Another Radial - this time 18 Cylinders

Home Model Engine Machinist Forum

Help Support Home Model Engine Machinist Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
My goal was, as you said, to first get the all the cylinders to a consistent bore diameter. I had a lot of time invested in their external features that I didn't want to risk, but I knew I could eventually lap them all to a common diameter even though at the time I didn't know what that diameter would be. It would just be a matter of me putting the necessary time into a process that removes material very slowly.
After finishing the cylinders I decided to make the pistons. The rings could just as easily done at this time instead since there really wasn't any close fits involving the pistons with either the cylinders or rings. By 'close fits' I mean fits that have to be verified by actually trying them because they may be too close to rely only on my measurement ability.
After finishing the pistons I'm now moving onto the rings. Even though I now know what their o.d.'s need to be after measuring the common i.d. of the cylinders, I won't be able to test such a close fit in the cylinders without risking damage to their finished walls until the rings are gapped. And, since the heat treatment may change their shape slightly I really won't be able to check their final fit until after heat treating. Then I'll use a light test to verify their contact patch to the cylinder wall. If the batch fails my light test it will be scrapped, and I'll mitigate my losses by limiting the batch size to ten or so rings which is the number my heat treat fixture can handle at one time. Scrapping a few batches of rings in my particular case at this particular time is much preferable to scrapping even one completed cylinder. The truth is, the scrapped rings will go into a labeled box and may end up perfectly fitting a cylinder in some future project.
My comment that you asked about concerning an undersize or oversize ring was trying to say that I felt it was better to be slightly over-size on the ring o.d. than it is to be undersize. The light pattern will show 2-3 very narrow point contacts on a slightly oversize ring that will quickly wear down to give a perfectly fitted ring. A slightly undersize ring will have a wide non-contact and a wide contact area that will take much longer to wear down and seal to the cylinder bore. It is totally a judgement call on what to accept and what to scrap. I will likely start out rejecting any ring that shows any light at all until my scrap rate is so high that it starts wearing me down. I'll then start accepting some slightly oversize rings.
Now I'll try to answer the main question of your post. If you're planning on using commercial rings I see nothing wrong with lapping your cylinders to the o.d.'s required by the rings. There is a slightly greater rusk to your cylinders in doing this compared with what I did, but what you are proposing is very reasonable and done by others all the time. If it were me, I would have the actual rings in hand before I started lapping and, in addition to actually measuring the lapped bores, I would perform a light test with the ring in its actual cylinder when I got close to the finished value. Once I was satisfied with the light test results I would make sure those rings stayed with those particular cylinders. - Terry
There certainly are some clever people in this world, and you sure are one of them, love your tooling and the indent explanation,
Cant wait to hear this baby fire up
For my piston rings I gathered up a couple one foot lengths of one inch diameter cast iron obtained from long-forgotten sources several years ago. Some of this 'aged' material has already been used for rings for a couple earlier engines including my H-9. Class 40 gray or just 'gray' cast iron is a theoretically good choice for piston rings because of its close grain structure, natural lubricity, and good machinability. Except for the messy, abrasive residue left behind, the excellent surface finishes that are readily obtainable with this material makes it a good candidate for sealing surfaces. One inch cast iron rounds are cast some 10% over-size; and so, by design, there is sufficient excess material from which to machine one inch diameter parts.
I've used George Trimble's method for making most of my rings, and my own experience has been that the most difficult step in his process is creating an ideal cylinder of the proper diameter from which to slice the rings. Once the finished blanks are completed, the rest of his process is pretty straight forward. I briefly experimented with an alternative method for making rings during my H-9 build that included a final o.d. turning operation on the completed ring. This method requires yet another precision fixture to be machined to the cylinder's exact i.d.. The reasons I went back to Trimble's method, after only a few parts, were the difficulties I had with polishing the o.d. as well as using the fixture for more than one part at a time.
I started the rings for my T-18 by saw-cutting my one inch stock into ten 2-1/2" long pieces in order to create starting blanks for ten batches of rings. I need a minimum of 36 compression rings and 18 oil rings; and, with luck, each blank could potentially yield more than half of what I need of either ring. From experience, though, I've learned there will likely be a high scrap rate while machining the blanks.
The finished o.d. and i.d. targets for my rings are .9978" and .9120", respectively. The o.d. target was set .0004" greater than the i.d. of my cylinders to help me avoid ending up undersize. Using my 12X36 Enco lathe I rough turned the o.d.'s of the blanks to 1.015" and then drilled out their centers with a .812" drill. These two roughing operations removed a lot of material and likely changed the distribution of any stresses remaining in the cast material after it was cooled.
George Trimble's recipe requires a final stress relieving heat treatment to set the expanded shape of the ring. If Trimble's method is closely followed, the heat treated ring will, theoretically, fit the cylinder bore perfectly when it is compressed by being inserted into it. If there are stresses related to the casting process remaining in the metal after the ring's machining is completed, the shape of the ring could change from Trimble's theoretical shape during this final heat treatment when those stresses are relieved. Therefore, I performed an initial stress relieving operation on the rough blanks before the ring machining was actually started. The ten blanks were heated to 920F for one hour and then allowed to slowly cool overnight before doing any finish machining on them. For this heat cycle, the blanks were wrapped in a double-folded stainless steel envelope filled with argon to avoid scaling.
After the blanks cooled they were taken to my 9X20 lathe which has tight bearings and the capability of turning nearly perfectly round parts. Here, the centers of the blanks were bored out to the finished i.d. of my rings. This left a cylindrical workpiece with only a .051" wall. I did not, however, bore completely through to the portion of the blank inside the lathe's collet chuck. I used a high rake Korloy carbide insert designed for aluminum in order to give nice surface finish to the bore. The critical o.d.'s were then turned to .9983" which is .0005" over the ring's finished o.d.. I then used abrasive papers in 400g and 600g to polish the o.d.'s of the blanks to .9978" +0/-.0001" over their full lengths outside the chuck. I tracked my progress on worksheets by carefully measuring at three points with respect to a reference mark on the cylinder and working down the high areas.
Finishing the o.d.'s is the most critical step in the blank creation process, and there are at least three potential issues with which to be aware.
The easiest issue to deal with is the surface finish. Just as with my valves, I removed the machining marks by polishing the o.d.'s to a final bright finish; but with the ring blanks I stopped with 600 grit paper.
Matching the finished o.d. of the blank to the cylinder's i.d. is a moderately difficult step. In my shop it is primarily a measurement issue of trying to match the o.d. of one part to the i.d. of another part using two different measuring instruments at a precision to which neither is actually capable. I used the same micrometer for measuring the o.d. of the blank that was used to measure over the cylinder's bore indicator. I also allowed the blanks to cool before measuring them whenever the polishing managed to get them too hot to touch.
The most difficult issue that I dealt with while creating the blanks, though, was turning them truly round. Lots of subtle sources affecting this error seem to creep into the process. My 9X20's spindle bearings, limit the minimum circularity error to .0001". While gripping these thin-walled cylindrical blanks my economy 5C collets can add up to another .0005" error over a significant portion of the blank even if the collet chuck is only moderately tightened. I bored only the portion of the blank outside the chuck to the final ring's o.d. in order to leave the portion of the blank inside the collet with a thicker wall in order to help reduce the collet error.
It seems that in some batches of cast iron even my initial stress relief heat treatment can't seem to stop the thin-walled cylinder's shape from moving around uncontrollably as it is machined. My lathe's spindle vibration at the high speeds I use for polishing also contributes circularity errors. Fortunately, I can continuously vary the spindle speed and find 'quiet' windows by lightly laying my hand on the workpiece.
In this project I came across a new source of circularity error. The photo of the o.d. finishing operation show the outer end of a metal slug dampener I inserted into the end of the blank during machining to prevent squeal-induced chatter. After boring the i.d. the blank becomes a chime just waiting to ring when it makes contact with the o.d. turning tool. This slug of metal, wrapped with plastic electrical tape was loosely inserted into the workpiece and completely eliminated the squeal. However, the undetectable wobble it evidently created resulted in a whopping .0007" circularity error on three of my blanks before I realized what was going on. I later found that a dampener made from a short piece of wood dowel, wrapped with tape, and inserted into the tube also deadened the cutter squeal just as effectively as the metal slug without affecting the circularity of the blank.
The final photo shows the completed blanks ready to be sliced into rings. The areas marked in red are the areas I've tentatively chosen to scrap due, primarily, to circularity errors. I ended up with only a 50% yield of 'good' material. My arbitrary cut-off point for this 'good' material was at a .0003" maximum circularity error. The top compression rings will likely be first selected from this material. At this point, though, I don't really know how big of an error is too big and whether I may be discarding useable material. I hope to run some 'eclipse' tests by shining light through a test cylinder containing a ringed dummy piston in order to compare ring/cylinder contact patches for various degrees of circularity and diametrical errors. - Terry










