Valve recession
There is a situation that is paradoxically the most significant to vintage Honda owners and yet at the same time the least appreciated and understood. Namely, the fact that Honda put very soft valves in their motorcycles from the late 1960s through the mid 1980s. These valves are in fact plated (actually plasma-coated) for wear purposes. Honda’s use of soft valves is very significant. Because of it, all of these engines suffer low cylinder compression until the valves are replaced.

Honda published a dealer bulletin, Service Letter #84, that warned that these valves cannot be refaced on the traditional automotive valve refacing wheel because the risk of going through the hardness-providing Stellite coating is very real. At factory Honda tech school we cut a valve in half lengthwise, then dipped it in warm battery acid, and it was easy to see (due to dissimilar oxidation) the two metals. And just like the drawing in the Honda bulletin, we could see how thick the coating was. It is on the tip, and it is on the face (sealing surface) of the valve. And it is about 0.020" thick, that is, roughly half a millimeter. Too thick to be bothered by lapping, if that is your thing, but also too thin to risk removal when the valve is being refaced, which of course is the subject of the bulletin.

Though speaking only very briefly about the coating and warning against refacing, this official notice let the proverbial cat out of the bag. It’s plain that Honda's supplier changed their manufacturing method and began making softer valves. This proved to be the case in the practical world. Career Honda and Kawasaki mechanics can tell you stories about this. We called it, "Ninja syndrome", because it affected the ZX600 even more than most Hondas. But by 15,000 miles vintage Hondas also suffer significant compression loss due to a ridge worn onto the valve face, technically known as recession. Recession is actually a normal occurrence. But rapid recession quickly reduces cylinder compression and increases valve heat as the valve sinks into its seat, reducing clearance. And strangely, the Stellite coating doesn’t help much. The valve face recedes incredibly fast and the valve's tip gets pocketed very consistently on rocker arm engines.

All of Honda’s 1960s and 1970s production four-stroke engines and a smattering of their 80s bikes are included in this problem, from the tiny 50s and 90s through the GL1200 Gold Wing. However, it happens that this extreme recession due to softness is slower on the DOHC 450 because of another little-known Honda thing: Honda put soft valve seats in at least two engines, the GL1000/1100/1200 and the 450/500T. Having soft valve seats is an inexplicable departure from normal for Honda, who has traditionally used very hard valve seats, so hard in fact that the Titanium valves in their V4 Interceptor-based factory roadracers wore out precipitously. The 450’s softer than normal seats wear along with the valve, which results in slower valve recession. But they still recede. Whenever possible I use stainless steel replacement valves in vintage Honda engine rebuilds.

The valve job
After repeated closing against its “seat”, wear appears on the valve as a ridge circumscribing its face. This is called valve recession, and it gradually impairs the valve’s sealing. The valve seat wears too of course, by widening and cratering. And valve guides wear and occasionally must be replaced. But most of Honda cylinder head rehabilitation focuses on the valve seats.

The seat has three angles. The sealing surface, where the valve will reside, is a 45 degree angle. This is the working part of the seat. Valve seats also exhibit two additional, adjacent angles: a 60 degree surface transitioning from port to seat, and a 30 degree one transitioning from seat to combustion chamber. These are sort of oversize chamfers that ease airflow across the valve seat. So critical are these angles that the Superflow corporation, makers of the most widely-used cylinder head airflow measuring equipment in the industry, holds that the valve seat has the most influence on cylinder head flow performance.

The first step in a valve job is to lightly machine the seat’s 45 degree surface to square it up relative to the valve guide and remove pitting. Then Prussian Blue is carefully smeared on the seat, and the valve pressed against it then removed to “register” where on the valve’s face the seat is contacting it. This contact must be at the correct height on the valve’s face. If it is off-center, then the seat’s supplementary 30 and 60 degree angles are machined to move the contact where it belongs. If for example the contact is off-center toward the combustion chamber, then the 30 degree angle is machined slightly to move the contact point in the other direction. If on the other hand the contact zone is off-center in the opposite direction—toward the valve guide—then the seat’s 60 degree angle is machined a bit to move the contact appropriately. Once the seat is centered, there’s still more. Now that 45 degree contact surface needs to be smoothed, and widened or narrowed as needed for good sealing and long life.

