The Knucklehead's valves were machined from 303 stainless. Compared with those in other multi-cylinder engines I've built, these valves are huge at .625" diameter. Earlier when I machined the valve cages, though, I reduced them to .570" so the edges of the combustion chambers would have a bit more safety stock.
When I was making valves by the dozens for my radials, I learned to machine them in one large lot that I shepherded through six machining operations. Before jumping into production, however, I completed at least one valve that I trial fitted into all the cylinders so I could check for machining inconsistencies. There's often a number of machining operations that can affect the requirement for a valve's length, and some of these can be hard to control. Although it may sound risky to put off lapping and testing until after all the valves are machined, I probably saved a lot of time and avoided errors by not changing and recalibrating the set-ups for each and every valve.
The first operation is to create the starting workpieces. For the Knucklehead this involved sawing a 5/8" diameter 303 rod into short lengths so a complete valve could be machined on either end of a central one inch work-holding spigot.
The second operation involves roughing out the valves on both ends of all the workpieces. This is my favorite step because I can turn my little Wabeco CNC lathe loose with very little handholding. I typically use a CCMT 21.51 roughing insert or sometimes even a dull finishing insert that's been retired from service. Surface finish isn't yet important because this step leaves a diametrical excess of .020" stock. A tailstock isn't bothered with since stem taper will be dealt with later. However, before starting the operation the ends of the workpieces are center drilled to prepare for the next step.
The next step is the finishing operation, and it uses a very sharp DCMT 21.51 insert typically intended for aluminum. Three passes are used to remove the remaining stock except for the last 1 to 1-1/2 thousandths. Either the tailstock is used for this operation, or I will back up the free end of the workpiece with a piece of leather held in my right hand. (I realize this sounds wonky, but after a little practice it actually works pretty well.) The speed and feed are adjusted for the best possible surface finish. During this step my lathe requires some hand-holding. Upon completion of each valve, its stem diameter is measured, and the lathe's work offset is corrected as needed before taking on the next valve.
The valve stems are brought to their finished length in a fourth (lathe) operation. The workpiece is inserted through the rear end of a 5C collet so the valve to be trimmed can be gripped close to the end of its stem. The groove for the retainer clip is also cut in this step.
In the fifth step the workpiece is re-chucked on its spigot so the last thousandth or so can be polished from the valve stems. I start with 400 grit paper if I have more than a thousandth to remove. The last thousandth is removed with 600 grit paper, and the stem is mic'd along its length during the process. When finished, a dab of white buffing compound on a rolled-up paper towel moved along the entire spinning valve brilliantly colors it.
In the last operation, the two valves are sawed free of the spigot which is gripped in a vise attached to my bandsaw. With some care, the larger diameter spigot prevents the polished valves from being marred. The stem of each freed valve is then gripped in a 5C collet up close to the valve head so its face can be finished in the lathe. Again, light passes with a sharp finishing insert produce a nice surface finish without deforming the completed valve.
Before starting my little production run, I used a test valve to search for an optimum stem length for both the intake and exhaust valves in both heads. The length I was looking for was one that would rotate the pushrod rockers perpendicular to their pushrods with the rocker arms resting on their closed valves. Somewhere between the valve and pushrod drilling operations, though, one of my compound angled setups must have been off. As a result, two valve lengths will be needed - one for the valves in the left side of the rocker boxes and another for the valves in the right side of the boxes.
The need for two valve lengths created another issue. Measurements using my previously finished springs showed the pushrod forces now approaching four pounds in the valve trains with the short valves. I had already been uneasy about these relatively high forces when all the valves were the same length and longer. As a result, I decided to make a new set of valve springs.
I machined the two different length valves in a mini production run. I also machined a lap which was nothing more than a completed and polished valve that wasn't parted off from its spigot.
Unlike what I've done on full-size engines, I like to finish model engine valve seats with a separate lap rather than lapping them to their mating valves. After all, the valves come off the lathe with correct and concentric geometry, and they're brilliantly polished as well. The seats also have the correct geometry, and they're concentric with their guides thanks to the piloted seat cutter used to cut them. Their sealing surfaces, though, have marks left on them by the seat cutter that, at the scale of a model engine, can create leaks.
Over time I've 'honed' a technique that helps to minimize the marks left by the piloted seat cutters I use. I coat their teeth with cutting oil and their pilot with motor oil. The oil reduces the chatter that can sometimes be felt while turning the cutter. I'm also careful to apply only a minimum force to the cutter. Pushing too hard can gouge the seat requiring it to be made wider than necessary. I usually orient the head so the cutter is vertical and allow its own weight to do the cutting. I try to keep the final seat width on the order of .005" in order to minimize the amount of material that will have to be removed to clean up the marks.
Lapping can damage a perfectly good valve especially if the marks on the seat are prominent. I feel these marks are best removed using a sacrificial lap and a very fine lapping compound such as Timesaver. The whole process takes less than ten minutes including the time used for leak testing along the way. I perform my leak-checks by pulling a vacuum in the port behind the closed valve under test using a Mity-Vac. My own subjective goal is to achieve a 25 to 15 inHg leak-down time of ten seconds or more. This is probably overkill but it's not difficult to achieve when the seats have the proper starting geometry. This check can also take leaky valves off the table as a potential problem when trying to start an engine for the first time. If valve cages are used, their geometry can be verified with a very light pass of the seat cutter before the cage is installed in the head.
After leak-testing the valves, I turned my attention back to the valve springs. I wound a new set of springs on the mandrel used for the first set, but this time I used .026" diameter spring wire. I ended up with approximately the same i.d., o.d., and length but with one less turn. The finished springs had a measured spring rate of 5 lb/inch which reduced the worst-case pushrod loads to a bit more reasonable 2.3 lbs.
This finally completes the work on the heads. I'll likely move on to the cylinders next as I work my way down from the top of the engine. - Terry
