CarTalk.com Blogs Car Info Our Show Deals Mechanics Files Vehicle Donation

A Comment On Mchine Tolerances

An oft quoted statement is that modern cars have tighter tolerances, machining is much more precise now, etc, etc. As some of the regulars know, I I disagree with that premise. Ultra-precision might be had on a Bugatti Veyron but on a garden variety Ford Taurus, Toyota Camry, or Chevy Malibu, etc that will not happen.
Using crankshaft oil clearances and end play for an example due to that being the starting point for an engine assembly I will state that if one tore down any modern car that had never been started one would find that bearing oil clearances, end play, etc will vary a bit; all within spec, but varying.

I’m in the middle of an engine building SNAFU that I WILL resolve and this involves main bearings from a very reputable manufacturer. I might add that this manufacturer is gargantuan, worldwide, and they actually produce many types of engine parts, gaskets, suspension components, and so on for many different car manufacturers along with the aftermarket.

So here’s my question for those who think modern cars and manufacturing are extremely precise and this is in regards to main bearings although the principle carries over to every other part on the car.
What would you consider the maximum deviation, variance, or tolerance (hundred thousandths, ten thousandths, thousandths, etc) to be when comparing main bearing shells?

Just some food for thought and yes, I’ve omitted the details behind the current SNAFU for the time being. Anyway, I’m just curious about opinons on a +/- deviation.
and will weigh in later. :slight_smile:

The main bearing oil clearance on small block Fords was .0007 to .0030 compared to a Honda 2.2L that was .0005 to .0015. But the dimensional deviation that appears to have been greatly reduced is that of the pistons, piston rings and cylinder walls. The fit and finish of those components is much improved compared to 40 years ago. Filing down oversized rings to get the proper end gap was common long ago and the deviation in piston to cylinder wall clearance from one point to the next was extreme.

Geometric dimensioning and tolerancing has had a big impact on fit issues in many industries, including automobiles. By reducing process variability and using tolerances to tailor the mean size of fitting parts has made ill-fitting components much rarer that they used to be.

My point is not so much about the clearance itself because no matter if all clearances are within spec (be they .0008, 0025, or a mixed bag of the lot) will not affect motor longevity.

My point here is about arriving at those numbers and that involves the manufacturing of the actual parts used to get there.
Staying with the main bearing scenario, does everyone assume that every bearing out of the box is identical and if so, to within what tolerance would you say is acceptable? (Thousandths to hundred thousandths)

I would expect main bearing tolerance on a modern day (the last 10 years) to be within .0001".

I don’t think every bearing shell out of the box is identical. I don’t think the 12 year old kid in China making those is that interested in accuracy. But I hope that they would be within .0005".

I do think that it matters much more now than on the old V8’s from the 60’s.

Can’t say what crank or rod bearing clearance should be; am not an engine designer or tester. There are other factors affecting crank and rod bearing life such as crankshaft material, permissible out of round spec, oil viscosity, oil cleanliness, driver abuse or lack of that, crankpin and rod journal diameters, crank and rod journal width and surface finish, bearing insert design, a few things just off the top of my head. It might be possible that one mfr is minimizing diameters to reduce friction making a tighter clearance among other things such as width and surface finish more critical to maintain life while another may use other means elsewhere to reduce friction to permit a larger crank dia. tolerance variation. Who can say for sure unless they are an experienced engine designer? My VW, by the way, is still good at over 250k miles; has a crank and rod clearance spec of .0012 to .003". Permissible max out of round is .0012".

I recall the bad old days when a “Rod Out” was common in well used up cars from the 1930s and 1940s that were still in use when I was a kid. Ford flathead V8 rod bearings were quite good. They had full floating inserts. I can imagine old Henry Ford instructing his engineers do what it takes to get rid of rod big end bearing failure.

Tolerance variations are a necessary fact of manufacturing life.

I’m not sure what you are getting at, but my education is in manufacturing and much of my work life has been in that field. It looks like you are questioning the consistency of the dimension of the parts.

Many high volume parts are made on multiple machines or in multiple cavity molds. The parts off any given machine or from a specific mold cavity are likely to be extremely consistent from part to part. The use of statistical process control (SPC) techniques map the trends in dimensioning, that is the changes that occur over time due to wear and use and programs in corrections to compensate such as small changes in tool path or scheduled insert changes.

The bigger changes come from the different machines or mold cavities. For example, when I worked in a foundry casting cylinder heads, I was in the finishing (machining) department. One cylinder head we made had 25 different molds. All the heads went down the same transfer line and were machined on the same machines.

You would think that all the machined surfaces would have the same dimension from the reference plane. What happened was that all the heads from the same mold would have the same dimensions, within a couple hundredths of a millimeter, but with up to 20 molds in production at a given time, the heads from the different molds would go down the line in sequence as they came off the carousels, there would be a small variation from head to head due to the mold number, as much as a half millimeter. But every mold 2 for example would be identical to every other mold 2, same for mold 3 and mold 4, but they would be a little different from each other.

The total allowable tolerance was one millimeter (±0.5mm) so the goal was to keep the machine in the middle so all molds were in tolerance.

@keith, I agree that using SPC and GD&T will provide parts of the correct dimensions, and parts that will fit together every time. The two are not the same, as I’m sure you know.

