Ohrndorf 5 Cylinder Radial

[Roger B]

Looking good  there's a lot of work in those pieces 

[90LX_Notch]

It’s coming along very nicely petertha.  I’m really starting to think that I need to build a radial someday. 

-Bob

[bent]

With all the radials and rotaries being built lately, I too have had the urge to make one.  Then I lie down for awhile, and the urge passes. 

[petertha]

Next up are the liners. I should mention upfront that the path I took detoured a bit as things progressed, so the pictures might seem a bit out of sequence without some elaboration. When the CI liners were lapped to final bore & heat shrunk into the aluminum cylinders, the cylinders squeezed back under cooling. The bores reduced to the extent that they needed to be lapped all over again to return them to target dimension. I figured any shrinkage would be quite small, like within a few tenths, thus only requiring a quick lapping correction tweak. But they reduced more like 0.0005-0.0015” & also non-linearly down length of liner. Likely a function of the cylinder’s tapered shape squeezing it differently on the top vs. bottom of liner. Anyways the bottom line is my careful bore finishing work before mating into the cylinders kind of went out the window. I could have opted for something more like a slip fit, but I started to read articles suggesting some liner interference is required to achieve proper heat transfer. And as the engine warms up under running condition, a sliding fit will only get looser yet as the aluminum cylinder expands more than CI liner.

My initial workflow was to first establish the bores of the aluminum cylinders to a consistent diameter & finish, which was brought about by a reamer. It was an imperial size next closest to the nominal metric size specified on the plans. I was already modifying the cylinder barrels as mentioned earlier, so this change was incorporated. This resulted in a slightly thicker liner wall which I thought was fine, maybe even desirable. With the cylinder ID established, I would finish the liner OD to whatever dimension was required for the correct slight interference fit. The interference amount was driven by being able to place the mated liner/cylinder assembly into an oven at moderate soak temperature so that they would release from one another based on the different thermal expansion of the two materials. I’ve done this operation many times on RC engines with my small toaster oven to replace liners.

But let me back up a step. This is my first engine & I was intimidated by making good quality piston rings. This topic has been beaten to death in many other posts, so let’s just say I found myself at the same fork in the road I suspect others have arrived. I was aware of the Trimble ring method documented in Strictly IC magazine. There are also some excellent build posts on this forum where others followed Trimble’s procedures with great results. I find Terry Mayhugh’s (Mayhugh1) build posts on the other forum to be particularly informative. Making the ring fixtures represents some work, but didn’t seem too onerous. But I didn’t have access to heat treating equipment or related experience which seemed pretty important to success. I wanted this engine to run & rings are crucial to success. So, to my thinking, there are 2 main paths:

(A) Bring all liner bores to ‘whatever’ diameter they arrive at, as long as each are identical to one another & appropriate final finish. Using that resultant measured bore as an input value, all of the dimensions to make the ring blanks & heat set fixtures can be determined using Trimble’s equations. The advantage here is that all the liners can proceed along together somewhat as a group. They receive the same tool setup treatment one after another, especially up to the latter stage of finishing where it counts most.

(B) Purchase commercial rings, assuming they can be reliably sourced. This solves the ring making issue. But you need to make the bores exactly the same as the liners they were intended to run in. The O5 is a nominal 24mm bore which happens to be the same as an OS-56 4 stroke engine. Therefore, it seemed like a good idea at the time in my case to purchase 5 rings, including spares for unforeseen replacement. So, I somewhat naively, went down this path. Although it seems like a good plan, in reality it’s actually more work & higher potential for mess up. At least for multi-cylinder engines where the count increases. The issue is the dimensional target – trying to stay within say 0.0001” bore target and simultaneously arriving at that target with the appropriate finish. If the bore is inadvertently exceeded for whatever reason, that’s kind of the end of the trail as it will no longer match the commercial ring. Next engine I will likely go the Trimble route.

When the rings arrived, I measured the cross section against the Trimble values & they were quite closely which is assuring. I also got a new OS-56 liner to closely examine for fit, finish & use as a dedicated glorified bore gage. And added a piston to obtaining corresponding dimensions like ring groove, crown & skirt OD’s etc. to replicate for my pistons. Thus, the shopping list expanded but I figured I could sell the piston & liner as spares one day & recoup some costs. As of today, several years later, they are still sitting in my box. :/

[petertha]

So, onto the liners. They start out as drops of 1.25" nominal diameter Class 40 grey cast iron bought from Speedy Metal in USA. They arrive ~1.35" OD I think so you arrive at the good stuff under the skin. Previous to the real liners I also made some testers out of 12L14 & 1144 Stressproof. The 12L14 finishes beautifully but I was a bit concerned about corrosion in methanol fuel environment. The Stressproof machined well, likely a bit stronger & probably a good choice too according to others experience. But sourcing the appropriate diameter was more difficult at the time & sadly 90% of material core ends up in the swarf bin. CI seems to have a reputable track record in conjunction with CI rings. I’m sure wear will be just fine for my occasional running. CI is relatively inexpensive & available in progressive sizes for expected mess ups, so CI it was!

