In the process of gathering up all my bagged parts related to valve installation I ran across a note I had written to remind me about a cosmetic chamfer I had decided to add to the tops of the spring caps. So, I spent a couple hours finishing up some parts that I thought were already done. If you think this was an extravagant waste of time wait till you read below about what I did next.
Before starting the valve installation, though, it might be useful to review the parts involved since their fabrication has been spread over several weeks. First, the valves are completely finished. The heads and seats were turned concentrically with the stems, and the seat faces have been brilliantly polished. The valve cages, though, are only semi-finished. The seats have been cut to a width of just under .005" using a manual piloted seat cutter which insured they are concentric with the integral guides. However, the machining marks left in the seats by the seat cutter have not yet been addressed. I have two seat cutters - one shop made and one recently purchased - and they both leave machining marks that ultimately limit sealing. The one I've been using most recently is a ten flute 45 degree muzzle cutter from Brownells. To the naked eye it produces a nice surface finish, but under magnification tiny circumferential grooves can be seen in the seat's surface similar to the ones left by my shop-made cutter. The prominence of these grooves is important and at the end of the day their depths can't seem to be limited by manipulating the cutter. With either cutter used individually, I get about the same results. The leak-down times they produce are essentially identical and remarkably consistent across a large number of seats.
What I learned through some testing, however, is that if the seats are cut using both cutters - one alternated with the other - I can always get a significantly longer (up to 2X) leak-down time compared with what I can get with either cutter used individually. I eventually discovered that with the pair of cutters I could often reach 20 seconds, and with a time this long lapping really isn't necessary. It appears each cutter is knocking down the machining marks left by the other, and the net result is a better over-all surface finish. But, I've jumped ahead.
The cages have already been installed in the heads with high temperature Loctite and are free of any distortion due to installation. Just before installation each seat was cut with the Brownells cutter and then leak-checked using a single H-9 test valve. The average leak-down time was typically 10 seconds +/- 3 seconds for a vacuum leak of 25 Hgin to 15 Hgin. The total involved volume for this measurement including the vacuum gage and connecting hose was approximately .25 cubic inches.
This leak-down time is probably only meaningful to me because it became an empirically derived target while building my previous two multi-cylinder engines. It might be more interesting to others, though, if it is converted into something a little more familiar. With just a simple units conversion this 10 second leak-down can be shown to be equal to a 5 psi pressure drop over 10 secs or, equivalently, a 0.5 psi drop per second. Through sheer coincidence the volume of my combustion chamber at TDC is also approximately .25 cubic inch. So, without having to correct for a volume difference the effect of this leak on, say, a compression measurement can be roughly estimated for my shop-made peak-reading compression gauge. This gauge has negligible internal volume compared with that of my combustion chamber. Assuming a compression ratio of something over five, the expected pressure in the combustion chamber during a compression test will be about 5X atmospheric or about 75 psi. Scaling my measured pressure drop created by a single leaking valve to the increased internal pressure during a compression measurement gives, roughly, a 5X or 2.5 psi drop per second. With a cranking speed of 120 rpm or 1/2 second per 4-stroke cycle, a cylinder with this single leaking valve will have about 1/2 second (yes, I'm making some outrageous simplifications here) to lose roughly 2.5 psi or about 3% of its actual value.
The effect of this leak on combustion pressure during worst-case running can also be roughly estimated. Assuming the maximum combustion pressure is about 5 times the compression pressure (rough rule of thumb I found on the web) the pressure loss due to this single valve during running will scale up another factor of 5 to some 12.5 psi drop per second. With a worst-case minimum running speed of, say, 600 rpm or .2 sec per 4-stroke cycle, the cylinder will lose roughly 2.5 psi of its 375 peak pressure (simplification alert again) for a loss less than 1%.
So, it seems perfectly reasonable to leave things as they are. That is, I can with good conscience install the valves and let the pressure pulses from combustion beat the final machining marks on the valve seats into submission even though with this level of sealing the leaks really aren't significant. But, I like to experiment, and I'm interested in learning more about what's involved in getting to the next level in valve sealing. I have plenty of parts to play with, and I've never been one to stop at a point of diminishing returns.
