| A friend of mine, who shall remain
anonymous (his initials are B.J.) built an engine for his 73 914 the hard way. He
bought a box of parts at the Pomona Swap Meet that were allegedly once a 2.0 liter engine,
got a good deal on new euro pistons and cylinders, and spent a small fortune on everything
else he needed to make it run. He installed the fresh mill in his car, fired it up, and
found to his horror that it leaked oil in a steady drip from the passenger side of the oil
He partially dropped the engine (see his notes, at the end of
the article), removed the fan housing, and pulled the oil pump. He installed a new pump,
smearing sealant lightly on the gasket sealing surfaces, in defiance of the factory
manual. After reassembling the car, it still leaked, in exactly the same spot. Bee Jay was
one unhappy soul. He called me.
Cars, unlike people, are inanimate objects and follow the rules of
logic, engineering, and physics. They dont leak out of spite. It still took a while
to convince Bee Jay he shouldnt push his red roadster into the street and torch it.
We must pull the oil pump again, and examine the sealing surfaces. The odds were good that
the case or oil pump had some imperfections which were too big for the gasket to seal. By
now, Bee Jay was getting proficient at dropping the front of the engine and removing the
fan housing. He pulled the oil pump, and I slid under the car to have a look.
To properly check the flatness of the sealing surface on the case, I
wanted to dress it with a known flat reference plate. To do that, we had to remove the oil
pump mounting studs. Big problem. Bee Jay informed me his mechanic had mounted the studs
with red Loctite, which is one very aggressive locking chemical. Attempting to loosen
studs set with red Loctite usually results in a snapped studs or pulled threads in the
case. To avoid this, the Loctite must be killed by heating it above 400 degrees F. I did
not look forward to applying a propane flame to the end of a stud while lying on my back
under an assembled car. I envisioned myself being be nominated for a Darwin
Award, posthumously. Before firing up the Burnz-O-Matic, we tried double-nutting the
stud. This is the old trick of threading on two nuts, and using two wrenches to jam them
together. When Bee Jay attempted to remove the studs this way, all four came right out. I
was relieved, but puzzled. The Loctite didnt put up enough fight. When it came time
to reinstall the studs, the answer became obvious.
We now had clear but uncomfortable access to the sealing surface of
the case. See photo 1. I cleaned the surface with lacquer thinner, then I stuffed rags
into the cavity where the oil pump normally lives, to keep sanding grit out of the
crankcase. I also taped over the oil intake and output galleries in the case. To make any
irregularities stand out, I blackened the sealing surface with a felt-tip pen.
See photo 2. If I were a professional machinist, I would use Prussian
Blue or Dykem. (On the other hand, if I were a professional machinist, Id insist on
Bee Jay completely disassembling the engine for proper milling of the oil pump surface.)
My plan was to lightly sand the mating surface against a flat reference plate to find
irregularities. But, I needed a reference plate.
I made my reference plate by cutting a 3 ¼ inch by 4 inch rectangle
out of a chunk of scrap ¼ inch steel plate. I drilled a ¼ inch hole in the center, then
cut a wide countersink on one side. I pushed a flat-head ¼ inch bolt through the hole,
and threaded a bunch of nuts onto the opposite side to make a handle.
photo 3. The flat-head bolt sat slightly below the surface of the plate, and the old
plate had some rust pits in it, so I degreased it and smeared on a bit of two-part spot
putty to fill the holes. When the putty cured, I sanded the surface perfectly flat by
rubbing it against a thick glass plate with a sheet of 120 grit sand paper stuck to it.
The result was a crude but accurate surface plate with a small handle.
photo 4. Again, if I were a professional machinist, Id make a fancy reference
plate at 10 times the cost, which would look better but be no flatter.
Under the car again, I wrapped a strip of 240 grit sandpaper around
the reference plate and saturated it with WD-40. A few light passes against the sealing
surface revealed the cause of Bee Jays woes. See photo 5.
The light areas where the dye was rubbed off are high spots, and dark areas are low spots.
The immediate cause of the leak was a shallow dent in the sealing surface on the passenger
side of the engine. Oddly enough, this is just where Bee Jay observed the oil seeping out.
The dent shows up as the dark crescent (photo 6, arrow 1). The
dent was right next to the output passage of the oil pump, where oil pressure is highest.
Now look at the mating surface at the top of the oil pump
cavity, right where the case halves mate (see photo 7, arrow 1).
