You can always borrow tools. No one is going to lend you new P's & C's.

Yep, that's the Millimess. Here's what I did to measure the valve stems.

The Millimess is a comparator, not a direct-reading instrument. That means that it has to be zeroed with a gage block at exactly the check dimension, then you introduce the part and read the difference from spec. In this case, I used the 9mm gage block from my set of 87 Metric Gage blocks. First you clean all the oil and grease off the block. As this is a brand-new set for me, there was significant cosmoline that I stripped off with rubbing alcohol. The faces of the gage blocks that you measure with have been precision ground to a mirror finish, whereas the non-measurement faces have a dull finish and the dimension etched into the side. I should add that when working with gage blocks, you should wear powder-free gloves, because the oil from your skin can rust the block, altering its dimensions. And always clean the block after you've used it and coat it with a light oil before you put it away.

Anyway, I put the 9mm gage block under the Millimess pointer and zeroed it out by adjusting the height of the clamp first, then by wiggling the Millimess back and forth to move the stem a little, then tightening both clamps as tight as I could do by hand. There is also a fine adjustment screw on the right side which you can turn to set the pointer at exactly zero.

The valve stem was then introduced on the anvil and rolled back and forth under the indicator tip until the largest dimension was shown-- that way you know you've got the tip exactly tangent to the largest diameter of the valve stem.

As you can see, the gage reads -34 microns. One micron is one millionth of a meter, or .001mm (one thousandth of a thousandth is a millionth). So that's -0.034 from a gage block of 9.000mm, or 8.966mm. Now, a grade 2 gage block, which is what these are (you can spend more money for grade 0 but if you aren't working in a metrology lab, your grade 0 blocks will soon become grade 2 when you use them!) have a tolerance of +0.10, -0.05 microns. So that's under a tenth of a micron which I am considering negligible for the purposes of engine rebuilding even though we know it is there. But the point is the significant digits- that 9mm block is actually accurate to at least 9.000, because the tolerance is in tenths of microns, e.g. the block can range from 9.0001mm to 8.99995mm in size within that tolerance grade.

And now the little spec book copyright 1978 by Porsche AG

http://forums.pelicanparts.com/uploa...1175616154.jpg

The spec for mid-stem is 8.97- 0.012mm. Note that this is a one-sided tolerance, which makes sense, because if the stem is any bigger it will either bind or there won't be room for oil for cooling and lubrication. So the range really is 8.970 at the largest to 8.958mm at the smallest. Which is approximately where the markers on the face of the Millimess are. So back to the spec, this reads. . . 8.966mm, so it's just smaller than the max it can be. Which makes sense for an NOS intake valve.

On to the next valve! (It's a lot easier for the next five)

Now, I'm sure somebody is going to say that I should have used a v-block instead of an anvil to measure. If you use a v-block, then you should use a plug gage equal to the measuring diameter. . .which is a whole new set of tools! This is good enough for me!

Great work, John! I would expect that the stem would wear unevenly... did you take a few measurements long the stem to see how the stem wears?

Best regards,

Michael

Michael,

I haven't yet measured any worn valves with this setup, although I just received my spring compressor so hopefully I'll be able to diassemble the heads and start checking things. I'll check all along the stem, that's the beauty of the comparator, it's really fast. For fun I'll use the new .001mm micrometer to check my measurements vs. the comparator setup. Even more fun will be to check the accuracy of the mic against the gage blocks. . .we'll see if it's up to the standards of Mitutoyo!

The new 0.001mic is a 2-3" version with an electronic digital output. I think I should be able to use it to measure the rod bolts before and after stretch. I need to get a good look at the head of the bolt: if it contains an indentation, then perhaps pointed anvils would be required.

ARP specifies a stretch of:

0.0092" to 0.0096" for the 6003 rod bolts used in my 2,0.