I have watched most of your previous thread and all of this one and would like to say how much I enjoy all you do.I especialy like your self imposed quality controls not to mention your beautiful wrkmanship. I also enloy your depatures from the set plans and really like the results on the cylinders and heads. Thank you. Buchanan
This maybe a stupid question, but wouldn't the ring match the bore after it is run? so chasing 0.003 maybe in vain, I never built anything like this so I wouldn't know
Beautiful work, with an incredible eye for detail,
No doubt you're probably right. I've just never seen any real data showing what, in the real world, makes the difference between a good and a great piston ring.
For me, pretty much all the satisfaction of working on one of these long term projects is trying to understand and trying to do the best possible job I can with all the minutia that contribute to an engine's operation. It isn't at all important to me to quickly get one running so I can move on to the next one. I try to be careful in my write-ups to be clear that what I document is only what I've done and in no way is it the 'best' - whatever that means - way to do it. I go down lots of questionable paths for nothing more than the opportunity of a different path. It's much more important at this point in my life that I experience a thoughtful journey than it is that I arrive on time. - Terry

I too am old enough to agree with you. That was a great explanation !
Remember when somebody saw you limping and asked what happened and you got to tell a cool story about flipping a dirt bike on a hill climb ?
And now the explanation is more like " I sneezed in the shower and threw my back out"

Getting older is not all it's cracked up to be.

I'm recovering from a hernia operation right now.

Years of moving car engines and Bridgeport mills around didn't cause it.
Had a violent coughing fit while brushing my teeth and "busted a gut".


And now the explanation is more like " I sneezed in the shower and threw my back out"

Getting older is not all it's cracked up to be.

