Quarter Scale Merlin V-12

[petertha]

> so I reduced the axial groove width to .027" for use with a .026" ring

Terry, hard to tell from your pic, but are you using those Nikcole inserts to cut the ring grooves? Technique wise, do you cut the middle groove portion like parting & then dress each face with a skim cut to final gap dimension? And how about the groove bottom - do you traverse the tool laterally a bit to ensure its a clean profile? I found this operation to be a bit fiddly & was reluctant to go in there with sandpaper, any tips appreciated.

I like your clamping setup jig, going to steal that idea. Do you target the pocket bore diameter exactly equal to piston OD & then the slitting 'unsprings' it a bit? Or do you bore with a bit of oversize clearance? I'm slowly learning the hard way how easy it is to mar or distort delicate parts.

 

[mayhugh1]

> so I reduced the axial groove width to .027" for use with a .026" ring

Terry, hard to tell from your pic, but are you using those Nikcole inserts to cut the ring grooves? Technique wise, do you cut the middle groove portion like parting & then dress each face with a skim cut to final gap dimension? And how about the groove bottom - do you traverse the tool laterally a bit to ensure its a clean profile? I found this operation to be a bit fiddly & was reluctant to go in there with sandpaper, any tips appreciated.

I like your clamping setup jig, going to steal that idea. Do you target the pocket bore diameter exactly equal to piston OD & then the slitting 'unsprings' it a bit? Or do you bore with a bit of oversize clearance? I'm slowly learning the hard way how easy it is to mar or distort delicate parts.

Peter,
Yes, I take a pass down the middle and then skim the walls on either side. On a groove this narrow, there typically isn't room for much side-to-side. In this particular application the rear finish isn't important but for an o-ring it would be. And yes, I'm using Nicole inserts.
The piston clamping jig really is really something I also borrowed from somewhere. I bored the hole .002" oversize and I don't think I saw any spring either direction. I made one from 12L14 some time ago for another project and it did spring open a bit. The hole hoe the clamping bolt should also be drilled and tapped before cutting the relief grooves. - Terry

 

[geo]

Terry a it off topic toying with the idea of a tormach mill are you happy with your machine
Cheers
Geo

 

[mayhugh1]

Geo,
i've been very happy with it. If you have any specific questions, send me a pm. - Terry

 

[mayhugh1]

Although there's still lots of small PITA parts ahead, the connecting rods should be the last major engine components to be machined. The Merlin uses six pairs of fork/blade rods that include big-end split bronze bearings. A single pair of these rods is about as complex as a radial's master rod, and so I spent time thinking about how I was going to go about machining them in quantity. I decided to use 7075 aluminum since it has some 80% greater tensile strength when compared with 6061 and is 50% harder. The greater hardness should provide some additional wear resistance against the hardened wrist pins. Unlike the full-scale Merlin, the Quarter Scale doesn't use bronze bearings in the rods' small ends.

I started with the simpler blade rods so I could work out the fixturing that I'll also need later to machine the more difficult fork rods. The plan was to machine the rods in cookie sheet fashion in groups of three. I prepared enough workpieces to machine three groups of three rods so I would have spares to cover mistakes along the way.

Each workpiece was created by bolting two half-inch thick rectangular bars together so the rods and their caps could be integrally machined. The junction between the two bars would become the seam between the rods and their caps, and so the workpiece ends were finish-machined before being bolted together. The layouts of the three rods within the assembled workpieces were planned so the bolts holding the two bars together were actually the rods' 5-40 cap bolts. The heads of the bolts were sunk deep in counterbores inside the workpiece so they would not be cut up while machining the ends of the caps. Even still, extra stock had to be left on the ends of the caps and removed later in a separate operation.