In metric, that's

0.23368mm to .024384mm, so let's call the midpoint of the stretch range that you shoot for as 0.239mm, with a tolerance of .004mm either way. So 0.239 + .004 = .243, that's inside the high limit at .00956", or .239 -.004 = .235, inside the low limit at .00925".

Anyway, none of this can be achieved with a .01 resolution micrometer, because the difference between the high end of the stretch range ann the low is .01mm, so you'd end up in between the lines. You'd be stuck buying an English-reading ARP bolt stretch gage, and you'd have to zero it against a new bolt. At a minimum, if you were going to do this, you should measure all the bolts and figure out the deviation, so you would know if the particular bolt you were measuring was an outlier, sometimes happens. That or get the unstretched specification from ARP directly, or if using a Verbus bolt, use the diagram posted in the other thread (if you have a 2,4 with 73mm bolt, that is).

No offense intended but to me this should be called Operation Overkill. It's a car, not a space ship. If you enjoy it go nuts, but I hope no-one reads this thread and thinks this is necessary.

I'm not unimpressed, I just think it's overkill. Porsche, or not.

Excellent! Sounds like the type of work done by the late "Maestro" Harry Pellow.

I also have a bit of measurement fetish when it comes to working the lathe, unfortunately its only accurate to .0002 at 12"

PBH, I'm not offended in the slightest. Certainly plenty of engines have been built and run well without measuring to microns.

An interesting side note is Henry Schmidt's thread about the QC that he does on the QSC cylinders. In another thread, he mentions that he's measured new Mahle Cylinders and determined them to be out-of-round by .0005" to .0008" whereas his own cylinders leave with less than .0002" ovality. Mahle uses air gages to measure, which are capable of measuring in millionths of an inch, without contacting the cylinder bore, avoiding the associated "penetration" when a hard carbide bore gage distorts a comparatively soft aluminum cylinder wall.

Does it matter? Hard to say, because that cylinder distorts into a banana shape when it's torqued down, heated up to 180F and has a piston changing direction 233 times a second inside it. It goes back to the criticism of one Pelican who insisted on using a torque plate for honing cylinders even though it's impossible to replicate the wild thermal load the cylinder sees: you might be doing more harm than good, you don't know until you measure, and you can't measure! Note: many NASCAR engine builders use a torque plate heated to 220F for "hot honing" because they realize that imprecision causes friction which costs horsepower. But that's in a discipline where 1HP may win the race.

Also, Mahle has a different issue in that they use Statistical Process Control to manufacture large batches of cylinders, and it makes economic sense for them to substitute capital cost, in the form of Mahr air gages, for labor cost, in the form of more inspection time. Particularly in Germany, where the labor cost is very high.

That being said, here's my rationale for Project OVERKILL:

1) Most of the factory specs are in 0.001 and they actually show somebody using a Millimess to measure piston diameter in the old workshop manual;

2) I'm not a professional engine builder. I figure if I err on the side of being more accurate and double-check everything, there's less likelihood that I'll screw something up. Also, it's a confidence thing: when I turn the engine over for the first time, I want to hit the starter knowing that I know the condition of every part in the engine; and

3) I, too have a preference for cool measurement instruments.

So no, no offense taken, and thanks for stopping by.

I guess I prefer to go by feel. When my engine smokes or leaks I know it was caused by poor/worn components, no way was it poor assembly. You will have no excuses. :)

Now that you have used the term, please rename this thread as Operation Overkill.

Hey John

I'm loving this thread!

One question

Maybe I am missing something but...Why didn't you use a 0-1" OD micrometer to measure the valve OD? I've done this before and it would seem to be much easier and more accurate.

Tom,

Just wait till I start posting the standard deviation of the measurements. . . -ZZZ- :)

Certainly one could use a traditional outside mic with an 0.01 resolution. You would just have to round the Factory limits in the right direction, e.g. you would scrap any valve that measured less than 8.96, because the low limit is 8.958. I personally wouldn't put a valve in that measured at the low limit anyway.

