to verify the limits of the timing adjuster and to check the consistency of the edges of the sensor signals. Since I had some earlier difficulties with the sensors, and any aperture machining errors could create timing inconsistencies, I wanted to make sure the static timing jitter was acceptable before moving on with the rest of the build. I have the option of adding custom circuitry between the Hall sensor and the CDI module, and so I'm free to use the sensor's most consistent signal edge to fire the plugs.

When I started work on the degree wheel I found I had been thrown a curve that I hadn't seen coming. The only available shaft for mounting a degree wheel during final assembly will be the prop shaft. Gear reduction is used between the crankshaft and the prop shaft in the Quarter Scale and its ratio is a "closer to scale" 48:21. This screwy ratio required the 720 degree crankshaft cycle to be mapped into just 315 degrees of the degree wheel. Because the ratio isn't a nice integral number like 2:1, there was also a 45 degree gap in the wheel that's equivalent to just under 103 crankshaft degrees. If I had realized this little surprise was on the horizon, I might have considered altering the ratio to at least reduce the gap. Unfortunately, that train left the station a long time ago. The biggest issue with the wheel turned out to be wrapping my head around its construction. It wasn't all that difficult to use once I made my peace with its limitations and gained some experience using it

A cam timing diagram mapped onto a geared-down degree wheel can be easier to read, but the loss in resolution will limit the accuracy of the valve adjustments. The referenced piston's TDC will remain synch'd to the degree wheel only during the revolution on which the two were initially referenced. This needs to be kept in mind to avoid some serious head scratching later. I made a second diagram without the valve information for the opposite face of the wheel to use for checking the timing. The wheel gap causes the firing edge measurements to slip some 17 crankshaft degrees per each complete wheel revolution, and so this is another landmine that's best stepped around.

At the end of the day, the sensor's turn-OFF edges appeared to be consistent and spot on with respect to one another within the two degree measurement resolution I probably had to work with. After completing the measurements, I chose to turn the sensor's led indicator ON when the trigger disk aperture exposes the sensor to the magnet and to trigger the CDI on the opposite edge when the led turns OFF. This is the same convention used on my two radials, and it mimics the old Kettering systems where the led ON time is a mock dwell. I verified the timing adjuster was capable of advancing the timing a maximum of 18 crankshaft degrees beyond its initial setting and the plugs continued to consistently fire at this maximum advance.

For testing, I used a fixture made during my last radial build. In addition to an led driver it contains an old style S/S Engineering CDI module. The starboard distributor worked as expected, and the plugs fired in their 1-4-2-6-3-5 order while the prop was manually rotated in its normal running direction. Similar tests with similar results were obtained on the port side distributor. Before moving on to the cylinder liners, the next step will be to construct and test the actual ignition modules that will be used to run the engine. - Terry

 

[mayhugh1]

For the Merlin's ignitions I re-used some of the previous development work from my 18-cylinder radial build. The CDI modules used on that engine were older generation units previously sold by Roy Sholl. These particular units had been modified by Roy to have a half-size discharge capacitor in order to obtain an output of 30k sparks/minute. The Merlin uses the same CDI's, but without the capacitor modification. Since each Merlin ignition will be responsible for only six cylinders running at a maximum 3600 crankshaft rpm, the unmodified units should easily support the Quarter Scale's 11k sparks/minute requirement. Theoretically, going back to the original capacitor should allow these CDI's to output twice the energy per plug compared with the modified units. The additional plug current that will be available should help mitigate the Vipers' tiny plug gaps (.009" vs. .018" for the CM-6's). Although smaller gaps can offer some advantage in a rich-running cylinder, they are more susceptible to oil fouling.

Each ignition includes a small circuit board of my own design. Basically, it's just an LED driver that provides an indication of the Hall sensor's output state without the need to power up the CDI. A 'Mag' toggle switch on this board allows the engine to be timed using the Hall sensor without having to worry about generating unintentional and potentially damaging sparks. Both of my radial ignitions included a similar PCB, but after the boards were designed and populated I decided to add a pair of transient protectors across the Hall sensor inputs. These large components had to be kludged into the radial's ignition enclosures, and the results weren't pretty. I re-did the layout of the circuit boards for the Merlin ignitions and increased their size in order to accommodate the transient protectors.