I typically prefer to support workpieces, requiring both top and bottom machining, in a vise because I seem to get better results when re-registering the parts. My supply of 7075 was pretty sparse, however. The widths of the material I had on hand wouldn't have provided the margin required around the rods to stabilize the workpiece if supported in a vise during the final machining passes. As a result, I wound up with several inefficient set-ups and lots of workpiece preparation. Since I didn't have material to waste on workpiece mistakes, this preparation amounted to machining pairs of parallel surfaces with no precise dimensions. In my shop, trying to achieve both of these at the same time is pretty risky.

The first step in preparing the blanks was to parallel machine the long edges of both bars that make up each workpiece. Exact dimensions weren't important, but since the workpieces would be flipped over for top/bottom machining, parallelism was important. The top and bottom mill-rolled surfaces of the workpiece material was already as parallel as I could hope to achieve, and so those surfaces didn't receive any special attention before drilling the holes. The clearance holes and counterbores for the cap bolts were first drilled through the edges of the narrow bars. An extra hole was included for a pressed-in gage pin that was used to keep track of the locations of the bolt holes within the workpieces. Each pin was collinear with its row of holes and spaced a known distance from them.

The tapped holes for the bolts were then drilled into the finished ends of the wide workpieces that will contain the rods. After bolting the workpiece halves together, the top and bottom surfaces of the assemblies were lightly faced to give consistent parallel surfaces for later fixturing. For these facing operations, the surfaces of the narrow bars were used to indicate the assemblies in the vise to insure the final faces of the workpieces wound up parallel to the rows of bolt holes. To complete the workpieces, their left hand sides were squared up. At this point, three of the workpieces' dimensions had been changed after their holes were drilled, and all three workpieces were slightly different. Only the gage pins knew exactly where in the workpieces the bolt holes were and where the rods would eventually wind up.

The first operation was to bore the rods' big and small ends using the gage pins to determine their exact positions. After completion, the middle rod's big-end bore was used as X0-Y0 for all subsequent operations, but Z0 remained with the gage pin somewhere inside the workpiece. The boring operations were done with the workpieces supported in a vise. At this point I was still optimistic about using a vise for the rest of the machining, and so I bolted a huge chunk of steel to the underside of the workpiece for reinforcement. It did a good job of damping the vibrations of the thin workpiece straddling the parallels and probably improved the surface finish of the bores. However, it was still no cure for the lack of stabilizing stock between the rods and the rods and vise jaws.

After giving up on using a vise for the rest of the machining, I re-surfaced a sacrificial block that's been in my shop for years. The workpieces were bolted down to it for their top side machining which included engraving unique identifiers on each rod/cap pair. The diameter of the bolt heads was only .007" less than the finished thickness of the rod caps, and so it was important that they end up well centered in the rods. Watching them emerge from the workpiece without a scratch during machining gave me the confidence I needed to carry the gage pin references over to the fork rod machining later. Before flipping the workpieces over to complete the bottom side machining, the partially cut out rods were glued to the workpiece using Devcon 5-minute Super Gel. This epoxy kept the rods attached to the workpiece frame while they were being cut free during the bottom side machining. All the bottom side filleting was done while the parts were attached to the workpiece by only the epoxy. When the bottom side machining was completed, the workpieces were heated to 250F, and the parts literally fell free.

The final step was to unbolt the caps from the rods so the caps could be finish machined. To save time, three caps were machined simultaneously. A set of extra-small pattern steel washers were turned to fit under the heads of the steel SHCS's. The actual machining time on each blade rod/cap pair worked out to about two hours. About half of this time was spent on the cosmetic filleting that will never be seen after assembly. - Terry

 

[kvom]

How did you machine the rounding along the edges?

 

[goldstar31]

I have just been with a Merlin engine fitter whom I served with in 1948-50 in RAF31 ( The Goldstars) Squadron when it was at RAF Hendon.

Whilst there were 2 Squadrons 601 and 604, we had 'care' of three belonging to high ranking officers. Boothman of Schneider Trophy fame but perhaps more importantly SL-721 belonging to Air Chief Marshall Sir James Robb- a 16 which is still airworthy and in Canada.