That’s exactly what I’m getting at keith; consistency of the parts.
This deal started out when I noticed that one of the plain main bearing shells had been double-stamped at the tang; which then boogered up part of the bearing surface. An irritant at most because that meant ordering another set of bearings.
Noting a difference in material between the 2 halves of the thrust bearing pair, on a whim I decided to put the micrometer to them.
The thickness of one shell (meaning the part the crank journal rides on) was .0015 thicker than the other shell. As an analogy, think of cutting a 15" and 16" wheel rim into halves and trying to mate the half rounds. When oil clearance is .008 to .0025 how would one ever work that out when the radius is different; and it got worse.

This led to measuring the widths of the thrust bearings at the thrust surfaces. Not only is one half .003 thicker than the other as measured at the middle, both have hourglass shapes. The thinner bearing is .001 thicker on each end and the thicker bearing half is. 002 thicker on each end.
What this means in setting up crankshaft end play is that it won’t stay there very long because one small patch on each end of one shell will be hammered by the crank while none of the rest of the thrust surface on that shell and none at all on the other shell is making any contact at.

So, after returning the bearings the counter guy said he would order 3 more sets; just to make sure.
When I returned after the parts came in it was with mike in hand and all 3 sets were the same as the first.

After a day of fretting I contacted the manufacturer. An email to them led to a request from them for me to call and sort this out. Several calls later, the following info was provided; in short form.
Eventually ending up on hold, they pulled the engineering blueprint and stated that the .0015 was “acceptable”.
The .003 thicker thrust bearing was deemed “somewhat excessive”.
It wasn’t excessive by much because according to the blueprint the “accepted deviation” is .00251, and I personally consider that smaller number to be out of the park.

These are not white box Chinese knock-off parts and this company provides crank bearings by the container load to many different car makers along with countless other engine parts, electrical, suspension parts, gaskets, and so on.
Anyhoo, that’s why I dispute the tighter machine tolerance premise.

Will the engine run and run well with those bearings? Probably. It’s debateable how it would do over the long haul and I simply can’t bring myself to put something together like that after cringing at the micrometers.
A different set and brand of bearings has been ordered so we’ll see how those look when they arrive. :slight_smile:

.00251 accepted deviance? Maybe for an old flathead, but in this day and age?

What are you working on?

To your original question, I was going to say .001 In the world of mass produced car parts, that’s all you are going to get…And in this same universe, that’s all you need…

The engine is an early 90s Ford 5.0.

ok4450- please keep us posted.

I will, and the new different brand bearings should be in next week. No one around stocks or has .010s for some reason.

Oops.

A different reason for the modern cars have tighter tolerances idea is the complex added features run by oil pressure, like variable cam timing.

Non-conforming parts are made by vendors quite often. This is why the end user does Receiving Inspection on a basis dictated by; the criticality of the part, the historical vendor performance and other criteria.

Just because you received non-conforming parts directly from a vendor supporting aftermarket does not mean those poor quality parts would ever see a new production engine. Mistakes do happen but it’s the rare exception not the rule.

Consider the difficulty in verifying the dimensions of a curved part. We have parabolic parts that are super difficult to accurately measure. The part vendor is incapable of doing it and we decided not to fund them to do it because it was economically more sensible for us to just screen them in house. If you bought this part directly from them, you’d be taking on the risk which is about a 93% yield today.

There are 2 problems here: The specs - and how often the specs are met.

As time has gone on, the ability to make parts that comply with the specs has increased - and that’s why the phrase has been used.

The crankshaft has been turned .010 but the crank and oil clearances are almost moot points. The problem is the wide variance in the bearings themselves and as I stated orignally; the 3 additional sets that were ordered showed to be the same.

A few 10 thousandths is not an issue but when one shell is 1.5 thousandths thicker and when the thrust surfaces vary by 3 to 5 thousandths (depending on how it’s looked at) it becomes a serious problem in my opinion.
To be honest, I was a bit surprised when I was told that those numbers are acceptable and as I told them on the phone; I’m not mad at all about this, just puzzled.

If for example a crank journal oil clearance has a spec of .008 to .0016 then how in the world is a consistent figure to be reached when the shell thickness varies by .0015, which is almost the upper limit?
Same with end play, which has a spec of .004 to .008. With only a tiny fraction of the bearing thrust surface available it won’t take long for the crank to beat that down and increase the end play.
If it’s set up at .006 and the upper limit is .0098 then by the time it beats that original .005 down the end play will be at .011.

As to the wavy thrust bearing surface, I was told that when clamped in the block by the bearing cap that the bearing would straighten out. Not buying this for a second I clamped the bearing shell in the block sans crankshaft just out of curiosity and it altered nothing.

Are you certain you’re not misplacing a decimal point? I’ve never heard of cutting a crank journal .010, usually it’s .001, .002 and a max of .003. I’d certainly not be surprised nobody would stock a bearing for a crank cut .010 undersize.

And to address the following:
“If for example a crank journal oil clearance has a spec of .008 to .0016 then how in the world is a consistent figure to be reached when the shell thickness varies by .0015, which is almost the upper limit?”

That looks to me like a misplaced decimal as well. I can’t believe ANY crank bearing has a clearance of .008. I think that spec should read " .0008 to .0016". You’ve got it “.0016 to .0080” and I’m having trouble believing that’s right.

Forgive me if I’m wrong, I have the greatest respect for you on this board, but this looks like a case of misplaced decimal points to me.