One side of the case is shiny, and the other is black. There was a step in the mating
surface at the case split line! To exacerbate the problem, the aluminum around the studs
was raised above the mating surface. This shows up as shiny rings around the stud holes (see arrow 2 in photo 6,
arrow 2 in photo 7, and photo 8). The raised metal around the studs has the effect of
holding the pump away from the sealing surface, encouraging leaks. The poor gasket
didnt have a chance to seal against such an uneven mating surface!
Because the pump mating surface is surrounded by raised
bosses, I could only get bout a half inch of stroke as I sanded the mating surface flat. I
was also lying on my back and reaching over my head, which doesnt help either.
(Heres an old trick: when youre working on your back, use an empty antifreeze
jug as a pillow. Its easy to keep clean, its free, and it eliminates neck
strain.) After about an hour of careful sanding, the surface was perfectly flat, as shown
in photo 9, photo 10, and
photo 11. I used 180 grit wet-or-dry paper and kept it saturated
with WD-40. I also wiped up the gritty sludge from the case every couple of minutes. For
any sanding job, do yourself a favor and change the paper frequently. Worn paper gives the
wrong finish texture and wastes your time.
Once the surface was flat, I took a countersink and beveled
the edges of the stud holes to a depth of about .040 inches. This relieves the stress at
the surface of the casting and prevents the metal around the hole from being raised. This
is a really good idea for any stud, even in a cast iron block or steel plate.
I cleaned the area thoroughly with lacquer thinner and
lint-free rags (i.e., a well-worn T-shirt) to remove all traces of grit and oil. As a side
note, never assume new or freshly machined parts are clean clean them yourself. A
friend of mine received his bored and honed Chevy block from the machine shop, and
assembled it without cleaning it up first. Then, he was shocked when it wore out in 18,000
miles. All the machining grit was packed into the honing marks in the cylinder walls, and
it ate the new rings! Machine shops dont clean parts thoroughly, because nobody
wants to pay them their hourly rate to clean parts. New parts are often covered in oil to
protect them from corrosion. The sticky oil collects dust and machining swarf clean
it off with degreaser!
Before remounting the studs, I cleaned their threaded mounting
holes with Q-tips and lacquer thinner. The mystery of the non-sticking Loctite became
instantly clear. On my first pass, the Q-tips came out black with engine oil and machining
chips! The mechanic who assembled the engine obviously hadnt bothered to clean the
holes. Like other glues and sealants, Loctite doesnt stick to oily surfaces. Lacquer
thinner and brake cleaner are excellent solvents for prepping parts before using Loctite;
keep swirling solvent-soaked Q-tips around in the stud holes until they come out perfectly
clean. Dont forget to clean the studs, too - cleanliness is next to stickiness. Put
a couple drops of red Loctite on the threads of the stud, and install them until they just
bottom in their holes. Thats right - just finger tight! Over torquing can result in
With the studs installed, Bee Jay reassembled the oil pump and
the rest of the engine, and fired it up. The leak was gone, and the townsfolk danced in
So, where did the problem come from in the first place? When
the engine left the factory, all the mating surfaces were very accurately machined. Over
the years, the case deteriorated. We know the engine had been rebuilt at least once
before, because the oil gallery plugs had been welded over. The plugs tend to loosen and
leak with age, but the preferred fix is to remove the plugs, tap the holes, and insert
threaded pipe plugs. This way, the plugs can be removed during an engine rebuild, and the
oil galleries can be cleaned with a rifle bore brush.
Anyway, the disassembled case was treated with less respect
than a giveaway T-shirt. Bee Jay let it slide around in the bed of his pickup for a week.
This might be OK for Chevy blocks made of Detroit Wonder Metal (cast iron), but it is a
good way to damage an aluminum casting. The dent that caused the leak might have been
caused by the case sliding into something in the pickup bed, or the case could have been
bumped before Bee Jay bought it, or someone might have pried the oil pump out by inserting
a screwdriver between the pump and the mating surface.
The raised metal around the studs is a common problem. The
stud stretches under load, and tries to pull the threads out of the case. The load on the
threads is highest near the surface, and decreases as you go deeper into the stud hole.
Under this load, the aluminum at the surface of the case may yield locally, raising a
mound around the stud. Overtorquing the nuts on the studs (like, attempting to fix a
leak!) increases the probability of raising a ring. It can also happen with time due to
creep (slow deformation under load) of the case. As I said earlier, the raised ring tends
to hold the pump away from the mating surface, decreasing the clamping force on the
gasket, and increasing the chance of a leak. The fix is to file or mill the raised ring
flat, then countersink the stud hole. Countersinking places the highly stressed first
threads deeper into the case, where there is more metal around them to absorb the stress
and resist yielding.