Ironically, we were both a mere 18 years of age then-- and now 86

Ex Corporal Norman Atkinson

 

[mayhugh1]

How did you machine the rounding along the edges?

Kvom,
The filleting was done with many passes of an 1/8" dia. ball mill.
Terry

 

[nel2lar]

Terry
I've watch the progress for a long time and you still amaze me on your expertise and craftsmanship. The love of machine and man, you surely have achieved the highest.
Nothing less than beautiful.
Nelson

 

[petertha]

Terry, I find your epoxy backfilling technique very intriguing. What sort of oven temp do you bring the part up to for glue removal stage? Do you treat the aluminum in any way prior, like an oil wipe acting as a release agent? Or do you endeavor to clean the metal to encourage a good adhesive bond during machining & just rely on the heat?

 

[wirralcnc]

On the subject of epoxy can you post picture of the epoxy you use.

 

[mayhugh1]

All,
Here is an Amazon link to the epoxy:

https://www.amazon.com/dp/B002YCACJ6/?tag=skimlinks_replacement-20

If you Google 'Devcon 5 minute gel' you'll find many places that sell it. I get mine locally from Lowes which is a home building supply house. Devcon also makes a non-gel product which works about as well but with less tensile strength, but the gel is often more convenient because it won't flow into places you may not want it to go.

I thoroughly clean and dry the parts before use. The vegetable-base coolant I use for machining has to be thoroughly removed. I used to use acetone but recently switched to hot water and dish detergent which seems to do just as good of a job. Heat is the only release agent I use. The epoxy is only good to about 200F, and so an hour in an oven at 250F cleanly releases the parts. Sometimes I also just use a torch to warm up small parts for release.

I spent a lot of time several years ago trying various epoxies that were strong enough to withstand what I wanted to do with them, but would cleanly release with moderate heat. One lesson I learned was to stay away from JBWeld. Somewhere in the bottom of my scrap pile are some JBWeld epoxied test parts that I never could get to release. - Terry

 

[mayhugh1]

Forked connecting rods were commonly used in high performance aero-engines like the Merlin to help reduce length and weight of the power plant. For a given number of cylinders, the crankshaft needed only half the number of throws compared with a conventional V-engine. V-engines using conventional rods on shared crankshaft pins, however, required an offset between the cylinder banks. The Merlin's forked rods were split in two at their big ends, and the blade rods for the opposing cylinders were thinned down to fit in between the tines of the forks. This arrangement didn't require an offset between the cylinder banks.

The Merlin's forked rod has a single, wide bearing sleeve that spans the whole width of the rod including its central gap. The blade rod doesn't run directly on the crankpin, but instead runs on the outside of the sleeve. The two rods oscillate but don't rotate relative to each other. The bearing is relatively lightly loaded and doesn't run at high speed, but the resulting reciprocating motion complicates its lubrication. As in the full-scale engine, the crankshaft's pressurized oil system will provide a protective oil film between the sleeve and the crankpin, but the Quarter Scale's blade rod will depend upon splash lubrication from sump windage.

I started work on the fork rods while the fixturing that was worked out during the Quarter Scale's blade rod machining was still fresh in my mind. I had some longer pieces of 3/4" thick 7075 material, and these allowed more separation between the parts for a more rigid workpiece. They also allowed the use of larger cutters to reduce the extra machining time required by the thicker workpieces. The width of the material, though, was identical to that of the half inch material used for the blade rods, and so pretty much all the workpiece preparation and many of the machining steps wound up being similar. The big and small end bores are the most critical operations in the rod machining. They must be parallel to each other to avoid binding the rods in the cylinder bores. Identical big end bores will also ease the bearing fitting process later.

Each fork rod has two pairs of 3-48 steel cap bolts, and so there were two rows of bolt holes in the workpiece layout. The gage pins were placed between the bolt hole rows and on the centerline of the final parts. Most of the text in the previous post describing the blade rod machining also applies to the fork rods, and so the photos should be pretty much self-explanatory.

