After completing the Merlin's valves I will have made a total of 92 valves, including spares, for three different model engines during the past four years. That's a lot of valves for someone who used to hate to make more than one of anything. Even though they were all made using essentially the same technique, it seems I always learn something new from each batch that helps me with the next.
I started by cutting a number of short lengths of half inch diameter 303 stainless rod so I could machine a valve on each end and waste only one spigot between them. I've used my 9x20 CNC lathe to make all my valves, and I machine them in a single batch in four steps.
The first (roughing) step is my favorite because with .020" excess stock left behind for finishing, each part can be run relatively unattended without the hand-holding required for a precision part. This step includes my screwy technique of using a greased leather pad between the end of the valve stem and the tailstock to dampen chatter. My live center is the only center I have that will give the tool holder and inserts that I use for this operation access to the end of the stem, but its run-out produces a larger taper than I can typically get with my leather pad kludge. With all the valves that I've made I probably should have created a custom center by now, but the leather actually seems to work pretty well. The blanks were all rough-machined over a couple days using a Seco DCMT21.51 MF2 TP250 insert. Each end of the blank required about ten minutes machining time, and a single edge of the insert showed negligible wear after roughing out all 28 valves.
However, I had to interrupt the build and devote some time to my lathe. During the T-18 build my 9x20 Mach3-based CNC lathe began acting up, and the spindle motor would sometimes take an inordinate amount of time to start up, or it would not start up at all. This occasionally caused the tool to crash into a non-rotating workpiece. At that time I replaced a couple relays on the break-out board that I thought might be sticking, but that didn't solve the problem. I then made a bus monitor to display the various control signals between the very expensive and poorly supported German-made Hamming motor and the Lathe's breakout board. The problem showed up only intermittently over the past year and, it seemed, only after the lathe had been powered up for a few hours. However, while turning the Merlin's valve spring retainers and clips the problem became much more frequent. According to my monitor the problem appears to be within the motor, and maybe it is heat related.
The motor inside the lathe's enclosure has a powerful fan mounted on one of its ends. Air is blown over a heat sink wrapped around the motor to which the VFD circuitry is likely internally attached. Warm air, chips, and coolant from inside the enclosure is blown through the heatsink and must occasionally be cleaned out.
In the middle of the spring retainer machining I stopped and cut a five inch vent in the rear of the enclosure and after fabricating a bezel and adapter for a large plastic plumbing elbow I had cool uncontaminated outside air flowing over the motor. This seemed to help, but while machining the valves the problem became even worse and, finishing the valves became difficult. The west coast distributer for the lathe informed me the motor is repairable only in Germany and will likely require several months, and so I've decided to order a replacement.
The next step used a Kenametal DCMT21.51UF finishing insert, and because of my lathe issues I decided to leave .002" stock instead of my normal .001" stock for later polishing. This higher precision step required more of my involvement. The lathe is typically consistent enough to hold a thousandth if I measure and readjust the work offset between each part. The tailstock wasn't used at all in this step, and I backed up the valve stem with a leather pad in my fingers. This really isn't as dumb as it probably sounds. I usually finish machine all the valves in one sitting, and after the first few parts I can usually maintain just the right amount of pressure on the workpiece to obtain final valve stem diameters that are consistent to within a half thousandth. Because the lathe now forced me to run much smaller batches there were a lot more 'first few' parts, and these diameters ended up all over the place. Typical finishing time was about four minutes.
The machining marks left by the last machining pass were polished out in a third step using 400g and 600g papers followed by metal polish. The valves' rear sealing surfaces were polished with 600g and 1000g before using the metal polish. The sealing surfaces, after close inspection, may be polished one last time later during the fourth operation when the valves are brought to their finished length and the retainer grooves are cut. The polishing time was typically eight minutes.
Before actually finishing the valve stem diameters I had to decide on a target clearance to their guides. I used .0005" +.00025"/-.0000" on both of my radials, but these engines are air cooled, use valve cages, and don't have well-defined top-end lubrication systems. A small clearance seemed appropriate for them in order to prevent un-metered air from being drawn past the intake valve stem. The Merlin's top-end will likely be well-oiled, and some of this oil will end up on the valve stem to help to seal it. If the clearance is too great, oil can be sucked into the combustion chambers during the intake stroke, and the engine will smoke.
I'm expecting the Merlin heads with their valve-cover'd heads and suspect liquid cooling system to run much hotter than the finned radial heads which sit well up in the prop wash. Since the exhaust valves will likely run much hotter, temperature expansion of the stem diameters is a concern. This, coupled with the use of individual seats and guides rather than valve cages, may result in larger temperature differentials between the stems and their guides so that binding could become an issue. As a result I chose a target clearance of .001" +.0005/-.0000" for the Merlin valve guide clearance Because of my lathe distractions, several ended up closer to .002", and they were marked with a red Sharpie so I could keep track of them during the fitting process.
Leak checking the valves in the Merlin heads is complicated by the individual seats and guides as well as the engine's giant ports. I was tempted to skip this step but finally decided it would be interesting to give it a try.
I purchased a two-part silicone molding kit from a local craft store. After shoving temporary hard rubber plugs through the valve seats and up against the valve guides to create an air space inside the silicone behind the valves, I poured the silicone into the ports. After it cured, the hard rubber plugs were (easily) removed, and I was left with silicone molded plugs that perfectly filled the ports. Just behind each valve was an open space with a volume that very closely matched the volume I would normally leak-check in a valve cage. These equivalent volumes will allow me to compare my leak down times with past results from other builds.
With the silicone mold plugs inserted into the ports, the leak-down times can be measured either of two ways. A vacuum can be drawn through the rear of the guides with the slotted test valve in place. Or, the vacuum can be pulled through a metal tube inserted through the silicone and into the air space behind the cylinder's actual valve while it is held in place. I decided to test a seat and valve pair to see just how well this second check might work.
After some measurements to verify the final valve length I realized my CAD model was accurate enough to use, and so I determined the final length through trial and error with the valve in place in the actual valve train. The lash-setting eccentrics are difficult to use and don't provide much adjustment range, and so the finished valve length needs to be fairly precise. Fortunately the valve train components seem to be consistent enough that the valves won't need to be fitted to their locations.
The total machining time per valve averaged just over 30 minutes which is probably even more than what a real machinist would need using all manual equipment.
The seat was cut to a .005" width using the manual seat cutter, lapped with TimeSaver against a dedicated steel lap, and then polished using a felt bob and a dab of metal polish. I found that greasing the cutter's pilot to reduce its clearance in the valve guide produced a better finish than I could get without using it.
To reduce the measurement to a two-handed operation the silicone plug was held in position with a pair of miniature clamps, and its integrity was verified before the measurement by pulling a vacuum against my thumb on the seat. The rear of the guide must be capped even with a valve in place. Even though any leak past the valve stem would become a part of this leak-down measurement it is not a component of a combustion chamber leak which is the leak I'm ultimately interested in.
My first leak-down test resulted in a twenty second time which seems like a reasonable goal I'll likely set for the rest of the valves. The volume being checked behind the valve is only .05 c.i. which is a small fraction of the total involved volume. The volume of the interconnecting hose between the port and the MityVac is .20 c.i. - Terry