This entire procedure would have been much easier to do on an
unassembled engine sitting on a workbench. When you are building an engine, treat the case
with respect, and check all the mating surfaces before assembling the engine. File down
raised rings around studs, and countersink the holes. Clean everything thoroughly.
The technique of flattening mating surfaces with sandpaper and
a reference plate can be used on dozens of applications. Carburetion problems are often
caused by air leaking through warped mating surfaces on the carbs, manifolds, and heads.
If your oil sump plate leaks, check it for warps. On small parts, I prefer to use a plate
of thick glass as my reference surface. Place a piece of wet-or-dry sandpaper on it (grit
up), flood the sandpaper with WD-40, and rub the part across the paper. Use a circular or
figure-8 motion to avoid sanding the surface to a convex shape. Give your gaskets a flat
surface to seal against, and enjoy a dry engine!
Note that Bee Jay used an aftermarket (Melling) oil pump, which resembles a VW type 1
pump. Heres Bee Jays procedure for pulling a 914 oil pump without removing the
engine from the car:
1. Jack the back of the car way up, insert two jack stands under the two rear jacking
2. Remove front engine tin, alternator, fan, and fan housing.
3. Support engine with jack. Spread the load of the jack by using a piece of wood
between the jack and the engine. Unbolt engine mount bar at engine and body.
4. Jack engine up about six inches, remove engine mount bar from engine and slide it as
far forward as possible over the shifter bar.
5. Lower the engine onto a jack stand, with a block of wood between the jack stand and
the engine case so you dont ding up the case.
6. Loosen case bolts around the oil pump.
7. Unbolt oil pump. Remove oil pump cover and internal gears. Attempt to remove oil
pump case with large water pump pliers. DO NOT PRY BETWEEN OIL PUMP AND ENGINE CASE!
8. After fixing the oil pump problem, installing the pump, retorquing the case bolts,
and refilling the engine with oil, you can check for leaks by running the engine for NO
MORE THAN 60 seconds (no fan or alternator).
9. Reverse above procedure to remount the engine.
Photo 1: The mating surface for the oil pump. You are looking at the front of the engine
from under the car. The drivers side is to the right in the photo. Bee Jay partially
dropped the engine for access to the oil pump. The mating surface doesnt look too
Photo 2: To help visualize the irregularities of the crankcase, I used a felt marker to
dye the mating surface. Next I needed a flat reference surface tool to sand off the high
Photo 3: This is the back side of our home made flattening tool. It is simply a 3 ¼ inch
by 4 inch piece of ¼ inch steel plate, with a ¼ inch flathead bolt sticking through it.
Crude, but effective.
Photo 4: The front of the tool. To ensure the pitted surface of the tool was perfectly
flat, I filled the pits with two-part body putty. After the putty cured, I flattened the
surface by rubbing it against a piece of 120-grit sandpaper stuck to a thick piece of
Photo 5: I wrapped the tool with 240 grit sandpaper and lightly rubbed the oil pump mating
surface. The sandpaper removed the dye from the high points, revealing the horrific truth.
The mating surface was grossly uneven.
Photo 6: Arrow 1 points to a crescent-shaped dent (dark area) on the passenger side of the
mating surface just where the oil leak was! Arrow 2 shows a ring of bright metal
surrounding a stud hole. The aluminum around the stud was raised, preventing the cover
plate from being pulled down against the gasket evenly.
Photo 7: More bad news. Arrow 1 points to the joint between the case halves. Note the
drivers side of the case is black, while the passenger side is shiny. This is due to
a significant difference in HEIGHT of the mating surface across the joint. Arrow 2 points
to raised metal around the upper drivers side stud hole.
Photo 8: A close up of the raised metal ring around the lower passenger-side stud hole.
Photo 9: After about an hour of careful sanding using the flattening tool, 180 grit
sandpaper, and a little WD-40, the mating surface looks much better.
Photo 10: The raised metal around the stud hole has been removed. Note the generous
chamfer around the stud hole, cut with a countersink. The mating surface was sanded down
enough to remove the dent on the passenger side.
Photo 11: The step at the crankcase joint is gone. So is the raised metal around the upper
drivers side stud hole.
Photo 12: The masking tape is removed. All surfaces and threaded holes are thoroughly
cleaned with lacquer thinner, Q-tips, and lint-free rags. The studs are installed finger
tight with a couple drops of red Loctite.
Photo 13: Bee Jay torques the oil pump cover into place. He tightened the nuts in a cross
pattern, and sneaked up on the final torque in several steps, increasing torque on each
round of tightening.