Extra care had to be taken while machining the slot between the tines of the fork rods because a 3/16" diameter end mill with a 1-1/4" stick out was required. From experience, I've learned that this is a recipe for chatter. With only a few thousandths clearance specified on either side of the blade rod, a smooth surface finish on the walls of the slot was important. The speed, feed and d.o.c. were experimentally determined on a test part before the actual machining was begun. In order to obtain an acceptable surface finish, a roughing pass down the center of the slot followed by light finish passes on each wall were required.

Devcon may have changed the recipe for their epoxy gel since the last time I purchased it. The new product I bought for the fork rods was in a different color package and behaved differently. After baking the finished workpieces to remove the finished rods, I found they still released easily so long as the workpieces were at 120F or so. But, I allowed one to cool back to room temperature, and the epoxy hardened and regained its grip on the parts. After re-warming the parts until the gel became flexible again, the rods easily released. Unfortunately (for me) Devcon may be trying to improve their product's high temperature performance.

The final machining on the ends of the rod caps was done with the caps clamped together in groups of four. A torus end mill with a .015" corner radius was used to machine the smooth contour using many fine steps.

I didn't feel comfortable with torquing the steel rod bolts down against the aluminum rod caps, and so I turned three dozen .155" diameter steel washers with a very close i.d. fit to the bolts for use under their heads. - Terry

 

[Parksy]

Superb work as always Terry!

I always learn something from your posts.

 

[Ghosty]

Hi All,
Only found this site the other day. Have been reading through this thread.
You amaze me on your expertise and craftsmanship. Will follow to the end.
Many years ago I came across a 1/5 build, done some looking and found it
http://enginehistory.org/ModelEngines/merlin_xx.shtml
[ame]https://www.youtube.com/watch?v=0xe1LL1IC7Y[/ame]

Cheers
Andrew

 

[mayhugh1]

To date, the only split sleeve bearings I've made were the main bearings done earlier for this engine. For those, I took a whimsical departure from common practice and went down an involved but fun path that included rolling out pure silver plate and forming shell bearings using my own shop-made press dies. The final results with their pressed-in oil grooves turned out better than expected, and the experience was certainly worth the time and effort put into it. But, having been there and done that, I wanted to try something different and, hopefully, much simpler for the rod bearings.

While researching piston ring fabrication techniques several years ago, I came across a method being used by a model builder who was making his own cast iron rings. His technique was to measure the diameter of his cylinder bore from which he calculated a circumference, and to this circumference he added the width of the slitting saw that would be used to cut the ring's gap. This new circumference was then divided by pi and used to calculate the o.d. of the blank from which the rings were parted. Although the builder claimed his rings fit his cylinders perfectly, it seemed to me that most of the contour change needed to even get one of these rings inside its cylinder would occur at its highest stressed point directly across from its gap. His fits were evidently good enough to get his engines started, but I doubt they were perfect. I was more impressed by G. Trimble's quantitative arguments and adopted his method for making my own rings. I felt, however, that someday I might apply this builder's technique to making split sleeve bearings. With two individual bearing halves there are no concentrated areas of stress. Only one of two saw kerfs would have to be absorbed in each split bearing's contour adjustment when inserted into its rod. Since the absorption areas are much wider, a more uniform contour adjustment is possible.

I decided to finally adapt this ring-making technique to the machining of the Quarter Scale's rod bearings. To calculate the o.d. of the bearing blank, I started with the diameter of a rod's big-end bore which I multiplied by pi to obtain its circumference. To this I added twice the kerf width of the saw that will be used to slit the blank plus an additional thousandth for 'crush'.