So a valve job, though not difficult, is also not simple. There are many steps, some juggling of geometry, and some fairly specialized tools. A long time ago there was a shortcut technique in the car world called interference valve angles. It was a hack, that is, an ill-informed and low-quality shortcut. Specifically, the valve seat was made 46 degrees, one degree different from the valve’s 45 degrees. This wedge-like geometry produced at least the potential of a seal, but without the painstaking adjustments of angles and widths just described. However, the fallout of interference valve angles was rapid valve recession because on starting the engine the valve immediately began sinking into the head. It was supported on only a knife edge of a seat, remember. But since car engines had hydraulic lifters, despite this abrupt recession the valve train happily self-adjusted. And it worked. Kind of. Not really, though; you could not trust either the sealing ability or the long-term durability of the interference valve angle hack. In fact, on solid lifter motorcycle engines, things went to hell very quickly. When interference valve angles were employed on these engines, rapid recession wasn’t the invisible compromise it was on cars. It very quickly resulted in zero tappet clearance and burnt valves. Harley-Davidson warned in its official iron Sportster manual that readjustment of valve clearances had to be undertaken very shortly after the bike was put back into service after this kind of a “valve job”.

Then there is valve lapping. Valve lapping had been practiced before this, but it went prime time as a way to make interference angles more reasonable, and of course it is advocated by most manufacturers today completely apart from inteference angles, but with the same sad ethic behind it: making up for poor work. That is, lapping began as a way to mitigate a hack, is itself another hack, and it continues today as a hack because now manufacturers advocate it to compensate for skill shortcomings of a different but still related sort, improper valve seat machining or even no machining at all. Valve lapping is not part of a valve job or a substitute for a valve job; professional powersports engine rebuilders test their work using a vacuum pump and no lapped valve will pass this test. Sure, bike owners just want to improve valve sealing and not invest in expensive tools. But where lapping is concerned that’s wishful thinking. The whole reason for valve work in the first place is valve recession. When a valve is lapped, what do you think happens to that recession? It increases, of course; it gets worse. Thus absolutely nothing is solved. The ridge in the valve face is not decreased and sealing is not improved or is improved incrementally if you’re lucky.

Granted, valve seat machining tooling is relatively expense and not likely to be found in the DIY toolbox. It is best to find a machine shop you can trust. But it is not completely out of the realm of DIY providing you can and want to invest in the tools. Very simple hand tools have been endorsed by and at times retailed by the Big Five. The Neway valve seat cutter is probably the best-known. This carbide rasp cutter costs about $100 per angle, resulting in about a $300 investment per valve size, so $600 for both intake and exhaust in cases where the same cutter can’t be adjusted to both valve sizes, which in some cases it can. Add to that the necessary pilot(s) and driver and you’ve spent a between $600 and $700. The next step (and a significant one) up from the Neway is the current Honda diamond stone, at a little more cost—about $150 per angle—with the total including pilot/driver combination coming in between $700 and $1000, again depending on whether one cutter will service both intake and exhaust. Honda’s then is the best choice at the low end of the market. Unfortunately Honda has discontinued their system and has already run out of certain stone sizes, though China is now making their own diamond stones at a mere $15-20 per stone, so this technology is still available. The next step up in DIY valve seat tool expense is the motorized abrasive stone of the Sioux and Black and Decker variety. However, despite tradition, these tools are hard to use accurately and are actually inferior to the hand-turned diamond stone. Ultimately, the very best in the home mechanic range of hand-operated valve seat machining tools is the Serdi and its look-alikes.

Professional seat machining tools
The weakest part of any valve seat machining method is the valve guide. The guide is typically relied upon to provide two important things: location and loading. The location part is indispensable; you cannot hope to have machining that is concentric to the seat without starting at the guide. Loading however presents an inescapable problem. Valve guides have between 0.001 and 0.003 inch clearance in them; the valve during engine operation naturally self-centers in this clearance. However, a seat machining tool does not self-center, instead shifting significantly with cutting load and thereby resulting in a cut that is not concentric to the valve and thus a machined seat that isn’t either and will not seal optimally. It was probably the German Ludwig Hunger who came up with the solution. Hunger devised a valve seat cutter that like all cutting systems located on the guide but loaded on the cylinder head casting, by being attached to the head’s gasket surface. The French company Serdi refined this a bit and now is famous for it (and Hunger not!). Head loading valve seat machining tools are at the high end of the available solutions and provide the very best end result.

