Sunday, December 3, 2006

Leaky Oil Seal

Kid comes in the shop wringing his hands, all upset. He's just paid a lot of money for someone to rebuild the engine in his bug and it's leaking oil from the front seal. The guy who did the engine sez it's not his fault, it must be coming from the rear tranny seal, wants to sell the kid a rebuilt tranny. And besides, all VW's drip oil; no big deal. But the kid is sure it's leaking more than it did before it was overhauled and he's never had his tranny leak and the leaky stuff doesn't have that thick sulphury smell like tranny lube and his mom is all upset about the gunk on the driveway.

That last is the real reason he's here :-)

So you drop the engine, pull the old seal, check the end-play ('way off), clean the oil drain drilling, put in a new shim stack and a new oil seal. To further the kid's education you show him how the seal has been ruined; whoever installed it used a hammer.

Hammer = Drips

At least half the drippy oil seals I see are caused by improper installation of the seal. A few are due to improper engine assembly. The remainder are from excessive end-float; the thrust face of the #1 bearing is simply worn out, allowing the crankshaft to move back & forth. When it does, it acts as a nice oil pump, defeating the purpose of the lip-type seal.

The #1 main bearing oil seal is neoprene or silicone rubber bonded to a metal ring with a circular coiled spring inside to maintain a leak-free sliding fit around the center boss of the flywheel. The oil seal fits into a recess cast into the crankcase; it's a tight fit. Properly installed, the oil seal ends up slightly below the level of the casting.

The factory service manual shows the seal being pressed into place using a screw-type fixture, but an experienced mechanic can install one using repeated light taps from a plastic mallet. An unskilled mechanic will try to do it with a hammer and while it might look okay, about half the time the hammer blows cut the silicone rubber where it's molded over the invisible metal ring inside of the oil seal. Oil quickly discovers the cuts and you've got a leaker on your hands.

Too Much Sealant = Drips

A leaky seal due to improper engine assembly is a bit more subtle.

Like all of the main bearings, #1 is generously supplied with pressured oil form the main oil gallery. In normal operation the oil lubricates the journal and escapes from both sides of the bearing. On the flywheel-side of the bearing the oil collects between the bearing and oil seal, flowing back to the sump via a drilling in the left half of the crankcase. Want to guess what happens if that drilling gets blocked? (Be careful, it's a trick question.)

When the #1 main bearing oil return is blocked, oil pressure will build up behind the seal and the thing will eventually leak. That's the obvious answer. But a blocked oil return port also results in accelerated wear since the thrust face of the bearing and the associated shim stack is not being provided with a circulating supply of oil.

Most often, the oil return passage is blocked by an over zealous application of sealant when the crankcase halves are joined. Here's how it happens: The left half of the crankcase is in the fixture, parting line up, the guy swabs on about four times more sealant than needed and when he drops the right half of the crankcase into place it squeezes the sealant out, which flows downhill into the oil return passage from the #1 main bearing. And that's just on the inside of the crankcase. Outside, the sealant is oozing all over, including down into the recess for the oil seal. Being in the 'corner' of the recess, the oil return passage gets more than its share when it shouldn't have gotten any at all. (Hint: After closing the crankcase for the last time, check the oil return drilling with a Q-tip to ensure it is clear.)

I've also seen engines with sealant deliberately painted into the oil seal seat, apparently hoping to stave off leaks. In those cases the oil seal itself was always damaged by hammer blows.

Think about that for a minute. The guy builds an engine, hammers in the oil seal and sure enough, the puppy leaks like a sieve. So the next time he globs on a lot of sealant, hammers in another oil seal and this time it leaks even worse, convincing him it's impossible to keep a VW engine from leaking. (And besides, everyone sez VW's leak. Conventional Wisdom wins again.)

Indeed, when you combine an improperly installed oil seal with an improperly assembled engine (ie, the blocked drain hole) the engine doesn't just leak, it gushes. In effect, the builder has just created a direct path from the oil pump to the ground under the engine.

So why do people install oil seals with a hammer? First, because they see a real mechanic do it successfully and never understand that it takes considerable skill to do it right. Secondly, they do it because most of the manuals say it's okay to hammer it in... and about half the manuals show the seal installed incorrectly, flush to the outside of the crankcase. Correctly installed, the seal will be slightly below that level.

