ABEC 7 spindle and housing tolerance clarification sought

[rhb]

I was reading the "fits and tolerances" section of "Machinery's Handbook" today and it appears I've been interpreting what I've read incorrectly. To wit, I've been interpreting bearing tolerances as spindle and housing tolerances.

However, I'm unsure about the nomenclature. I *think* that for a small precision lathe spindle I want an LN1 fit. I *think* that is a light load fit. Looking at "Table 12. ANSI Standard Interference Location Fits ANSI B4.1-1967 (R2009)", p 637 in the 30th ed, I read the table as stating that the housing for a 25 x 52 x 15 bearing should be 0 to +0.7 thou over 52 mm and the spindle should be +0.5 to +1.0 thou over 25 mm.

Am I interpreting the table and fit class correctly?

Thanks,
Reg

 

[Bakafish]

I know you don't want to hear this, and starting out I was in the same boat as you are in trying to make my small lathe a precision instrument. The problem is that you are fighting basic material science. You just don't have the rigidity for any of this to matter, and you already have all the tools to prove this to yourself if you just took a minute to do so. The deflection that you can make with the pressure from your fingers is far greater than the tolerances that you are chasing. It sucks, I wish it weren't so, but just try it. You can deflect the chuck, you can deflect the bed, you can deflect the cross slide and the tailstock. You can twist it like a pretzel between your hands. 20-30 microns? More? The forces of the tooling is way higher than your finger. You wouldn't think that lump of cast iron would do that, it feels solid when you lug it around, but it is about an order of magnitude too small to even start on the path to real rigidity.

If you think, "Yeah it moves, but I'm starting with a super tight tolerance, so the parts will come out better than if it wasn't." Isn't at all true, as the micron or two you gained by ultra precision bearings (assuming you managed to assemble and pre-load them effectively, which is harder than you seem to think based on your test rod theory) is totally overwhelmed by the relatively massive errors introduced by the lack of rigidity of the machine.

People get wrapped up in their hobbies, I get it. I've seen grown men spend thousands of dollars on silver wire power cords for their mono-block amplifiers. And equal amounts trying to shave grams off of bicycles, when they had the physique of a walrus. Don't be that guy.

A healthy small lathe can still make great parts with good standard precision bearings. It can even make precision parts with luck, knowledge and enough effort. It will never be a precision machine though, and you are wasting your time and money trying to make it one.

 

[rhb]

I looked in "Essential Concepts of Bearing Technology" by Harris & Kotzalas. The chapter on fits and tolerances has tables on CD, but it only lists ABEC/RBEC 1. It gives +.0004 as Locational 2 for the shaft and -0.0002" as Transitional 8.5. And the reverse for the housing bore.

Not quite as gnarly as I'd initially thought, but pretty close. But still uncertain whether I'm interpreting the tables correctly.

Edit:

Harris & Kotzalas give all the equations to calculate the internal clearances of the bearings. I *can* do that, but it *really* seems overkill for something as mundane as this. I don't have large forces nor temperature changes. And I'm not trying to get less than one tenth TIR at 12" from the spindle face under all operating conditions. 3-6 tenths at 12" will be good enough. That's within the "toolroom" specs in Connelly. What's not acceptable is the 0.0015" TIR at 6" I have now.

 

[rhb]

After several hours of digging through catalogs I finally found an unambiguous statement that for ABEC 7 bearings the shaft and housing tolerances are the bearing ID and OD tolerances. So it is as ugly as I thought at first. 0 to 3 tenths oversize for the housing and 0 to 2 tenths undersize for the spindle. "Line to line" contact.

 

[CarbideBob]

I am reminded of a comment an old engineer told me when a new kid showed up and my life experiences.
"Look, he has the desk covered with books and is making spreadsheets and simulations on all the formulas.
Eventually he will learn this is the real world which is more like predicting the weather and full of butterflies."

3-6 tenths controlled TIR all along from face to 12 inches out will be a challenge on a one off build no matter how big the iron.
One spot checked is easy, end to end a different story.

Most spindle bearings are clamped onto a shaft and into their housing from the sides so .005 under/over has no effect on TIR once the chuck is cut/ground to the bearings.
The super tight tolerance is so that the pair will be aligned with each other.

Bakafish is spot on about rigidity.
Got a $680,000 18,000 lb grinder and it was very nice. Parts from one person often showed a higher variation.
Checked his part cleaning and loading and all sorts of stuff.
Turned out he would sometimes lean against the enclosure watching the part cycle.

"Drink deep from the well of knowledge as it is drinking that drunkens us it is largely drinking more that sobers us."
The OP is that new engineer still thinking that it should work the way the books say and the world is perfect.