Italian company PEG may have been working on a solution at about the same time and approached the loading problem in a completely different way, by making their machine cut the seat’s profile and not its plane. That is, the high speed stone slowly traverses the valve seat instead of plopping down onto it, thus eliminating the possibility of a bad seat influencing the machining operation. American manufacturer Hall-Toledo also went this way in their VIP model grinder. Both are “orbital” designs. However, both companies put too much faith in the valve guide in an unexpected way by locating the grinder on a fixed (“dead”) tapered pilot. So that despite the laudable and innovative attempt, the machine falls short of the potential of modern, live pilot and head-loading systems.

Though superior to every other system, the advantage of the head-loading system is in many cases very small in actual practice. The Honda diamond stone is an excellent all-around choice, offering as it does not only an accurate seat with a good finish at an approachable price but also a live pilot design wherein the pilot is not stationary but rotates with the stone, automatically improving tool centering as it is merely acting in the same way as the valve does.

At least when the valve seat is flat and undisturbed. And therein is the issue. The head-loading choice in tooling becomes virtually mandatory in the case of certain “problem” cylinder heads, and none more than those in which the seats have become tilted, that is, shifted slightly in the head casting. Honda’s hottest-running vintage engines in fact have this problem, with the CBX1000 perhaps the most extreme example, and the similarly engineered first-generation DOHC fours (750, 900, 1000 and 1100) following close behind. The valve seats in these heads classically present shifted out of plane. In the CBX head the seats exhibit varying levels of shifting out of square with the head beginning at the outer combustion chambers and getting worse in the middle two, and with the exhaust seats being twice as bad as the intakes. When doing a valve job on these heads therefore, a valve seat machining system that, like most, loads on the guide, will not accurately cut these seats but only perpetuate their defect. The cutter simply deflects in the guide until it conforms to the bad seat, resulting in a newly-machined tilted seat that is still tilted, and thus not the best work. Without tools such as the Serdi and its copies the Hunger, Mira, Manek and FM that through head loading and live pilot completely eliminate the influence on the machining operation that both the clearance in the guide and a tilted seat will have, problem heads can be extremely difficult to do good valve seat work on.

While there are many ways to do a valve job, with methods ranging from archaic 1940s tech to state of the art computerization, and much in-between (and I own several of them), some are definitely better than others. See a much more thorough comparison of these various types of valve seat machining tools in the form of an illustrated article at this link. It is useful to have the information to make good choices. However, I regret that this won’t be very helpful in choosing a machine shop to do the work for you as any shop will ensure you that they know what to do, when in fact they may be working to very old theories and using tradtional but not very accurate machining methods. But forewarned is forearmed I believe and if nothing else you now have a better appreciation of the complexities of proper cylinder head servicing.

Compression and leakdown tests
Cylinder compression is marvelous! It’s an efficiency thing; squeeze the combustible harder and it will more completely convert the fuel’s thermal energy into pressure. It’ll push back harder. Compression also determines how your engine uses fuel; a mechanically sound engine is less wasteful and thus carburets better. In fact, vintage Honda performance shortcomings that virtually everyone blames on carburetion can in many cases be traced to low compression. Unfortunately, compression is a problem on 40-50-year-old Hondas. Back to that soft valve thing we spoke of earlier.

Every vintage Honda owner needs a compression tester. And try to spend more than $49 on one. Compression testers having screw-on adapters often indicate low quality. No professional uses that kind of tester, not because of the meaningless debate raging on forums about the volume of the adapter or the hose, but because the adapters are glitchy sources of leaking and durability issues, and they most often exemplify very cheaply made compression tester kits.