But the most common reason for all those drippy engines is the fact everyone assumes that hammering requires no skill.

Push or Pull = No Drips

Oil seals are designed to be pressed or pulled into their seats. It's possible for an experienced mechanic to install them with a plastic mallet, or even a hammer in the case of some axle seals, but it's also possible for a skilled surgeon to do an appendectomy with a pocket-knife. The emphasis here is on the skill, not the tools.

Oil seals aren't expensive and they don't look very sophisticated but there's more to them than meets the eye. If you toss an old one on the barby and wait for a while you'll be able to examine what's under the rubber. You'll see that most of them start out as a segmented ring of thin sharp steel. Cover that with rubber, tap on it with a hammer and it cuts the rubber as neatly as a knife.

So press them in. Or pull them. You can make a dandy puller-presser for your front brake drums (the seals you'll replace most often) using nothing more than a length of all-thread, some washers and three nuts. And you can buy a screw-type pressor for the #1 main bearing seal, although it's easy enough to make one, assuming you have a lathe. (See the drawing at the top of this article.)

Or you may drive them in with a 'seal-seater,' if the seal is small. By distributing the force of the hammer blows uniformly, a seal driver lets you pop the things into place with one or two well placed blows of a hammer. If you've a lathe, making oil seal drivers is a spare-time sorta thing; all are simple turnings, and aluminum or even hardwood works as well as steel.

Big seals are different. Because of their tendency to cock in the bore, large-diameter seals are best installed with a press or fixture. Rear axle seals are especially troublesome due to their deeply recessed position in the seal cover. Because of their proximity to the brakes and the fact that any leak could leave you without brakes, the wiser course is to always press-in rear axle seals.


With the exception of the Muir manual, books on maintaining your Volkswagen assume a certain level of competence. Learning to tap a seal into place with a plastic mallet isn't difficult; it's one of the many minor skills acquired during the apprenticeship all mechanics must endure. It is also one of the minor skills many self-taught mechanics never bother to master. (Hint: Start with an old seal. And an old engine case. When you can tap the thing in a dozen times in a row without damaging either the case or the seal you're probably ready to try it with a new seal on a good case. Along the way, you will have learned how to remove the thing as well.) But installing a seal with a hammer falls into the category of 'Field Repairs;' things a skilled mechanic must do when the proper tools are not available. Pressing the seal into place is not only safer, it's usually faster. And a pressed-into-place seal is cheap insurance against oil leaks.

Copyright © 1995 Robert S. Hoover

One Engine's-Worth of Parts

In response to one of Rocky's messages I mentioned a number of things that can effect compression ratio. Sunday I go to check the mail and there's this buncha guys peering in my window all saying pretty much the same thing:

I don't see how ... (you fill in the blank) can have any effect on CR.

A minor variation on the theme was:

(Your favorite expert's name goes here)... sez to do it like ( whatever) and never mentions (...various unmentionables...).

Please accept the following as a general answer for all.


The four holes in the crankcase that accept the cylinder barrels are called spigot bores. The area around each bore is called the deck and serves to support the cylinder. The decks of all four spigot bores must be the same distance from the center-line of the crankshaft. This is something you check before you start building any VW engine even when using a new crankcase because sometimes the axis of the crankshaft is machined slightly eccentric, meaning the main bearing bores are a little bit deeper in one half of the crankcase than the other. Or more rarely, machined at a slight angle, with the clutch-end being more to the left, the pulley-end to the right (or visa-versa). Not often but it happens. So you check it.

With any used crankcase the spigot bore decks will have been re-faced -- re-machined to get rid of the shuffle marks. Good shops with the right equipment always machine the case decks so all four will match but if you buy a used crankcase from a shade-tree mechanic or a shop that caters to the kiddie trade you're liable to find almost anything. I've seen cases with as much as sixty thou variation in the spigot deck height from one side to the other... and almost that much on the same side of some cases, which tells you the case came from a drill-press operation (ie, a shop that doesn't have a milling machine).

Your jugs sit on the deck around the spigot bores. If there is any difference in their height it will be reflected in the height of the cylinders. And since the con-rod extension is relative to the center-line of the crankcase, any variation in the height of the cylinders will show up as a difference in the deck-height of the piston at TDC.

And that will effect your Compression Ratio.