The circuit boards were designed in SolidWorks, and the single-sided copper layer was milled on my Tormach using a 1/16" diameter end mill in a 3x spindle multiplier. The .030" diameter holes were drilled using an eBay circuit board drill. Fabricating a simple board in this manner is reasonable and fairly quick, but laying one out in a general purpose CAD program like SolidWorks was a long and frustrating experience. After designing the original board for the radial I didn't think I'd ever do another one in that way, but I ended up spending several more hours on it adding those two extra components.

The enclosures I designed for the radial had to be increased in size to accommodate the new circuit boards. The Merlin's enclosures were machined from a block of gray PVC that I had acquired long ago. I'd never before milled PVC in my shop, and I ended up learning a 'shocking' lesson while working with it. Sometimes when machining plastic or wood on my Tormach, I'll occasionally vacuum up some of the chips with my Shop-Vac while the machine is working. Instead of flood coolant I use a Micro-drop system, and its compressed air spray can churn up the lightweight chips and blow them everywhere. This began to get out of hand during the machining of the enclosures after a thick layer of PVC chips had accumulated over the vise and much of the table. A few seconds after beginning to vacuum up the chips, a one inch stinging arc jumped between the vise and my hand that was holding the Shop-Vac hose. This was pretty surprising since the humidity here in central Texas has been insufferable during the past month. Mach 3 went off into the weeds and took the machine's work offsets with it. I was able to recover the workpiece, but I put the Shop-Vac away and finished up using a chip brush. I'd run a static drain wire down though the Shop-Vac's hose if it weren't for the fact that it has to frequently come off so I can clear metal chip blockages at the entrance to the main canister. I've used this Shop-Vac on large black Delrin machining jobs in the past with no issues, and so this experience might be evidence of (black) Delrin's inferior dielectric properties compared with PVC.

The rear face of each enclosure includes a machined boss into which a short Futaba J female connectorized cable was epoxied. This connector mates with the male sensor cable from the distributor so the sensor can be powered up and its signal brought inside the ignition module. A short piece of shrink-fit tubing will be used to form a slip-fit cover over the junction between the two connectors in order to protect the electrical interface from the exhaust. The Mag switch protrudes through the lid of each enclosure along with a pair of LED indicators. An amber LED indicates the output state of the sensor, and the red LED warns that the 'mag' is alive. The red indicator and its dropping resistor are actually soldered to a tiny breadboard area available in one of the corners of the CDI. There's actually a pair of amber LED's so the state of the sensor can be seen from either side of the panel on which the enclosure will be mounted. I had a bit of fun with the design of the enclosure's lid and machined it to look like a finned heat sink.

The center conductors of the coil output leads were soldered to some mysterious phosphor bronze spring terminals that were in my electronic scrap collection. I made up the coil boots by supergluing a short length of 1/8" rubber tubing to the end of a 3/8" automotive vacuum plug.

After the ignition modules were completed, I used them to replace the previous distributor test set so they could be tested with the distributors and the CM-6 plugs. There was a noticeable increase in the intensity of the sparks in the plug gaps which was probably due to the higher energy output of the unmodified CDI modules. (The CDI in the distributor test set contained the half-size capacitor.) During one of my tests I became careless and forgot to connect the ignition module's engine ground to the return on the spark plug fixture and quickly blew out the sensor. Not connecting the high voltage return is, of course, a major no-no. But, I was really disappointed to learn that the transient protectors that had taken so much time and effort to add evidently don't protect the sensors after all. - Terry

 

[10K Pete]

What can I say but... WOW! Really great work.

Pete

 

[JaamieG]

Superb work!

 

[Ken I]

You surge protectors are only going to arrest the HT after it has passed through the sensor (or its wiring).

Murphy's law is never on your side.

A delicate transistor protected by a fast blowing fuse invariably becomes a fuse protected by fast blowing transistor.

Epic build and thread.

Regards
Ken

 

[mayhugh1]

The next step in this build was to finalize the design of the cylinder liners so the designs for the connecting rods and pistons could also be finalized. These are the last major components to be machined, and the interactions among them have to be carefully checked. This particular step wasn't one of my favorites because it involved sitting in front of a computer for the last week instead of making parts and progress in my shop. But it was an important step, and for anyone with a set of these castings it may save a lot of headaches and re-work later on. I've been told by Richard Maheu that there are between 50 and 100 sets of his castings out here somewhere.