So, the goal in front of me was to remove the machining marks from the valve seats in order to obtain the best possible seal with the valves. In previous builds I've learned that lapping valves to their seats immediately after cutting them is not always a good idea because if the machining marks left on the seats are prominent, lapping will just transfer them over onto the valves and score their beautifully polished surfaces. I've also learned that the grit of the lapping compound needs to be scaled for the narrow seats typically found in a model engine. For my own lapping I now use nothing coarser than metal polish. In a model engine the effect of using typical automotive valve lapping compounds can be inconsistent, and even disastrous.
I made three double-ended metal laps for lapping my seats while I was making the valves for this engine. My intentions were to use them instead of my valves to do any needed lapping. But, I began having second thoughts about them wearing too quickly even with metal polish as a lapping compound. The hardness and narrow seat widths of my new phosphor bronze valve cages were my concern. When I made my H-9 valves I inadvertently used a softer bearing bronze for the cages and was forced to cut the seats to a width of .015" - .020" in order to clean up distortion created by pressing them into the heads. The two double-ended laps that I made for that project lasted long enough to lap all the seats in that engine plus a few spares. With the much narrower and harder seats in these cages, though, I'm afraid the laps will wear out before I finish the whole lot of 48 valves.
The first alternative I came up with was a new piloted lap turned from a wood dowel rod. Two photos show the lap and the seat in a spare valve cage after one minute of lapping with Honda (motorcycle) metal polish. The grooves in this particular seat were almost entirely polished out as evidenced by its initial 10 second leak-down time increasing to over a minute. One minute is actually close to the noise floor of my measurement set-up. Using the same lap on a second valve cage, though, improved the leak down time to only 20 seconds and then on a third there was no improvement and, in fact, minor damage to the seat. Examination of the lap under magnification showed that even with super fine metal polish the narrow seat had eroded away the surface of the lap and destroyed its profile. It would be possible to make a separate lap for each valve - something not unreasonable for a smaller engine - but I wanted to see if I could avoid making another 50 somethings for this engine. I have a note of caution to anyone who might want to try a wood lap. I found it only useful with metal polish as a lapping compound. Some testing I did showed that even fine TimeSaver lapping compound is much too aggressive for wood and can destroy the lap during its first time use.
As a second alternative, I made some protective covers for my metal laps. I tried several fabrics but eventually settled on strips of thin suede leather purchased from a craft store. I punched a hole in the center of the strip through which I inserted the stem of my lap. With the thin leather patch coated with metal polish and sandwiched between the lap and the seat I was again able to increase the leak-down time on a few test valves to 20-30 seconds after one minute of lapping. These leather patches need to be replaced after lapping four or five valves, but they are very simple to make. The thin suede stretches nicely over the end of the lapping tool so there is no puckering in the seat area to give an inconsistent contact patch. One of the photos shows some of the patches I used. The combination of the leather patch and metal lap is a bit awkward to deal with after the valves have been installed in the heads, but it works really well on cages that have not yet been installed. After only several seconds of lapping a bronze colored ring shows up on the patch to indicate where to add more polish.