If that's not clear, make a drawing and work it out but the message here is that you have to know what your case-deck-height is. You can't guess. You need to blueprint the case and record your findings, whatever they are, because you're about to build on that foundation and by the time you get out to the heads you will have stacked up half a dozen components and even the smallest variations will have become significant because of the stack-up.

Major point here is that there is always some amount of deviation from spec in the parts going into your engine. With an army of inspectors to insure the quality of every step in the manufacturing process, for original Volkswagen parts the variations would tend to cancel each other out rather than stack up. That's not true with after-market parts. The only way to know what you have is to measure what you got. Some guys call this 'blueprinting' and make a big deal out of it but it's mostly common sense.


Set the crank up in vee blocks or with fitted bearings in a known-true case half and check the length of the throws, even if it's a good crank you've just sent out for a polish. Sometimes the grinder will have a bad day and you'll end up with a crank having a slightly different stroke on one (or more!) of the journals. So you check it to within the accuracy of your tooling and record the results. Usually, cranks are pretty good. Some of those cranks coming in from China are as good as any I've seen. But some are trash. Ditto for a lot of welded strokers aimed at the Kiddie Trade, with examples of every problem you can name being woefully common. You have to check and record what you find even when any variation falls within acceptable limits because that variation, whatever it is, will add to or subtract from the finished dimensions of the engine.


To ‘rebuild' a rod you re-bush the little end, hone the bush to spec then pull apart the big end, use a surface grinder to remove a little metal from the parting line, torque it back together and machine the big-end back to a true circle relative to the little end. That is, you try to keep the distance between the center of the big end to the center of the little end the same as for a new rod fresh from the factory.

Sunnen (brand name) honer that has been properly maintained, skilled machinist... you can produce a pretty good rod. Shops that cater to the kiddie trade... wetback labor... worn-out or poorly maintained machine tools... Forget about it.

So what's the spec for a stock length rod? I donno... 137mm? Something like that.

Doesn't really matter. (!!) What matters is that all four of your rods must be of identical length. That's what matters. Long or short, you can deal with that but only if they are all the same.

But they won't be. There will be some variation in their center-to-center length, center of mass and over-all mass. You'll take care of the weigh differences during balancing but right now you need to know the variation in their center-to-center length, which is pretty easy to measure even with simple tools if you use one journal of a crankcase as your center on the big end and a well fitted wrist pin on the other.

Con-rods are numbered. Use their number in your records when you record the difference in their lengths. SOP is to identify the shortest rod then simply record the differences of the other three as 'pluses.' Good rods, you'll be working in tenths... +8, +4 (ie, +0.0008, +0.0004, etc.)

What's a well fitted wrist pin? Oiled and at room temperature, you should be able to slide the pin into the little-end with your hands. Once in, it should fit well enough so that the pin takes at least two or three seconds to slide out when the rod is held horizontally (and the pin is installed flush). Slower is better. At running temps the rod will expand more than the pin so a good fit is one that is damned tight at room temperature.

A lot of rods aimed at the kiddie trade or used by lo-buck rebuilders aren't even overhauled. They only knurl the bushing then hone it back to size and simply hit the big-end with a hammer before honing, if they bother to hone it at all.

Shop by price, you'll end up buying junk. Good shops are proud of the quality of their work, offer no objection if you want to mike a part now & then. Ditto for good dealers. The other kind don't want anything to do with real mechanics. And get their wish :-)

The point here is that the length of the connecting rods effects the Compression Ratio.


Pistons & cylinders are manufactured individually then sorted according to their finished diameter (for jugs) and weight (for pistons). The different sizes and weights are identified by dots of colored paint on the pistons.

In manufacturing a cylinder barrel the raw casting is first machined then the machined barrel is honed to remove the tool marks. In the process of machining a given number of cylinders, the finished bore will become gradually smaller as the tool-bit wears down. When it gets to a certain minimum size they stop the machine and set it back up with a new boring tool. That means the inside diameter of the jugs will fall across a certain range of diameters. This is normal.

The honed jugs are measured and divided into groups according to some standard deviation in their diameter, typically about a thousandth of an inch. But even with that small a standard, with four jugs from the same size-group you can expect to find a variation in their diameter. It won't be much but you need check it.