Once the Millimess is set up, though, it's faster than a mic and doesn't rely on accurate "gaging pressure" the way a ratcheting mic does. In other words, the inspector can't alter the measurement by using too much, or too little, gaging force-- you just lay it on the anvil and the spring in the instrument does the rest. But I admit this is a minor advantage especially if you have a ratcheting barrel on the mic that will prevent you from cranking it down too hard. We know (theoretically) that the part distorts, the anvil distorts and the frame bends open, but those are imperceptible outside of the lab- but you CAN get different readings with a mic if you crank it hard vs. barely get the barrel to slip.

Here is a good online article describing gaging pressure.

http://www.mmsonline.com/articles/0300gage.html

http://forums.pelicanparts.com/uploa...1176121213.jpg

As the above photo shows, the particular crank journal measures 56.471mm. The mic was zeroed on a 50mm standard, because it's a 2-3" version so the scale doesn't go all the way to zero. Of course this is a journal that has been reground 0.5mm under.

During the 20 minutes or so it took to measure the crank, the instrument held its zero nicely. At the end of a run of measurements I would put the standard back in and it lined up at 0.000mm once again. Very reassuring.

Of course there is measuring "feel" that has to be developed: when the standard is inserted you should "wring" it against the micrometer anvil to assure that it's square and that there's no foreign matter on it. Some light oil helps in this regard, but don't overdo it.

I see I made a mistake in the measurement of the valve stems above. According to the Fundamentals of Dimensional metrology book, comparators like the Millimess are most accurate when they are pointing at zero. The further from zero they are displaced, the more error is introduced. So later I'm going to check the Millimess using a gage block to zero it out, then another that is .050mm smaller, then one that is .050mm larger. This way I can verify how much error is in the full-scale swing of the indicator.

When I say I made a mistake, I should have used a gage block stack that was exactly equal to the valve stem diameter, instead of using one bigger and relying on linearity of the scale out at 75% of the indicator's range of motion to give me the measurement.

The rubber is meeting the road finally: the heads are going out and these measurements are required to determine whether the valves are usable, or whether new ones are required at $135 each. A reading of 8.93 when the low limit is 8.938 only causes frustration.

Hey John

Nice charts!!

Hey! You posted before I could give you full credit. :) I added a couple things like the mean and fiddled with the code a little bit. But here goes: "A big shout-out to Tom Butler for creating a spreadsheet that allows you to visualize the specifications vs. the factory limits! Thanks Tom!

Let me know if that's too syrupy. :) In any event, the crank is coming in below the new specs but above the wear limit, and the values are pretty well centered around the mean. So it's off to be magnafluxed, and if it passes that test, a micropolish and balance.

I'm sure that grinding 0.50mm off the rod journals removed the hardening, but 20,000 miles and 25 years haven't hurt things much, unless the journals were barely touched, which doesn't seem likely given how tight the measurements were to a line EXACTLY 0.50mm less than the factory low limit.

Noah: here's the Cliffs Notes version:

1) When you rebuild an engine you should measure all the parts to make sure they are within factory specifications. You don't want to put worn-out parts back in the engine unless you are prepared to run full-page ads in Panorama advertising as much;

2) In order to measure repeatably, you need the proper tools and experience. In my own opinion, the measuring instruments should have a resolution that matches that of the factory specifications. The good news is that these aren't very expensive, particularly if buying used.

On the experience side, the only way you can learn is by doing it, but it helps to have somebody who's done it before to guide you. That's what this thread is all about, an effort to get all the guys with machining and rebuilding experience to share it with the novices like me.

I tried to condense that into one sentence but my fingers got tired. . . :)

Love the graphs. Sure makes it easy to read compared with a list of numbers like I made.

quote: "You don't want to put worn-out parts back in the engine unless you are prepared to run full-page ads in Panorama advertising as much;"

That is Classic!!!!

Why put worn out parts in an engine when MM will do it for you?

From ignorance, not malice: Is the valve length tolerance such that that set-up is required? And what is your zero reference tool?