I took a skim cut to get through the crust, faced the end, then pilot drilled 0.375” to 0.875. On my prior testers I experienced a bit of harmonic ringing & minor chatter which I assumed was because I turned the OD to size first & then the bore work. This time I reversed & did the boring first while there was more meat on the wall. Seems to have helped but could also be CI itself vs the prior steel alloys. I found I could hold dimensions quite well as long as one account for any heat buildup. CI is a bit messy so I cover the ways.

[petertha]

I used the thickest section boring bar with carbide insert & bored to 0.940” ID. This lands me with 0.005" left to remove for target 0.945" finish bore.

[petertha]

I rough turned the main OD slightly oversize, then did the upper liner features. The crown is an extended lip with an undercut so the liner registers onto the cylinder top deck. Then I used some homebrew sanding sticks made from 2" wide MDF boards with wet-o-dry paper bonded with 3M spray adhesive. I found this to be an expedient way to remove the turning grooves & work the material down to size in a controlled manner. The board width spans the entire liner length which helps correct & minimize undulations that can result from traversing a narrower abrasive belt strip back & forth because the dwell time & tension can be different across the length. For reference the OS-56 liner was within a tenth OD all along the surface.

[petertha]

The final 0.0010-0.0015" came off with a homebrew OD lapping tool of sorts. This was kind of an experimental venture driven by how much time it took me to achieve acceptable surfaces on my prior test liners. I hoped these generic lapping blanks could be utilized used for this job & future projects if they worked out. Commercial tools are available but they are expensive, especially in larger sizes.

The idea was to have some aluminum blanks cut with the radial pie segment slit pattern you see. Then they could be bored out to the requisite ID & either lap directly on the bored surface, or perhaps in conjunction with a sacrificial split lapping collar to get more utility out of the bore size, kind of like how a collet grips a part. I made a CAD drawing & send it to a waterjet outfit. They cut the slit profile including end holes leaving me to drill & tap for a cross screw for setting lap pressure. I tried different lapping compounds & settled on some cheapo oil based AliExpress diamond paste that comes in a syringe tube.

I guess I could say my lapping tool worked out OK in that any diameter undulations (hilltops) are worked down & the diameter can become quite consistent. But I find lapping to be messy business & best confined to removing small material thickness. It also requires a bit of hand technique like stroke & dwell time & re-charging the lapping goop & fiddling with the clamp tension. Then you have to clean everything spotless, measure at various spots down the length & go at it again. It all takes time. I left the last 0.0005" for 1000 & 1200 paper using my full width backing boards & in all honesty might be just as quick as lapping. Eventually I arrived there. It’s Interesting how CI can get a finish, eh?

[petertha]

The skirt ID has a shallow chamfer to provide clearance for the rod motion. One last check of finish dimensions, lightly knock down any corner edges, then part off the liner. Then apply a single layer of protection tape, hold reversed in collet chuck & face the lip surface to final length.

[petertha]

Onto finishing the bores. A while back I acquired a Themac tool post grinder (TPG) grinder & played around with it on my previous test liners. I now have mixed feelings about TPG’s. I suppose it suites this kind of application where relatively short length work sticks out of the headstock cantilever mode. Accurate surface dimension & finish combinations can be achieved with the right technique. If the work involves hardened materials, grinding may be the only way to go, but I don’t consider CI to fall in this class. I also had hopes of using TPG for several other applications & that’s where some shortcomings & complications become apparent, especially if the project involves tailstock support. The TPG assembly just wants to occupy the same real estate as TS assembly which really is difficult to work around. I also had a tough time finding suitable wheels, meaning the right diameter & grit & composition combination. I ended up buying a surface grinder wheel from KBC & had a water jet cutter make me a bunch of wheel blanks. I was a bit apprehensive about them blowing up but so far so good. Dumore is another popular name. I can’t speak for their wheel arbors but the Themac has an oddball taper that doesn’t match any common taper, seems to be unique to them. Ultimately, I figured it out, but suffice to say making your own arbors is more work.

Most of the time you will see a TPG lathe setup where people pre-align the lathe compound at a shallow angle & finely increment cross travel depth that way. The magic trig number for convenience is 5.739 degrees compound angle (off of spindle axis) which yields the relationship of 0.001” increment on compound dial equates to 0.0001” of cross bed travel, times 2 equals 0.0002” in bore increment. Just remember that the same rule of removing backlash applies. And this is not exact because unlike a cutting tool, the wheel is slowly eroding in diameter as grinding proceeds. So, you have to measure & re-calibrate more so than other lathe operations.