Let me start off by saying how much I enjoy following your build. Knowing the time and effort it takes to document a large project like your radial I can appreciate what you're doing.
Your processes are much similar to mine with the exception of owning and using bore gauges. I have made up an attachment to hold a dial test indicator that is a poor man's I.D. measuring tool.
I use brass laps like you have used and find them to work very well.
As far as piston ring making I too use the Trimble method. I have read and tried several of the other processes and they do wok but the extra cutting and fixtures required just isn't worth the time.
After installing my rings and running them in on the bench I find that after about 1/2 hour of running time the rings usually have the darkened color from the heat treat process polished off. Sometimes there's an area of say 10-15 degrees that is still dark so I'm guessing we're talking .00005 difference and will polish out once the engine is running.
In the construction of an I.C. engine I have found that the ring/bore fit isn't quite as critical for good operation as getting the valves to seat properly. When you consider that most of these engines will idle somewhere around 1000 rpm the ring leakage should be minimal but with a leaky valve it never goes away.
Keep up the great work.
I had to make/gather up some tools and gages before continuing with the rings. The first 'tool' is a simple piece of aluminum round stock with a turned end such that a ring can be slipped onto it and held consistently by its i.d. against a flat surface while the sides of the rings are being lapped. This lapping will bring the width of the ring to its finished value so there is a proper clearance between it and the walls of the piston grooves. Lapping will also create the ring sealing surfaces that bear against the piston groove walls.
The second fixture is short piece of scrap steel with a bored i.d. that matches the engine cylinders' i.d.. This gage will be used to set and measure the running gap of the finished ring after heat treatment. An actual cylinder could be used instead, but with so many rings to deal with, there is risk of unnecessarily scratching up the bore of a pristine cylinder.
The third fixture is, in my case, an overly elaborate tool that I made several years ago to cleave rings just before they are heat treated. For my first two engines I simply broke the rings over a piece of 16 gauge wire on a surface plate by applying thumb pressure to the ring on either side of the wire. After moving up to more performant oil control rings, I needed a way to better control the actual location of the break. I allowed the construction of this tool get out of hand, and it turned into a full weekend project several years ago.
The heat treat fixture for 'setting' the uncompressed ring shape is a subtly important tool, and its Trimble specific design details can be found on the web. If the theory behind Trimble's method is carefully studied it may become apparent that there is more to the design of the fixture's gap spreading wedge than is noticed at first glance. This fixture is designed around a particular diameter ring, and a new one is required for a different size ring. I made a small modification to the original design by lengthening it so I could heat treat large batches of rings for my H-9. I also made a spacer so the same fixture could also handle small batches.
The last tool is new for me and will be used to examine the precision of the final fit of a completed ring in an actual cylinder. Although a compression check is the final word on the fit, a light leak test may be useful for eliminating poorly fitting rings before an engine is actually assembled. (Remember buying new cars in the good old days when oil consumption was a crap shoot? I had more than one dealer in the eighties tell me that burning a quart of oil every thousand miles, even after 4000 miles on the odometer, was within factory spec and not covered by a new car warranty.) Basically, the ring under test is inserted onto the end of a short length of black Delrin rod that has been turned to the pistons' diameter. The top end of the rod was turned to match the i.d. of the ring. In use, the ring is slipped onto the end of the Delrin rod, and the pair are inserted into the cylinder through its bottom and pushed up until the ring is near its position in the cylinder during TDC. A white light is then shown into the bottom of the cylinder, and any light escaping from the top of the combination is an indication of an imperfectly fitting ring. The give-away Harbor Freight led flashlights are not bright enough for this application. I had to use a 250 lumen miniature flashlight spaced away from the bottom of the cylinder by a trial-and-error derived distance in order to get enough light into the gap between the dummy Delrin piston and the cylinder wall in order to produce consistent results.
The difficulty with the process with which I'm planning to experiment will be judging the difference between a useable and an un-useable piston ring. I'm hoping that the rings from my measured good blanks will pass no light, and the rings cut from the imperfect blank areas will pass varying degrees of light for which I can, over time, develop some sort of pass-fail criteria for 'good enough' rings. I'm not expecting things to come out that 'black-or-white' though, and so this may turn into one of those interesting but not really useful exercises.
I took two preliminary test photos while developing the fixture. The first photo shows the light escaping from the top of my .001" oversize cylinder. The ring in the cylinder was cut from one of my on-size scrap blanks that had a .0004" circularity error. The ring was lapped but un-gapped, and it had not yet been heat treated. This test was used to decide if our household flashlight was bright enough to get a meaningful result. The second photo is the light passing through one of my production cylinders using the same ring. There is a clear difference in the light pattern between the two cylinders. Because the end of the Delrin piston was machined to give the same ring back clearance as an actual piston, the rings are not necessarily centered in the cylinders' bores. If the rings had been gapped and heat treated the patterns would have been somewhat different since a portion of the ring would have been sprung into contact with the cylinder wall. The total emitted light in each case, though, would have remained about the same. I think these early tests show that more experiments are worth continuing on finished rings. - Terry