It also helps to have a mechanic’s view of compression testing, which is equal to a doctor’s blood pressure test. Part of a routine maintenance service, that’s the pro view. In fact, undertaking the test intelligently is crucial. First, a reliable, quality tool. Next, have a well-charged battery. I sometimes will recharge the battery between cylinders simply to ensure the most accurate results. Crank the engine with the kill switch turned off and the throttle held wide open and stop when the compression gauge stops increasing. Do not put oil in the cylinder and do not warm up the engine. If you’re above sea level you will need to correct your readings for altitude; at 5,000 feet you’ve already lost 15 percent of the available oxygen and this gets worse as you go up, naturally. And realize something important: Having multicylinder compression results whose values are close together is a nice thing (mostly relevant to idle smoothness), but if they are also low what good is that? 1970s four-stroke Hondas were mostly manufactured with 165-170 psi cylinder pressure. Some say the book is wrong out of frustration with their results and an unawareness of the soft valve issue. But these numbers are correct and career Honda mechanics have confirmed them through several generations of engine testing and rebuilding. Naturally, fifty years later we can’t expect non-rebuilt engines to yield as-manufactured compression. Therefore consider the decision point for a 50-year old engine that had 165-170 psi new to be 150 psi today. If you have that, you are in good shape. If less, don’t spend money on carbs or ignition or whatever. Fix the engine first. On a related note, it is a fallacy to believe that a freshly-rebuilt engine can have lower than expected cylinder compression because it, “has to be broken in”. Proper machining and factory parts result in exactly the Honda spec and the most compression the engine will ever have by the end of the first test ride if not earlier.

A test related to compression is cylinder leakdown. A leakdown test is easy. The cylinder being leak-tested is put at top dead center on the compression stroke (TDCC) and air pumped in from an air compressor. Before attaching the tool I usually watch the intake valve open and close so I know that the next thing in rotating the crankshaft is this important place. Twin-cylinder engines can sometimes lack the internal friction needed to make the piston stay at top dead center—you may need to hold the crankshaft with a wrench.

Just as with the compression test, do not put oil in the cylinder, and do not pre-warm the engine. Normal in-service leakdown readings are 10-15%. A recently rebuilt engine will show around 5%. Readings below that are very unusual because they’re in the realm of racing engines especially modified (zero-gap rings, gas-ported pistons) to get that low a reading. Results above 10% indicate some wear but nothing terrible. But beyond 15% indicates a level of wear that demands rebuilding sooner or later. While the test is underway I like to roll up a piece of paper and use it to listen to the air escaping from various places. Hissing from the carburetor of course points to intake valves and at the exhaust to exhaust valves. Air sounds at the oil dipstick hole or under the valve cover point to rings and cylinder wear. This air movement is normal and a concern only when the leakdown reading is high. Some folks have experienced strange readings from their leakdown tester, such as 1% leakdown in a non-rebuilt engine. This is not believable and means either the tester is not reliable or the tool’s input air pressure is too high. 80 to 100 psi is the normal leakdown tester input pressure.

Some people view a leakdown test as just another form of compression test, albeit somewhat more sophisticated. That misses the point. A compression test can sometimes be somewhat limited, but it is far from dispensable. The correct diagnostic approach is to use the two together—first the compression test, then the leakdown test. Here are examples. Let's say the engine is running poorly. A compression test indicates good compression, over 150 psi. However, a subsequent leakdown test is not so promising, showing a 20-30 percent loss. Wait. How can this be? How can the engine have good compression but also high leakdown? Easy. Receded valves. Even badly worn valves sometimes seal adequately when repeatedly banged open and closed during a compression test. However, the completely static nature of a leakdown test will expose those tricky (usually extremely worn) valves. Now another example from the opposite viewpoint. How about low leakdown combined with low compression? Huh? Is this even possible? It sure is, but what does it mean? This is the classic symptom of retarded valve timing, typically from worn cams (common on vintage Hondas whose cams can wear half a millimeter in just 6,000 miles) or worn or mistimed cam chains or belts, all of which result in low cylinder compression despite the good valve and cylinder seal.

Two out-of-the-box techniques add even further to the leakdown test’s usefulness. One of them is to lower the input pressure until it is possible to rotate the crankshaft without the air pressure taking over and moving the piston. This Smokey Yunick (Google him) technique is used to find cylinder flaws and was how I found a damaged cylinder many years ago in a ZX11. A compression test overlooked the wear (returned a good reading) because cylinder compression is built up in the upper one-third of the cylinder and the trench in this cylinder was below that, and incidentally causing massive oil consumption. Another advanced technique is to, while the leakdown test is underway, carefully rap on the valve tappet and watch the leakdown tester’s gauge. I do this whenever seeing a high reading. If a high leakdown reading decreases and stays decreased, you know that it was mostly due to carbon buildup and not bad valves.