The Quarter Scale's stock liners are probably faithfully scaled replicas from the full-size Merlin's top end and, of the parts remaining to be machined, these will likely be the most difficult. In addition to the features required to seal the combustion chambers and the coolant jackets around them, the stock slip-fit liners have a worrisome wall thickness of only .035". The Quarter Scale's stock bore is 1.35", and with a stroke of 1.5" the total displacement is nearly 26 cubic inches.

My concern with the Merlin's thin wall slip-in liners arises from my experience with the cast iron liners in the Howell V-4 which was my second engine build. Finishing the i.d.'s of its 1/16" liners to within two to three tenths of being truly circular for a proper fit to its cast iron rings ended up being much more difficult than I had expected. All the machining and finishing had to be done outside the engine block since the liners weren't pressed into place but were supported with o-rings. Not only were the machining steps and their order important, but work-holding was especially critical to prevent workpiece distortion during both machining and honing. Material selection was also important not only from a functional perspective but also for dimensional stability during and after machining. I machined liners from three different lots of cast iron before I finally had four acceptable parts. I'm sure that being a newbie had a lot to do with my problems, but the larger physical size of the Merlin's liners makes their thinner walls scary even with the experience I've accumulated since my Howell days. For this engine, I'm going to need to end up with at least a dozen usable parts.

In addition to wall thickness, there are other reasons to re-consider the design of the stock liners. Poor statistics are available on the Quarter Scale's ability to run without overheating, and John Ramm may be the only builder to have accomplished this. The stock cylinder design provides for less than a teaspoon of coolant around each cylinder, and the lack of space between bores prevents any enlargement of the jackets inside the cylinder blocks. The only way to increase the volume of coolant around the liners is to reduce their diameter, and this will also reduce the engine's displacement.

The liners also affect the static compression ratio. If the Quarter Scale's c.r. is calculated using the typical simplified model of a stock piston sitting at TDC in a cylindrical combustion chamber, the result is 7.9 compared with 6 for the full-scale Merlin. I did a closer calculation using measurements on my completed heads to account for volume reductions created by the valve heads and coolant tunnels that protrude into the combustion chambers. The result was a whopping 9.7.

High compression ratios create needless problems in multi-cylinder four stroke model engines intended to run as displays. High cylinder pressures aggravate marginal cooling systems, and they increase starting torque requirements. Starting torque is a particular concern in the Quarter Scale because of its unproven electrical starting system. John's last video,
http://www.homemodelenginemachinist.com/showthread.php?t=25754, shows him hand starting his engine because of starting system difficulties that arose from increased cylinder pressures after the rings and valves bedded in.

Finally, measurements made on my engine's completed head assemblies show that it is currently within .005" of becoming an interference engine if completed using the stock rods and pistons. Although not fatal, while doing engine timing adjustments the fully open valves will come very close to extending into the space in which the pistons move.

All these potential issues are a result of the interactions among the designs of the liners, rods, and pistons as well as an accumulation of the machining tolerances of many, many already machined parts. The liner design can be adjusted to compensate for all these issues, but changes to the liners will likely introduce clearance issues between them and the connecting rods. A reliable method for verifying the clearances between the rods and the stationary parts of the engine is needed before making any rod design changes. Rod clearance issues typically don't occur at TDC or BDC which are easy positions to sketch on paper. In order to continuously visualize what will be going on during an entire crankshaft revolution, a CAD assembly model was used. A partial SolidWorks assembly was created that included a pair of cylinders, liners, heads, pistons, rods, and the crankshaft. The virtual crankshaft was rotated in order to spot clearance issues throughout the entire range of complex motions of both the blade and fork rods. Simple models for the heads, cylinder blocks, and crankcase were created using actual measurements taken from my already machined castings. Since I already had a full model of the crankshaft, I had only to create additional full models for the as yet un-machined rods, pistons, and liners.

The first issue revealed by the modeling was a minimum clearance of only .005" between the stock rods and the bottom edges of the stock liners during their closest approach. This insured that any reduction in the liner i.d. would definitely require a change to the stock rod design. I've included a few cross-sectional CAD snapshots taken showing the stock rods and pistons in a few locations of interest inside the stock liners.