I continued my ring construction with the goal of creating the best possible initial fitting rings by measuring and recording one of my good blank's dimensions one last time before slicing it up into 26 candidate compression rings. I decided to polish off the excess .0003" that I had previously left on this blank, though, so it would exactly match the i.d. of my cylinders. I did this so I could establish some kind of baseline for the light tests I had planned.
I used my 9X20 lathe, a .019" carbide grooving insert, and a few lines of scratch-written code to uniformly part off the individual rings. I ran the spindle at 1000rpm which was a little high for one inch cast iron, but I used an extremely low (.0001 inch/rev) feed-rate hoping to not raise any burrs on the rings' corners. My tape-wrapped wood dampener eliminated the flash on the parted off rings by holding them in place until they were completely cut free.
But, as the first ring photos shows, there were still slight burrs raised on the o.d. and i.d. corners of each ring. I removed the i.d. burrs by manually working a very hard 1/4" diameter cylindrical ceramic stone around both i.d. corners of each ring in order to break them. I removed the o.d. burrs by manually rotating the rings' o.d. corners against a sheet of 1000 grit paper for a couple revolutions using light pressure.
The rings came off the lathe having a width equal to that of the piston ring groove. The lapping tool was then used with 1200 grit lapping grease against a glass plate to remove .0005" from each side of the ring. Lapping reduced the total width by .001" which is my target piston groove clearance. (By the way, I think it is this clearance plus the ring gap through which crankcase oil flows in a stored radial engine to accumulate in the combustion chamber and possibly hydrolock the piston.) At this point, as a second ring photo shows, all the rings' surfaces were nicely polished.
The final step before heat treatment was breaking the rings using my shop-made cleaver. A folded-over two inch long strip of 1000 grit paper was pulled through the break just once to lightly clean up the faces. The actual running gap was set after heat treatment. The de-burring, lapping, cleaning, and measuring added some 10 minutes of tedious work to each ring; and so cleaving became a celebratory step in the process. The whole first batch of 26 rings was loaded into the heat treat fixture, and it was enclosed in a double-folded stainless steel foil bag filled with argon gas. The package was raised to a temperature of 970F (this is considerably lower than the Trimble's 1475F) for 1-3/4 hours and then allowed to cool overnight to room temperature. I've found that if the stainless foil bag isn't pre-conditioned with a heat cycle of its own, the rings will come out of the fixture stuck together, and they will have to be carefully separated with an X-Acto knife. In addition there will be a thin powdery coating on them that isn't difficult to remove. I believe this happens because the stainless foil probably comes with some sort of oil coating on it that is burned off in the heat cycle.
The running gap was then set for each ring using a diamond file and the cylindrical bore gage mentioned earlier. The purpose of the ring's running gap is to allow the ring to expand without breaking when it's exposed to the high temperatures of combustion. I use the same .004" gap that many others use. Although it might come as a surprise to a few, this ring gap creates an all but invisible leak in the combustion chamber, and this can be demonstrated with some simple math. The opening to the crankcase created by the ring gap is bordered by the o.d. of the piston, the i.d. of the cylinder, and the two ends of the gapped ring. For a .003" diametrical piston clearance and a .004" gap the area of this leak is only 6 millionths of a square inch. It's difficult to appreciate the significance of a leak this small in the combustion chamber; but if the volume of the combustion chamber at TDC is scaled up to the volume of a 50 gallon drum, this leak would be equivalent to a hole the size of a lead pencil punched in its bottom. Even under combustion pressure, it's only the amount of air that can escape from the drum during some worst-case tenth second idle power pulse that is lost.
After setting the running gap I lightly re-lapped the sides of the rings and de-burred the area around the gap. The third ring photo shows a typical finished ring. I checked several of the rings in my light fixture using various cylinders and could see absolutely no light escaping anywhere except for a brilliant pinhole of light coming from the ring gap. I was pleasantly surprised because I thought there might be some visible light on the ring's perimeter due to the theoretical impossibility of obtaining a perfect fit with the .0001" circularity errors in both my cylinders and rings. My biggest surprise came later when I got exactly the same light test results using the same ring with my .001" oversize cylinder. With the ring sprung outward against the wall of the cylinder, the light loss picture improved dramatically compared with the same un-gapped ring in the same oversize cylinder shown in the photo of my previous post. Where did all the leak indicating light go? It appears that even a .001" diametrical error doesn't create a sufficient curvature mismatch with the cylinder wall to create a measurable leak. Most of the area mismatch that allowed light to pass by the ring in the previous light photo is now hidden within the piston groove. As long as there is an adequate seal between the ring and the lower wall of the piston groove in actual use, this mismatch won't contribute to an actual combustion chamber leak. My conclusion is that a well-lapped ring sealed against the lower wall of the piston groove can cover up some otherwise loose machining on the ring's o.d. There isn't any way to verify the net leakage under these circumstances, though, without a compression test. Unfortunately, circularity errors won't behave in the same favorable way. They will just move slightly around the periphery while maintaining the same leak and light loss. Even worse, the theory behind Trimble's heat treatment falls apart when a significant circularity machining error is present, and the net result becomes fairly indeterminate.
After seeing these results and experiencing the friction of a single ring in a single cylinder of an 18 cylinder engine, I decided to polish off the excess .0003" from the remainder of my blanks before slicing them into rings. During engine assembly I plan to light test each cylinder with only the top compression ring installed on its piston. I'll log the location of each combination and see if there is any correlation between any differences in compression and the light test results for each cylinder at the end of assembly.
A second batch of 28 compression rings was cut from two more partial blanks. I completed these rings using the same process used for the first batch in order to end up with a total of 54 compression rings. This is enough for all pistons I've made including several spares. As it turns out, a spare ring set from my H-9 build will nearly perfectly fit my .001" oversize piston/cylinder combination, and so I won't have to make a special set of rings for it.
The oil rings are my final parts to make, They're more complicated but more interesting parts to make compared with the plain square compression rings. Thankfully, though, I only need to make half as many. - Terry











IThe heat treat fixture for 'setting' the uncompressed ring shape is a subtly important tool, and its Trimble specific design details can be found on the web. - Terry

Hi Terry. Many moons ago I took a crack at writing a spreadsheet with the Trimble design parameters from SIC article, more-so out of interest at that time. I seem to recall other individuals subsequently made corrections (or maybe 'suggested contributions' is a better phrase) to some of the metrics employed. It kind of left me in a limbo state. Sometimes I could back out some resultant numbers on a particular design that referenced Trimble & sometimes not. Do you have a go-to link with what you considered the complete method, or did you base your dimensions entirely on the SIC equations?