PLUGGING the LEAKY TIKKI PART II
As you remember from our last exciting adventure, the infamous Bee
Jay and I had flattened the seriously distorted oil pump mating surface on his 2.0 liter
914 engine. Providing a flat mating surface allowed the gasket to do its job, and the
annoying oil drip stopped. At least, the drip stopped at the oil pump, but Bee Jay
discovered another leak at the sump plate.
The sump plate of a 914 engine is held on by a single nut, which
looks like a big hollow bolt with female rather than male threads. This nut passes through
the central hole in the sump plate, and threads over a stud set into the cam bearing
saddle. Having a single centrally mounted nut ensures even clamping force on the gasket,
and eliminates the sump plate warping which has plagued 911, 356, and VW owners for ages.
Those other cars had multiple studs holding the sump plate on, and over enthusiastic
tightening of the nuts warped the plate, causing leaks.
Anyway, Bee Jays car leaked at the sump plate. An exchange of
e-mails with other 914 owners brought dire predictions of cracks in the crankcase where
the sump plate mounting stud threads in. Such a crack allows the plate to loosen up, and
can be fixed only by disassembling the case and welding. With this gloomy prospect in
mind, we drained the oil, removed the sump plate, and crawled under for a look.
The sump plate looked fine, no warps. I expected the mating surface
on the crankcase to be in excellent shape, since it is protected by a thick rib cast into
the case around the sump. The mating surface was smooth almost. Where the case
halves came together, there was a tiny step at the rear of the mating surface, and a
larger step at the front. Bee Jay ran his fingernail over the front step, and claimed it
felt like a speed bump. OK, he sometimes exaggerates a bit.
Eliminating the step would not have been easy, except Bee Jay had
just bought the perfect tool for the job. The Weltmeister deep sump kit consists of a
stout steel can which replaces the oil sump plate, a pickup tube extension, gaskets, and
mounting hardware. The deep sump is supposed to help keep the oil pump pickup tube
submerged in oil during vigorous cornering. I was more interested in using the steel can
as a tool for sanding the crankcase mating surface flat.
First, I checked the steel sump can to make sure its mating surface
was flat. I did this by placing a piece of 240-grit wet-or-dry sandpaper on a flat piece
of plate glass (yes, abrasive side up!) and flooding it with WD-40 as a lubricant. I then
rubbed the mating surface of the Weltmeister sump can against the sandpaper. Any warping
of the mating surface would show up as uneven polishing marks. The mating surface of the
steel can was perfectly flat, which is a lot more than I can say about a lot of other
aftermarket parts Ive checked.
To convert the steel sump can into a grinding tool, I cleaned its
mating surface with lacquer thinner, and gave it a thin coat of contact cement. I also
applied a ring of contact cement to the back side of a piece of 120-grit wet-or-dry
sandpaper. When the cement was nearly dry, I pressed the paper to the sump can, gluing the
paper to the mating surface. I carefully trimmed the sandpaper to the exact outer diameter
of the can, using an X-Acto knife with a fresh blade. I also cut a hole in the center of
the sandpaper, to clear the sump mounting stud in the crankcase.
We now had a disc of sandpaper neatly glued to the mating surface of
the steel sump can. I squirted some WD-40 on the sandpaper as a lubricant. To flatten the
crankcase mating surface, I pressed the sandpaper-coated mating surface of the sump can
against the mating surface of the case, and rotated the can back and forth. This meant, of
course, that I was working above my head. I found it easiest to press the can against the
case with my right hand, and rotate it with my left hand. I stopped every few minutes to
clean the excess sanding sludge off the case, squirt more WD-40 on the sandpaper, and rest
my arms. After a half hour my arms were sore, but the mating surface of the case was
perfectly flat no more step.
We peeled the sandpaper from the sump can, and cleaned the can and
crankcase with paper towels and lacquer thinner. No sense introducing sanding grit into
the oil pump! Bee Jay assembled the sump and filled the engine with fresh oil. The leak
was gone. Torque the sump nut to 9.5 ft-lb, and no more!
One must wonder why the crankcase had a step in the mating surface
of both the oil pump and the sump. It certainly didnt leave the factory that way.
Bee Jay had bought this engine disassembled at a swap meet. Was his crankcase in fact two
halves from different engines? Perhaps the locating dowels in the case halves were not a
snug fit. He had the case align bored, so the bearings should not be a problem.
Anyway, this is another illustration of the method for flattening
surfaces we used in Part I of the article. The method is not pretty, but it is cheap and
it works. Again, this would have been a lot easier to fix if we had caught the problem
before the engine was assembled an installed.