I began design changes on the CAD assembly model by shortening the pistons by .023" above their wrist pins in order to increase their clearances to the fully opened valves at TDC. This also had the effect of dropping the compression ratio from 9.7 to 8.6. The i.d.'s and o.d.'s of the liners were then reduced in order to increase the liners' wall thickness and the coolant volumes around them. The i.d. reduction was limited by the additional loss in compression ratio that I was willing to accept. After several iterations, I arrived at a new i.d. of 1.2" down from 1.35" and a new o.d. of 1.35" down from 1.42". This change increased the liner wall thickness from .035" to .075", and the coolant jacket volume by 80%. The new wall thickness was still a bit less than I had been hoping for, but the compression ratio had decreased to 6.2, and I wasn't willing to reduce it any further. This change resulted in the engine's displacement dropping from 27 c.i. to just over 21 c.i..

As expected, the liner i.d. reduction created an interference between the bottom edges of the liners and the stock connecting rods. Before modifying the rods, the liners were shortened by .075" which brought their bottom ends flush with the interior surfaces of the crankcase. I also added a 60 degree chamfer to the inside edge of the liners' lower ends in order to ease assembly with the ringed pistons later during final assembly. In addition to reducing the diameter of the pistons to fit the new liners, they were also shortened by .075" to match the new liners at BDC. The rods were finally modified to eliminate the interference that was created by the changes, and the minimum clearance ended up at .060" compared with .005" for the stock configuration. Removing the rod material weakened them slightly, but the mass removed from the pistons as well as the c.r. reduction most likely mitigated all the loss in strength.

Although the reductions in displacement and compression ratio may seem rather harsh, I believe they are now more in line with the expectations for a typical model engine, and hopefully they will improve its run-ability later when/if it runs. - Terry

 

[mayhugh1]

[Your surge protectors are only going to arrest the HT after it has passed through the sensor (or its wiring).

Murphy's law is never on your side.

A delicate transistor protected by a fast blowing fuse invariably becomes a fuse protected by fast blowing transistor.]

Ken,
I think you're correct. I had neglected to take into account the slow speed of the large area transient protectors I selected. I've been thinking about adding some protective spark gaps inside the ignition enclosures which is probably the only real solution. I've thought about doing this during my last few builds but have been hesitant because I really haven't had problems with blowing up Hall sensors, and the gaps will take some time to experimentally set properly so they don't become their own nuissances. - Terry

 

[petertha]

I machined liners from three different lots of cast iron before I finally had four acceptable parts.

Hi Terry, can you elaborate on this prior experience? You mean different CI grades or suppliers or property variations in the same stick...? What determined non-acceptability? Also I've heard people mention different machining properties within core of a stick vs. near its OD. I never paid much attention if this related to drawn rods vs. stress proof vs. CI. Do you think there is something to this in CI? So what will be your material/source for this build?

Can you elaborate on the sealing aspect. I can understand the O-ring on lower side, but why not anything on upper side (red arrow)? Also, maybe I missed the purpose of the extending ledge on your revision & then O-ring beneath that?

 

[mayhugh1]

Peter,
All the c.i. material used in the V-4 was class 40 cast iron, but it had been purchased at different times from different suppliers and was, in fact, remnants from other projects I had been involved with. I had trouble with the rings sealing on the V-4, and after checking the ring fit of my first set of liners with a light source, I traced the problem to the liners, themselves, which weren't perfectly circular. After discovering that I actually needed to measure the circularity of the liners with a dial bore gage after they were machined instead of just assuming they were perfect, I learned just how difficult it was to not only machine a thin-wall slip-in liner truly circular but also how difficult it was get it to remain so.
The upper stock coolant jacket seal is just a close metal-to metal seal. I have questions about it, myself and am thinking about including an o-ring. The bottom ledge looks a lot wider in comparison to the stock liner because the o.d. of the liner was reduced above it. There is an error in that cross-section drawing, though. The stock ring collar above the o-ring is no longer required, because it was integrated into that ledge. I forgot to remove the collar from the assembly drawing of the redesigned version. I currently plan to use Stress-proof for the liners.
I, too, have had suspicions about that outer layer of metal in a centrifugally-cast c.i. rod. Many use it and purchase a 1" diameter cast iron rod when tbey need to make a one inch diameter part since the rod will typically be another eighth or so larger in diameter. I, myself, machine the outer layer away rather than take a chance on using it. I figure the manufacturer probably doesn't consider it to be of the same quality as the underlying material or he wouldn't be giving it away for 'free'. - Terry