I found this link (which then refers 2 other sub-links named Feeney & Cirrus.

Id be happy to write & share a spreadsheet, but would really prefer it had an approval stamp of experience from someone like yourself. Or maybe you did this already?

I was a subscriber to Strictly IC magazine and my source for Trimble's method is his original 1989 articles in vol. 2 nos. 7,8, and 9. His design equations are in the second article.
By the way, George Trimble, in his original SIC no.9 article admitted that practically any of the current ring construction techniques of his day was capable of producing acceptable compression rings if one was willing to let the pressures of combustion wear the ring into the cylinder bore by running the engine under its own power for six to twenty hours. His assumption, of course, was that the ring was machined to yield sufficient initial compression to actually get the engine started.
The oil control rings are an entirely different matter as they see no combustion pressure. These rings rely totally on spring force to hold them against the cylinder wall. Trimble's theoretical work and the extensive testing he did was actually focused on developing a method for creating a perfectly contoured oil ring. Trimble's eventual claim was that his method for setting the open contour of such a ring was the only one at that time that could accomplish this goal. Of course, anyone who claims to have a recipe for perfection is opening himself up for a lot of debate, and that is just what happened. And, it is one of the reasons we have so many ring making religions today. After studying his articles, reading the various debates, and briefly trying at least one alternative, I adopted his method as my own. The only thing I changed was the temperature of the stress relieving step. I was swayed by the arguments of others having much more metallurgy knowledge than me that Trimble's 1475F was unnecessarily high and could create more problems that a more moderate 1000F would solve.
Your spreadsheet is a good idea and maybe it would be a good candidate for one of those 'how-to' articles that this forum has been soliciting. - Terry
I've been just a bit curious about the fact that of the 25 T-28 rings I've since tested using my light fixture, I've seen no light escaping past any of them except for in the area of the gap. I'd like to believe I've made a perfect batch of rings, but there was also the nagging possibility that the test is somehow flawed - maybe the light source isn't bright enough, for instance.
I went through my H-9 spares parts box and came up with three spare cylinders and fifteen spare rings. So, I tested all these rings in one of the spare cylinders for light leakage and got an array of results. Nine of the rings gave results similar to my T-18 cylinders. But six of the H-9 rings had obvious and significant amounts of light passing by them due to imperfect fits with the same cylinder. I've included four example photos of the results. (In these light photos as well as the others I've posted, it's necessary for me to angle and focus the camera to pick up the brightest leak, and in some of the photos there are actually other, smaller leaks that don't show up in the photo.) Both sets of rings came from similar material and received essentially the same final heat treatment using the same fixture. The H-9 blanks, however, did not receive the same pre-machining stress relief heat treatment that I performed on the T-18 blanks. The main differences between them are most likely their circular machining accuracies since my H-9 construction notes don't show me having made any circularity measurements on the blanks from which those rings were made. Since both sets of rings were made using the same equipment, I believe the improved T-18 results are just due to the better care and selection that was done. - Terry