Volkswagen used cast aluminum pistons from permanent molds. The density of cast aluminum varies slightly according to how much metal is in the smelting pot, its temperature and how long its been there. The castings are then machined to a given diameter, for the grooves where piston rings, for the wrist pin and for the top of the piston. All other surfaces are usually left as-cast. As with all machining operations, the finished dimensions will fall across a range of sizes.

The combination of differing density in the aluminum alloy and variations in the as-cast dimensions causes VW pistons to vary in weight by as much as an ounce (!) Even by 1930's standards that's a bit much so the pistons get sorted into three weight groups with each group having a maximum variation of ten grams.

The nominal dimension of the piston (i.e., its size group) is stamped on the top and a dot of colored paint is used to indicate which direction its actual dimension deviates from the stamped figure. Another dot of different colored paint is used to indicate the piston's weight group and a plus or minus symbol is stamped into the top of the piston to indicate if the piston's weight is above or below the nominal weight for that group.

The pistons are divided into groups according to their weight and within each weight group, are divided into groups according to their diameter, allowing them to be matched with suitable jugs, fitted with rings and packaged for shipment. Stock jugs used to be available individually; nowadays all you'll see are sets of four.

But your carton of new pistons & cylinders may arrive as a grossly mis-matched set of junk. Here's why: Some after-market retailers -- or the clerks who work for them -- tear open the boxes and shuffle sets around to make up sets having the largest bore diameter and identical weight markings. Some dealers even brag about this in their advertising, referring to such sets as the ‘pick of the litter' that need no further balancing. And sell such sets at inflated prices.

It's all bullshit of course, a minor deception aimed squarely at the Kiddie Trade. Why? Because with a weight group encompassing ten grams, with two divisions and a mark for high or low the best you can hope for is a spread of 2.5g... about 25x worse than a real balancing job. (Using an inexpensive electronic scale for measuring and a Dremel tool for removing metal, the average novice has no trouble matching four pistons to within a gram or two.)

But the most interesting point of all this is what happens after those sets of pistons have been pawed over by the clerks. They get tossed back into the boxes willy-nilly and sold to unsuspecting suckers, including other retailers.

The tricky bit here is that you can't balance a set of pistons if they span two weight groups. Pistons are provided with extra metal in the form of ‘balancing pads,' areas from which you may remove metal without effecting the strength of the piston. But the maximum amount you can remove is only a few grams. That isn't a problem when all of the pistons are from the same weight group. But when your P&C's are a mix of two (or more!) weight groups you're liable to see as much as 20 grams difference across your four 'brand new' jugs. Not only does that violate the factory spec of 10g, the difference is too large to be balanced out - - there simply isn't enough metal that can be safely removed.

You just paid good money for a set of new jugs that are junk.

But this is about compression ratio so let's get back to that.

First thing you gotta do is examine your new set of P&C's to make sure they are of the same size group (ie, the variation of diameter) and within the same weight group. That is, all four of the jugs in the box should have the same color code for dimension and the same color basic color code for weight group. The code for plus & minus doesn't matter because you're going to have them re-balanced to a finer standard of precision (i.e., typically +/- 0.1g across a set of 4).

You should do all that before you buy them. And yes, you can get royally screwed when buying through the mail. No, I won't recommend anyone -- I've been sued both ways on that one, once because a guy was unhappy with someone I recommended and another time by a dealer because I didn't recommend him. So go fish. And good luck. Because getting a set of P&C's that hasn't been tampered with is just the start of the story.

Once you have a set of P&C you'll need to put identifying marks on the jugs and record the marks and the dimensions in your notes. I file notches in the flat area of the upper-most fin. When you have more than one engine in the shop at a time, keeping their parts separate can be a problem. I use a series of adjoining notches to identify the set then one to four additional notches, spaced apart, to identify a particular jug within a set. The notches are cut with die-grinder as soon as I open the box. The pistons have to stay with their particular jug so you need to put a matching mark or number on the underside of that piston. I use a vibrating scriber.

Begin your measurements with the distance between the deck lip and the top of the cylinder barrel. The easy way to do this is to just stand the thing on its head and use a surface gauge to find the tallest barrel then record any difference in the other three. Here again, you can expect some small variation.

Barrel length is an especially critical dimension in an horizontally opposed engine since it is the foundation of the valve train geometry. This dimension is even more important in horizontally opposed engines like the Volkswagen which depend upon head studs (or stays) to maintain the seal between the cylinder and the head since any difference in the length of the barrels will impose an asymmetric load on the sealing surface leading to compression leaks.