In order to establish the bearing's thickness, I performed a similar calculation for the blank's i.d.. To the diameter of the rod journal I added .003" for a running clearance before calculating a circumference. To this circumference I added twice the width of the saw kerf and then divided the result by pi to obtain the i.d. of the bearing blank. The running clearance would have been closer to .0015" if my rod journals had ended up perfectly circular. Despite my best efforts during the crankshaft machining to prevent the Merlin's ten inch long 1144 alloy crankshaft from flexing while the journals were being turned, they wound up with circularity errors. On such a long crankshaft, grinding the journals would probably have given a better result.

The end of the rod blank was turned/bored for only one bearing at a time since I had decided to machine a custom size bearing for each journal. After machining its end to the calculated inner and outer diameters, the blank was moved to the mill and carefully slit on its exact center using a .010" slitting saw. In order to accurately locate the rather unwieldy thin bade, I marked the top of the bearing with a Sharpie and then carefully lowered the backwards spinning saw blade until it just rubbed the mark. The saw was then lowered half the diameter of the blank plus half the thickness of the blade before reversing the spindle and splitting the bearing.

After engraving the end of the bearing with an identifying number on each side of the slit, the blank was moved back to the lathe. A close-fitting Delrin rod, supported in a rotating tailstock chuck, was inserted into the end of the blank. The slit bearing was clamped around the rod using a simple shop-made Delrin clamp so everything would be nicely held together during the parting operation. After parting, but before removing the clamp, both ends of the bearing were manually chamfered with a large 90 degree countersink so the inner edges of the finished bearing would clear the journal's tiny inside corner fillets.

SAE 660, also known as 932 bronze, is probably the 'gold standard' for tin-leaded bearing bronzes, and it was originally specified for the Quarter Scale's rod bearings. This spec was later changed to a somewhat less common 936 bearing bronze which is about 10% softer than 932. I suspect this was done to reduce wear on the blade rods which were specified to be machined from 6061 aluminum. Since I used a harder 7075 alloy for my rods, this probably wouldn't have been an issue. However, I didn't have either alloy on hand in a large enough diameter, and so I ordered a $hort cored length of one inch diameter 936. I'd have wasted a lot less material with a 3/4" diameter workpiece, but no one seemed to have that diameter in stock when I placed my order.

Before launching the entire lot of bearings, I measured and recorded the diameters of all the rod journals as well as the bores of all the fork and blade rods. Using these measurements I pre-selected a best-fit rod pair for each journal and then machined a single trial bearing in order to test out the theory.

When placed side-by-side, the finished trial bearing halves don't appear to form a perfectly round circle. But, just like the commercial automotive bearings I've used, they literally snap into place inside each rod half. Under a microscope I checked the fit of the bearing's contours to the bores of the fork rod and cap before torquing the two together, and both sides appeared to match perfectly. The fact that the bearing halves are retained by the rod halves turned out to be a real convenience while installing the rods within the tight space inside the crankcase. The gap between the installed shell halves closed up to zero when the cap bolts were tigtened; and, as far as I could tell, the crush height came out as expected. I measured the inner and outer diameters of the bearing while installed in its fork rod and both agreed closely with the calculations.

When finally comfortable with the test bearing, I spent the next few days working as a Rolls fitter but without the skill and efficiency of those who performed the same task 75 years ago. After machining and checking the fit of each bearing in its rod pair while installed on the crankshaft inside the crankcase, I recorded the i.d. numbers and the orientations of all the components along with their journal location. Since the first few sets of parts involved installing and removing the rod bolts several times, I temporarily switched to a shorter set of bolts so I wouldn't add unnecessary wear the deeper threads that will be filled later by the actual rod bolts. I chose to not leave the rods in place after performing these preliminary fits. When the wrist pins are completed, I'll recheck the fits with the pistons installed and running temporarily in a pair of opposing cylinders. Checking these pairs systematically, one at a time, will make it much easier to locate any binds that could be caused by crank machining errors. Once the rods and bearings pass this test they can be finally installed on the crank. - Terry