Machining of the actual engine parts concludes with the oil rings. The oil ring design I'm using is straight out of Trimble's SIC no. 9 article. This single-piece double scraper ring with a central oil collection groove is a fairly old school design as far as automotive piston rings go, but it's nicely adaptable to a model engine. Hodgson, in fact, added it to the H-9 design several years ago.
Since the oil rings don't see the pressures of combustion they rely totally on the ring's outward spring force to hold them against the cylinder wall. The edges of the scrapers are slightly beveled to increase their pressure against the cylinder for better oil control. The oil grooves contain ten radially-drilled holes to allow oil collected in the groove to flow into the clearance space behind the ring. An even number of holes insures that one doesn't end up directly across from the ring gap. Another series of radial holes drilled through the piston groove allows this oil to escape into the interior of the piston and drain back into the crankcase. Two pairs of these holes are located on just either side of the wrist pin to aid its lubrication.
In order to do their best possible job the oil rings need to have the correct shape when they're installed. It isn't likely that the oil rings will ever be able to 'wear in' to their cylinders with the flood of oil they constantly experience combined with the lack of combustion pressure.
Of course, it isn't clear that a model engine operating with oil control rings operating at 100% efficiency is a good idea, either. Some oil consumption is beneficial to top-end cylinder and ring wear. What's disappointing, though, is oil fouled plugs and smoky exhausts that tattle poor oil control. The lower cylinders on a radial are particularly vulnerable since the bottoms of the cylinders are continually filled with oil. It's not at all obvious how one designs for the 'goldielocks' operating area where the amount of oil control is just right. In this engine, the approach I've decided to take is to aim for a perfect fitting oil ring in the center of the cylinder when the piston is at the top of its stroke. I'll then let the .0005" diameter outward taper that I've left in the lower portion of my cylinders control the amount of oil that is allowed to reach the second compression ring. Most likely this ideally fit ring will prevent excessive oil from reaching the combustion chamber.
Construction began by re-measuring the blanks I had left over after finishing the compression rings. I was pleasantly surprised to discover that the circularity errors on a couple of my reject blanks had improved to a state of usability during the past week while they were resting and recuperating from their finishing operations. I was disappointed, though, to learn that one of my perfect blanks had taken a turn for the worst and was now scrap. Fortunately, though, I now had more useable material than I had last week. This cast iron instability was reminiscent of my experience with the thin-wall cast iron slip-in, o-ringed liners I made for my Howell V-4. I spent nearly a month dealing with material and process problems that created significant circularity errors in those parts that prevented the rings from properly sealing. For this build I had hoped the stress relief heat treatment that I performed earlier on the ring blanks would eliminate this frustrating issue, but evidently it did not.
Construction resumed on the lathe by first turning all the oil grooves. The blanks were then transferred to the horizontal rotary on the mill where the .020" diameter radial oil holes were drilled. Each blank of 18 rings had 180 of these holes drilled using a carbide circuit board drill. Due to a senior moment that lasted an afternoon, I drilled holes in twice as many blanks as I actually needed. Before I was done I had drilled some 500 holes in three blanks using a single drill bit! After transferring the blanks back to the lathe, the scraper lands were chamfered with a 60 degree threading insert. The rings were then parted off with a carbide grooving insert. After the machining was completed, the rings received the same lapping and heat treatment that was previously performed on the compression rings. I ended up with a total of some 50 oil rings. All of them passed my light test even at the bottom of the cylinder where the i.d. was .0005" over the ring o.d..
Trimble's method for making rings is maligned by some who probably tried it and got poor results after following what they truly believed was his process. Unfortunately, his process has been paraphrased and passed along over the last 25 years with seemingly innocent modifications and 'improvements' that created these results. The gap setting dowel in his heat treat fixture is an example. The theory in his original article showed that its purpose is to resolve his calculated spreading forces into a single vector along the neutral axis of the ring. He is able to do this with a circular dowel of the proper diameter that is in a precise location and tangent to a truly radial ring break. The effect he is trying to produce can't be duplicated with a rectangular wedge and/or a non-radial snapped breaks that I and others have tried to use. In fact, in his article, he graphs possible resulting radial contour errors as large as .010" that can result. An alarming but counter-intuitive statement that he makes is that a ring completed using his method but ending up with a fit issue can't be corrected by re-machining the o.d.. Since he didn't provide any further explanation, I've never been able to understand the reasoning behind his claim.
Trimble's method is but one method for making rings; and if one wants to use it, the original three articles in the back issues of SIC are the best source material for his process. It's probably a mistake to mix portions of his process with portions of other processes without properly working out and testing the applicable theory. The theory he outlines in one of his original articles is not overly difficult to follow and can help the builder appreciate the subtleties involved in his process and its associated hardware. - - Terry










..where the .020" diameter radial oil holes were drilled. Each blank of 18 rings had 180 of these holes drilled using a carbide circuit board drill. - Terry

That's a lot-o-holes. I recognized the drill in your pic as similar to what I bought on ebay. So no problem on the cast iron, just straight in? In general on the CI ring machining, is it usually dry cutting or fluid depending on the job? You likely mentioned this in a prior post, but on the heat setting operation, do you use the sealed stainless foil & sacrificial burn paper to mitigate scale?
That's a lot-o-holes. I recognized the drill in your pic as similar to what I bought on ebay. So no problem on the cast iron, just straight in? In general on the CI ring machining, is it usually dry cutting or fluid depending on the job? You likely mentioned this in a prior post, but on the heat setting operation, do you use the sealed stainless foil & sacrificial burn paper to mitigate scale?