And so, after several days of experimenting with all my spare parts and nearly a third of my finished heads, I finally settled on a process I was happy with: 1) re-cut the seats using my two manual piloted seat cutters and alternate between them for several light (to avoid unnecessary widening of the seat) cycles until reaching a 15-20 second leak-down time, and then 2) lap the seat for one minute using metal polish on a thin suede patch sandwiched between the seat and a metal lap to obtain 20-30 seconds, and then 3) lap the valve against the seat using metal polish to extend the leak-down up to 60 seconds. (Skipping step two would never be noticed in an engine's actual use.) One of the photos shows a silicone cap that I used to grip the stems of the valves in their installed cages during lapping. (Harbor Freight used to sell a nice assortment of these caps in various sizes, and I've used the one I bought to solve lots of different problems around the shop.) The ultimate goal of my process became a 60 second leak-down with mInimum damage to the valve. All but two of my seats ended up less than .007" wide, and the histogram shows the distribution of leak-down times I actually achieved. This histogram includes several valves I went back to and re-worked after learning about the dual cutter technique. I wasn't able to reach my ridiculous goal on most of my valves, frankly, because after some 60 hours I was getting weary of the project. But I kept working until every valve reached at least 20 seconds just to prove to myself I could. On about the last dozen valves the dual cutters produced 20-30 second leak-down times on their own before any lapping. I think this was because I was developing a feel for using the cutters. Even though I could have skipped their lapping steps, I chose not to because I wanted to add more valves on the right side of the histogram.
I'm still curious as to why it was so difficult to reach one minute leak-downs on some valves but not others. Any further improvements and learning will have to be left to a future project where not so many parts are involved. Again, what turned out out to be a 40 hour diversion created by a ridiculous target was totally unnecessary and done only for the sake of learning.
I ran into a few frustrating issues involving my own poor quality control that ended up taking a lot of time to sort out. For example, I had evidently damaged a couple seats while drilling and de-burring the intersection of the valve cages with the cross-drilled intake/exhaust ports. Some tiny nicks located on a difficult to see area of the installed seats were only visible only under 10X magnification. One of the seat photos is actually of one of these seats. They were probably caused by slips of the X-Acto knife I had used for de-burring. Until I deepened the seats beyond those nicks I was not able to get leak-down times better than 5 seconds no matter how much lapping I did.
On several heads I found I had left burr remnants on the top of the valve guide bore while machining the spring cavity. My process for machining the heads included manually running a reamer in the guide bore to remove these, but I had evidently skipped this step for one of my batches. Even with only .0005" valve stem clearance this burr tilted the valve enough on its seat to limit the leak-down time to 15 seconds regardless of lapping. By the time I had discovered the source of the problem in the first head that had the problem I had leather-lapped the seat so many times that when the burr was finally removed I had absolutely no measurable leakage. After this experience every valve guide was re-reamed before starting the valve seating work.
After completing four heads and then reviewing the notes I made while sealing my H-9 valves I found a note I had made to remind me in the future to clean the seat surfaces with alcohol before making a leak test. I had found the metal polish could leave some residue on the seat that wasn't being buffed away by the Q-tips I was using, and this residue was sometimes preventing an ultimate seal. I verified the benefit of the alcohol on one head that was giving me inconsistent measurement results. When I found it stabilized my measurements on this particular head to 2X the best time previously measured on it, I added an alcohol swabbing step to my t-18 valve process.
I did a lot of other experiments - several of them destructive, but too numerous to include here - on the many extra parts that I had made. The really important things I learned while seating these 48 valves and those things that I'll apply to my next engine are:
1) Cut the seats using two different piloted seat cutters so that one can mitigate the machining marks left behind by the other. Alternate using them for 3 or 4 cycles until, with very light pressure, no further cutting action is felt,
2) Keep the seat width to .005" - .007",
3) If the concentricity's were well maintained during the original machining a leak-down time of 20 seconds or more should result from using the dual cutters with no lapping needed,
3) Blowing into a closed valve using one's lung power can only tell you that you have a leak so big that it should be visually very obvious,
4) Don't use lapping compound coarser than metal polish on a model engine,
4) If, during final assembly, the valve stems are lubed with a drop of light sewing machine oil don't get too excited when you come back later to re-check the leak-down times on the completed heads. You'll likely measure infinitely long leak-down times on all of them. What has really happened is that the oil has flowed down the stems and onto the seats where it has been wicked into what's left of the machining grooves leaving the valves with a false appearance that they are perfectly sealed. Don't ask me how I know this.
I don't plan to make much progress during the next week. Some of my grand kids are coming to spend a week with us and they aren't much interested in shop stuff. - Terry
But maybe this solves one problem but creates others?