After measuring the length of the barrels the pistons are removed and the pin height is measured. Follow the same general procedure; put the piston, head down, on a surface plate, use a gauge to find the tallest then record the difference between it and the other. (As a point of interest, in most cases there's nothing to record - - the dimensions match to within less than a thousandth of an inch and an amount that small is generally not significant. What I'm really looking for here is any radical departure from the norm.) But the fact remains, any dimensional variation in bore diameter, barrel length and piston pin height will have some effect on your Compression Ratio

The rings get removed and bunch of other work gets done but we're only talking CR here so I won't go into the other stuff.


As with the jugs, when measuring the heads you must first identify them. Through the course of assembling an engine the heads get a lot of work done to them and you need to keep good records. I stamp numbers on them, over by the right-hand exhaust stack (right-hand looking into the chambers, push-rods down). Doesn't really matter how you identify them just so you do. I use stamped numbers because in prepping a set of heads I usually replace some of the guides, run them through the blasting cabinet to roughen up certain areas then open up the chambers, unshroud the valves and do a few other things, most of which will destroy any kind of temporary markings.

On the chamber-side of the head casting you will find either a fully machined flat area surrounding the chambers (old style heads and some after-market types) or six machined bosses, three to each chamber. The horizontal plane defined by the machined surface, either of the bosses or of the flat area, is the base-line for all of your head dimensions.

You need to know the distance from that horizontal plane to the sealing surface of the combustion chamber. More specifically, you want that distance to be as close to identical as possible for both heads and, within a head, for both chambers.

This dimension can be all over the map if the heads have been opened up by a schlock shop. Good shop, any variation should only be a few tenths (ie, ten-thousandths of an inch) up to a max of half a thou (ie, fifty ten-thousandths). Shlock shop, using a cutter in a drill press, you won't believe the crap they turn out.

This dimension is especially critical in the fabrication of a good VW engine. If this distance varies by more than two thou between the chambers of the same head, or by five thou between a pair of heads, have the heads fly-cut by the minimum amount needed to arrive at a uniform figure for all four chambers.

With measurements for the case deck height, barrel length, rod length and piston head height, and knowing the compression ratio you are planning to use, measuring your chamber volumes tells you how much you will have to open them up to achieve the desired compression ratio. Indeed, once you've nailed down a few dimensions, setting up the correct compression ratio becomes something of a no-brainer.

And somewhere about now you'll realize this message wasn't about compression ratio at all :-)

There are two main reasons for doing the work described above. The first is to be able to identify good parts from bad parts. You can't make this determination by price nor the fact the part is new, rebuilt or whatever. Nowadays there is so much junk out there the wiser course is to assume you're dealing with shoddy goods until its specs prove otherwise.

As you progress through the measurement of the parts you begin to see ways in which you can combine those parts so as to arrive at the most dimensionally-uniform result. For example, a slightly short throw on the crank can be combined with a slightly long rod. The same is true for the jugs and the heads in that some combinations may be used to cancel out dimensional variations.

A nice point to keep in mind here is that the ‘assembly' of a ‘paper' engine is an arm-chair activity. You may take as long you wish, shuffling the numbers about in every possible combination until you arrive the one that makes the best possible use of that particular set of parts.


Did the light-bulb come on over your head? You see, the typical engine-builder can only afford one set of parts. And as much as I hate to say it, if you simply bolt them together the odds of getting a good engine are vanishingly small. Oh, it'll run. Veedubs are robust little buggers... almost anything will run. But if you simply throw the thing together it will not run as well as it should nor last as long as it could. And you won't know the difference.

But I'm not a machinist... (I heard someone shout).

Neither was W. Edwards Deming. He was a statistician with the Bureau of the Census. (Never heard of him? Your loss.)

The truth is, you don't need to be a machinist to build a better engine. You can do that by simply taking a few measurements and keeping good notes. That's enough to keep you from building a total piece of shit. When you subtract the POS Probability Factor from the engine building equation you automatically end up with a better engine. How much better? On average, about twice as good. Yeah, I know... nobody else believes it either. Except for the guys who have done it. (Didja read my article on dialing in your cam? Ditto.)

Up to you. It's your engine.

-Bob Hoover