I peck drilled the holes going .025" deep at a time, and I did all the cast iron drilling and cutting dry. The drill bit showed only a slight amount of wear after all those holes. I used the sealed stainless foil wrap, but I filled it with argon gas from one of my welding cylinders instead of using the burn paper. I tried the burn paper technique several times but always ended up with a sticky brown goo on the parts I was heat treating. I tried a couple different lots of brown wrapping paper as well as some white typing paper, but the results were always the same. I was going to try a wooden match stick but then got the idea of using an inert gas and that's what I've been using since. I don't think scaling is a real big problem below 1000F, though. -Terry
One of the most critical assembly steps in this engine is mating the finished heads to the finished cylinders. This is because the heads are permanently screwed onto the cylinders and are, hopefully, sealed by a .032" thick annealed aluminum head gasket. Once these pairs are assembled, the cylinder flanges can finally be drilled for the crankcase studs. In order to drill these flange holes the head/cylinder assembly must be supported so the head's intake flange is oriented precisely perpendicular to the crankshaft axis when the cylinder is installed on the crankcase. This is necessary so the intake tube assemblies will properly align with the intake ports on the crankcase. Once the flange holes are drilled, the head, cylinder, and head gasket are permanently married. If they are ever separated it will be extremely difficult to reassemble them and end up with the exact same head flange orientation. It's for this reason that it's a good idea to have some number of spare head assemblies tested and ready to go.
Since two-thirds of my build time on this engine has, in one way or another, been related to the heads and cylinders, I want the assurance of a final head/cylinder integrity check before drilling these holes. The valve seals in the heads have been previously measured and accepted, and so the spark plug and head gasket seals should be the only new variables in the completed assemblies. In order to verify these, I created a fixture to pressurize each head/cylinder assembly so the total leak-down time of each combination can be measured. The cylinder flange is safely secured to the fixture by a pair of screw-down clamps and sealed to it by an o-ring temporarily placed on the cylinder skirt beneath the flange. A filler block inserted into the cylinder reduces the total pressurized volume to that of the combustion chamber at TDC. This reduction in volume allows the measured leak down times in this test to be directly compared with the leak-down times previously measured and recorded for the individual valves. Since this volume has a direct influence on the maximum combustion pressure in an IC engine, it's a reasonable volume to use for a final leak check. In an earlier post, I predicted a worst-case 2.5 psi drop per second per valve at a cylinder compression test pressure of 75 psi assuming my arbitrary minimum acceptable Mity-Vac valve leak-down time. Using a 75 psi test pressure, a worst-case 5 psi drop per second would be expected if the only leaks in the head/cylinder assembly come from two worst-case valves.
The fixture, itself, was measured and found to leak less than .01 psi/sec at a 75 psi test pressure using a simple block-off plate. The fixture was then tried out on three unused spare H-9 head/cylinder assemblies that I have left over from that build. The leak-downs on those three cylinder assemblies measured 1.2, 1.8, and 4 psi per second; and these measurements correlate reasonably, but not perfectly, well with the valve leak-down results recorded for those heads. Although those numbers are fine, I'm expecting significantly better results from my T-18 assemblies since I put so much more effort into sealing those valves. But, I realize an assembly's leak-down time can be dominated by the worst-case valve especially if the valves' leak rates are well separated.
In order to torque the heads onto the cylinders, the cylinder skirts can be gripped in a vertical 5C collet chuck. The complex shape of the head, though, makes it very difficult to obtain a substantial grip for tightening the assembly even with a strap wrench. I eventually came up with a simple tool made from a brass block silver-soldered to a half inch ratchet adapter that mates with my torque wrench. In use, the block is inserted into the snug fitting space between the bottom pair of valve tower fins and the head is tightened onto the cylinder with a torque wrench. Soft brass was used for the block so it would conform to the radius'd edges of the fins and reduce the applied pressure that might damage them. I also wrapped the block with a protective layer of .002" stainless shim stock, but later used a strip of paper. I experimented with my two test heads and a scrap H-9 cylinder and finally decided upon a torque spec of 35 ft-lbs. I cringed the first few times I applied this tightening stress to the valve towers of my test heads. I was concerned I might be straining the delicate valve seals inside the towers. In order to check on this, I pulled a vacuum on the rear of one of the installed valves with my Mity-Vac and monitored the valve leak down in real time while I repeatedly torqued one of the test heads. I could see no effect on the dynamics of the leak-down, and so I felt a little better about the abuse I had planned for my heads. For peace of mind, though, I decided to continue the Mity-Vac measurements before and after torquing each assembly.
The fixture for drilling the flange holes may be the last significant fixture I will need to machine for this build. Again, the shape of the head complicates any scheme to secure the intake/exhaust flange parallel to one of the axes of my mill so the flange hole pattern can be accurately oriented. The photos show the two-part swan song fixture I came up with. The head's intake/exhaust flange is the important reference surface, and it is directly mounted to the fixture while clamped in the mill vise. The machined parallel surfaces of the fixture insure the parallelism of the intake/exhaust flange to the x-axis of the mill. An adjustable top plate helps to support the assembly during drilling and also helps to keep chips away from the completed assembly underneath. As an extra precaution, the assembly will be wrapped in plastic during drilling to help protect it from cutting oil and chips.
The assembly process starts by selecting a cylinder, head, and piston, all of which are uniquely numbered. Three piston rings are then selected and individually light tested in the cylinder. The rings are then installed on the piston, and it is then inserted into the oiled cylinder and pumped back and forth for about a minute using a spare connecting rod. This is just my sanity check on the fit and allows me to get a hands-on feel for the level of friction. It also kicks off the rather dirty process of rubbing the bluing from the cylinder wall. The piston is then removed, cleaned, and bagged so it can be returned to the same cylinder later.
The Mity-Vac will be used to measure the valve leak-down times before and after torquing the head onto the cylinder at 35 ft-lbs. The exact thickness of the head gasket used is also measured and recorded - just in case. An NGK CM-6 spark plug is installed in the head. I plan to use the same plug for as many assemblies as possible since I have only one of these on hand, and I've always wondered how many tightening cycles the compression washer can take. I've purchase a full box of Rxcel CM-6's that I'm saving to install during final assembly.
The assembly is then placed into the test fixture with the filler block, pressurized to 75 psi, and then total leak-down rate is measured and recorded. The assembly is then moved to the flange drilling fixture where the stud holes are spotted and drilled. Finally, each completed assembly will be set into a cylinder test position on the crankcase where the fit of a test intake tube will be verified. - Terry