 

[xpylonracer]

Great machining work and equally good write-up Terry.

xpylonracer

 

[dsage]

Hi Terry:
All I can say is I'm glad you chose the Trimble method to make your rings back when you made them. I've found rings to work perfectly using his method.
But I'm concerned about using this new method described to make circular elements including bearings.
I also read the article you mention about making the circumference (essentially the diameter) of the rings greater by the width of the slitting saw and think it is seriously flawed. Of course I may have misunderstood his writings.
I have perhaps an odd way of making sense of things in my mind by extending theories to crazy proportions to see if the idea still makes sense. So in this case lets assume (for rings) you wanted to use a slitting saw with a width of 1/8" Why would you add 1/8" to the circumference of the ring circle? The resulting diameter of the ring would obviously be WAY too large to fit the cylinder. Even if you could insert it in the cylinder it would obviously no longer be round.
I won't elaborate any further on my concerns because perhaps I'm misunderstanding the whole process.
You have obviously measured your results - especially the inside diameter - with great accuracy. I trust your superior workmanship so they must be correct. It just doesn't make any sense to me.
Question: Does the bearing half slide in from the side of the rod without effort (rather than snapping it in from the top - which forces it to compress out of round)? Would it make any sense to insert a precision ground rod in the bearing with some bearing blue on it to observe the contact points? (or assemble one rod in the engine with the bearing blue). I'm going to guess that the bearings are pinched at the parting line.
I'm just trying to understand this process better. Sorry for the distraction from your thread.

Very nice work - as usual.
Thanks

Sage

 

[kvom]

WRT Sage's question, I believe that the bearing deforms slightly when pressed into the rod end, and the extra circumference means that both halves meet with no gap from the kerf.

Obviously the rod end diameter was made larger than the crank journal by the thickness of the bearing material. Whether the bearing thickness is derived from the two measurements would be a question for Terry. Since he appears to have made each bearing individually, I'm guessing each was made to fit.

 

[mayhugh1]

Dave,
I often use the same technique of exaggerating an element of an explanation in order to help understand what's going on. In this case, increasing the kerf width shows that the kerf is, indeed, an error; and this technique for making bearings is a compromise solution. I really should have given more information in my post about why I used such a thin saw blade, and so I'll do that here.

The hope was that the kerf error would be distributed around a good portion of the circumference of each half shell and become negligible or at least comparable to my other machining errors. The actual diametrical error that I'm trying to distribute is not the kerf width but the kerf width divided by pi. When I machined the big-end bores on my rods I chose to interpolate them on my Tormach instead of using a boring bar. The backlash on my particular machine is sufficient to yield circularity errors of two to three tenths which I considered acceptable since the fork rods are not intended to spin on the bearings and so any circularity error will only help hold them in place. The blade rods, though, are spec'd for close sliding fits on the bearings, but they rotate only through an angle of less than 90 degrees. With selection, I was easily able to achieve smooth sliding fits of all the rods throughout this angle, and the selections were recorded so they could be reproduced at final assembly.

This brings me to the rod journals. As mentioned earlier, I started out having to deal with half thousandth circularity errors on some of the journals themselves due to the method used to machine the crankshaft. The measured i.d.'s of my compromise installed rod bearings ended up with their own circularity errors ranging from one to three tenths, and so my solution was to machine a custom bearing for each journal with as good of a fit as I could achieve. My worst-case clearances ended up at about three thousandths which was more than I liked, but after all this is a model engine. My back-up plan was to re-bore the bearing i.d.'s with a boring bar while the bearings were installed in the fork rod. For many applications this would have been a first order solution for making bearings. In this case, it didn't seem to be worth the effort or risk. I doubt my rod journals would have known the difference. To finally answer Dave's question - no, the bearing halves won't slide into the sides of the rods but are most easily pushed into place from the end of the rod. While free standing they have the right circumference but wrong diameter for the rod bore. - Terry