My compound leadscrew is in pretty good shape. But when I secure the compound dovetail between each pass, the clamping action itself can drift the actual position a bit. I’ve made some improvements to the lock, but it’s still there. So, I don’t completely trust this setup on my machine although I do like the fine incremental feed aspect. Alternatively, I made a fixture to hold a tenth’s reading dial gauge to bear against the end of the cross slide itself which directly measures Y infeed displacement that way. This removes both backlash & table lock issues simultaneously. A sensitive dial gage also acts as a sober indicator of machine vibration which is flowing through to the grinding wheel. The needle fluctuates on either side of the true reading. Ideally, deflection is low & somewhat dependent on where the indicator is mounted & how things are tightened down. I think the TPG is probably approaching the limits of my lathe rigidity & spindle bearing condition. If you don’t have dovetail locks, my opinion is that TPG might be the wrong weapon of choice because once the motor winds up, the sliding surfaces can become ‘buzzy’ & prone to free floating, even with a good quality TPG. A suitable DRO can measure displacement independent of dials so long as you have the right resolution. But consider even typical 0.0005” step increment on DRO display represents 0.001” of bore gone. That’s what made me a believer in a mounted analog indicator. So, after some trials I kind of considered the TPG as a ‘truing’ tool, not a finished bore tool. At least on my lathe & limited experience. That leaves the last thou or more for lapping but TPG is still a time saver. Whether this warrants the expense & setup of TPG is probably another discussion. So, who knows, it may get traded for a TIG welder one day which costs about as much & would see a lot more use in my shop, but that’s another story. What I REALLY need is a Sunnen hone LOL.

Back to liner grinding. I plugged off the back end of my collet chuck so grinding debris would not migrate in behind. I used a single wrap of tape on the liner OD for protection & gripped it in 5C collet chuck.
The wheel was dressed in-situ with a homebrew diamond tool holder. Cover the machine & use a vacuum for this operation please. Then it was a matter of selecting a low spindle RPM, power feed the TPG on low traverse setting & start opening the bore in small ~0.0005” bore increments. It is a somewhat satisfying experience to see partial geometry patches being removed under grinding, meaning non circular sections features you thought were quite true by a boring bar alone. The liner never got warm because of low DOC & more time measuring & futzing. I tried a fluid but t didn seem to be adding any value, the wheel looked clean at these removal rates. The bore finish was ‘ok’ not stunning. The geometry & roughness seemed acceptable but I think it exhibited skip or maybe secondary vibration. I’ve been told my grit selection was too fine, that actually coarser would be better. Grinding is a science in itself. But they popped off pretty quickly. I have been making 6 cylinders, 1 guinea pig & hopefully 5 keepers.

[petertha]

TPG grinding examples work in progress

[petertha]

At this point I was ready to lap the liners using a brass Acro brand lapping tool and a tenths reading bore gauge. I used my OS-56 liner as a glorified bore gage setting tool.

[petertha]

Many hours later of mind-numbing work I had 6 liners to a nice matt finish & within a tenth. And all this was basically a warm up run.

[petertha]

Now we arrive at the nasty bit. I put a cylinder in the oven at 450F for a set time & dropped in an ambient liner with no drama. But as mentioned previously, once the assembly cooled, the bore had reduced differentially. About +0.0015” near the top & ~ 0.0005” near the skirt. So not only a significant bore diameter change, but the barrel walls were no longer parallel either. I returned the assembly to the oven & repeated what I’d done a few times already on the tester; heat up & separate to evaluate what was going on. Well, this time even a light love tap after heating was not removing them. This always worked with my test cylinder.

When I carefully remeasured all the cylinders, I had more questions than answers. Maybe because the new cylinder design had more mass. Maybe the reaming was not quite as perfect as I imagined. Maybe the cylinders relaxed a little post-machining because I could measure as much as 0.0005” oval in spots. Or maybe there are micro undulations in the surfaces kind of acting like a screw thread where the mean distance between hill tops is correct, but the surface itself can act like a secondary grip once the shrink has occurred? I decided I wanted these parts separated to be utilized so had to resort to light torch heat. They finally let go. The liner came out a light tan color but amazingly neither seemed worse for wear. They had the same dimensions as when they started out.

[petertha]

At this point it was time to weigh my options. I already sunk some effort making the liner OD’s nice & consistent between one another. But the interference just seemed a bit excessive now. It was obviously shrinking the bores quite a bit. Had the liner bores remained unfinished at this stage, it would be of no real consequence, I could have just carried on. I no longer saw much merit pursuing the ability to swap in a replacement liner at some later time because I had already decided to make a complete 6th cylinder assembly spare to just bolt on the crankcase if something bad happened.

I tried mounting my test liner to an offshore expanding arbor held between centers in the lathe to see how easy it would be to somehow lap off a thou in a controlled manner. I’m not sure if I received a Monday arbor but it did not turn concentrically. I put a dial on the OD & it had about 0.002” TIR. I didn’t need more bad geometry problems so abandoned that idea. Glad I didn’t buy a complete arbor set!

So, I bought another Acro brass lap barrel to slightly enlarge the cylinder ID, thus reducing the amount of liner interference & simultaneously tuning up the surface geometry & finish. Hopefully I could use the liners as they were & reduce the amount of subsequent bore correction once shrunk again. Kind of bass-akwards to the original plan but seemed like the best option. This worked quite well. I haven’t lapped aluminum like this before with brass but it yielded a nice light matt finish & the bore gage said it was consistent down the length.