After measuring three T-18 assemblies things took an unexpected bad turn. Although I measured acceptable 5 psi/sec leak-down numbers on the first two assemblies, both were curiously identical and more than twice what I was expecting from the Mity-Vac data. The third assembly came in at an even worse 20 psi/sec, and a distinct hiss was coming from the spark plug. Initially, I assumed the compression washer on my used NGK CM-6 plug might have come to its end of life, so I borrowed two more NGK's from another engine. I used one of them to make up a 'test plug' by removing the compression washer and pressing a metal ring around its body for a rubber o-ring to do the sealing for these tests. I retested my third assembly, and the result was a much better 10 psi/sec. When I repeated the tests on my previous two T-18 assemblies using using my test plug I got 0.1 psi/sec. I then re-tested my three H-9 assemblies, and their measurements had also improved a bit and were now in better agreement with their Mity-Vac data. These results told me that my spark plug had been leaking air and had been dominating my leakage measurements.
To be honest, I wasn't entirely surprised to see better leak-down times for some of the assemblies in this test than was predicted by my Mity-Vac tests. When I ran those tests on the finally lapped valves I noticed that on some valves I could improve the results by pressing the valve hard into the seat with my thumb. The results I recorded were, for consistency, only with atmospheric pressure doing the pushing. In this test, the 75 psi pressurized head is adding another 10 pounds of force above atmospheric and probably over-whelming the slight surface irregularities that still remained after lapping. It could also be that the additional force is slightly bending the valve stem in some cases to allow a little better centering of the valve on its seat. In any event, these results showed that my Mity-Vac data should be treated as only a worst-case prediction in this final test.
I never expected to see an air leak at the spark plug, though. The problem with using a special test plug is that I could miss a potentially significant leak.
I decided it was time to break open my new box of Rcxel plugs and test each assembly using a new plug. I tried six of these plugs and could not get even one of them to seal to any of my three T-18 assemblies. The seal was so bad, in fact, that I could hear the tell-tale hiss of a major leak; and, according to my wife, my hearing isn't all that great. I took two photos comparing the compression gaskets of the Rcxel and NGK plugs. The NGK plug creates, in my opinion, a superior seal to the head thanks to a narrow sealing ridge spaced well away from the threads. The rear of the gasket is contoured for what appears to be a nice fit to a tapered area on the plug body adjacent to the hex. The Rcxel gasket, on the other hand, has a much broader sealing surface that is located very close to the threads. The rear sealing surface is similar. My experience with valve sealing has convinced me that a narrow sealing surface is usually a better performer than a wide sealing surface. I think the compression washer on the Rcxel plugs make them much less forgiving against an imperfectly machined engine-side mounting surface.
But, I didn't understand why the mounting surfaces on my heads weren't perfect. I machined, drilled, and tapped them in the same mill set-up. I used a spindle tap starter directly over the drilled holes, and the surface finish couldn't be better. I used the recommended 10mm x 1 tap for the CM-6, and the fit of the thread feels like it should. Still, soapy water showed the leak was coming from a poor seal between the plug and the head. Assuming my machining was somehow flawed, I made up a piloted scraper and manually re-surfaced the top .010" of the head mounting surface. The scraper, however, left machining marks in the surface that made the leak even worse, and so I also had to make a piloted lap to smooth out the surface again. After a full day of entertainment going down this dead end, I ended up with exactly the same unbelievably bad leak.
Since I couldn't see what anything wrong, I tried change the leak in some way hoping I could get another symptom to expose it. I tried Teflon tape on the threads. I tried .005" soft aluminum shim washers under the plugs' compression washers. I even machined and annealed my own compression washers although, in hindsight, they were much too thick at .070" to conform to the head with the torque I was willing to apply to the plug. The results were always the same. The Rxcel plugs hissed and caused immediate leak-downs, and two of the three NGK's sometimes managed to eek out a barely acceptable 10 psi/sec. Clearly, something was hiding in plain sight.
I won't list all the things I tried in order to affect the problem, but my notes show a total of 47 leak-down tests on that third assembly. I suspected my NGK plugs were also complicating the problem because theIr compression washers were, by now, worn out and obviously giving inconsistent results. I had completely given up on the Rcxels, and so I had a local auto parts store special order some new NGK's for me.
A possible cause of the leak came to me while I was studying my notes for the spark plug cavity machining. During machining a burr had been raised around the periphery of the plug hole by the drilling and tapping operations. I manually de-burred the hole at the bottom of the cavity using an oversized reamer since the cavity was too deep for any of my countersinks. It isn't at all obvious to the naked eye while looking into the head's spark plug cavity, but the microphotograph shows the de-burring removed an excess bit of metal near the start of the thread. This effectively left a slight 'gouge' in the head's surface with a depth and width that were dependent upon some variables I wasn't controlling very well. On my third assembly this gouge happens to extends underneath the Rcxel compression washer. However, it barely touches the sealing ridge of the NGK plug. I checked all my T-18 heads, and all of them have this problem to some degree.
I repaired the head of my third assembly by turning a .030" thick stainless steel washer and JB Welding it in the spark plug cavity over the old sealing surface. The i.d. of the disk closely matches the plug's threaded o.d. and the o.d. of the disk is just .010" smaller than the cavity i.d. The JB Weld's service temperature is 600F and should be fine in this application as a leak eliminating filler below the disk. I made a simple threaded fixture out of an Rcxel plug to secure the disk against the old sealing surface and to keep it normal to the threaded hole while the epoxy cures. This new surface will sink the plug about .040", but it is still remains un-shrouded in the combustion chamber. As the final photo shows, I ended up with a more robust sealing surface. The dark rings around the i.d. and o.d. is the JB Weld that flowed into the gaps between the disk and the head. The gouge created by the de-burring is even more prominent now.
After the epoxy cured, I re-measured the leakage with all the various plugs I had been using. My o-ringed test plug gave the best result at 2.2 psi/sec which was almost twice my Mity-Vac predictions. I tried my other two used NGK plugs, and they produced 3.5 psi/sec and 5 psi/sec. Soapy water around the base of these plugs showed that even though the inconsistent leakage numbers were acceptable, the compression washers are shot. The Rcxel plugs were still very disappointing. Only one new one was able to match the leakage of my best used NGK. Several brand new ones still leaked badly with an audible hiss.
My new NGK's arrived, and the three I checked yielded identical 1.2 psi/sec results that matched my Mity-Vac prediction for that assembly almost exactly. My plan is to repeat this fix on two more of my worst-looking heads. If the results still look good, I'll modify all of them including the first two that have already been tested. -Terry