I assume the OP knows the working conditions needed for spindle and bearing assembly at these numbers and the requirements for ultra filtered grease, lint prevention, skin cells, smoke particles and such.
Also that he knows that any bearing nut MUST be hand scraped in or at least checked at fit point tightness it the threads are not precision ground on both.
Bob

 

[rhb]

The attachment is from the Torrington service catalog. Not sure how I missed it given the amount of time I spent reading it.

Deflection of even a 1" piece of brass will be a lot more than a 2.5" x 3" piece of cast iron. And both will deflect much less than the 0.0015" TIR 2" from the spindle face with the OEM ABEC 1 30205 taper roller bearings.

Old engineers have repeated the same few years of experience for a long time. So they know a lot about that one job. Research scientists do things one time and move on to another problem. Any research project that doesn't involve failing many times before finding the answer is pretty trivial.

 

[CarbideBob]

Giggle.
You have all the answers and are sure of it so why come here?
You have made some reasonable points here and in other threads but you are looking rather sad.
I think you can read a lot and are not understanding how it applies. This is sad but is often.
Perhaps the Zone is a better board for you to stretch your thoughts and get feedback.
Bob

 

[rhb]

I ask questions when I don't know. If I don't get an answer I keep digging. If I find the answer, old school (pre WWW) etiquette dictates I state that I've found it and the answer.

Do you have credible evidence that what I have stated is incorrect? If so, please present it.

I've had PM members tell me that taper roller bearings can't take thrust loads and similar nonsense. Opinions are like watches, everyone has one, but none agree. I prefer peer validated facts.

It is a significant failing, but I don't suffer fools quietly. So, do us both a favor. Ignore my posts in the future.

 

[MCritchley]

rhb, what are you building?

Are you using a spring preloaded outer race somewhere?

 

[John Garner]

Take a look at the first Example figure. It looks like something's not right to me.

 

[rhb]

John Garner

"Take a look at the first Example figure. It looks like something's not right to me."

Sorry, but I can't figure out how this forum does quotes.

What doesn't look right? The scan from the Torrington Service Catalog? The entire purpose of this thread was to get help finding that. In the end I found it myself.

I'm converting an ABEC 1 taper roller bearing machine to use ABEC 7 angular contact bearings.

I've never done this before and so have been a bit uncertain about how to do it. The OEM bearing runout is unacceptable, ABEC 7 direct replacements are obscenely expensive and the spindle loads don't need them. Expected jobs are 1/2" stock extending 1/2" - 1" from a collet with tenth tolerances on the work. Primarily <10 GHz electronics connector gauges.

So far, the advice I've gotten from trusted sources is to make inside and outside laps and lap the spindle and housing to the 1-3 tenths required for a proper fit to ABEC 7 bearings. that seems pretty straight forward and easy to do.

 

[Milland]

So far, the advice I've gotten from trusted sources is to make inside and outside laps and lap the spindle and housing to the 1-3 tenths required for a proper fit to ABEC 7 bearings. that seems pretty straight forward and easy to do.

Well, if it turns out to be easy I hope you post your method...

 

[Bakafish]

The OEM bearing runout is unacceptable

Facts not in evidence... and even if true, have you tried just high quality, inexpensive, non-precision replacements first? I say this with the knowledge that you are dead set on your plan, facts and common sense be damned.

The thing is, normally I would love discussing precision and these sort of engineering challenges. But you never seem to clearly document the starting point or the actual problem you are trying to solve. Instead you shortcut to an ill defined solution that you've already decided on and ask for input on some tangental aspects of that fait accompli, dropping terminology that feels previously unacquainted with your mouth.

There seems little point to this performative posting and although we all (perhaps myself more than others) like to demonstrate our knowledge and show off our treasures, this seems different somehow. It's discomforting. It feels like someone who is clearly not a doctor asking how to do a complex medical procedure, and you are thinking to yourself, "What possible good can come of this?"

I know, "Just ignore me."
I'm sure once the spectacle of the car crash of whatever it is you are doing wears off I will, but maybe do a little introspection and see if you are really on the right track here, not about these bearings, but in general.

 

[TGTool]

Experience keeps a hard school, but some will learn in no other.

 

[jim rozen]

"I'm converting an ABEC 1 taper roller bearing machine to use ABEC 7 angular contact bearings.
I've never done this before and so have been a bit uncertain about how to do it. The OEM bearing runout is unacceptable, ABEC 7 direct replacements are obscenely expensive...."

Wait, what? If the abec7s are too expensive, then how are you retrofitting abec7s into the spindle? If you purchase two random abec7 angular contact bearings you'll need to set the preload spacers to get the preload right. Otherwise your new bearings will either a) not work or b) get trashed in short order. If doing it this way you'll need to have gaging that goes down around the micron level.

 

[Peter from Holland]

If it is originally a abec 1 machine with tolerances for abec 1 How do you measure the alignment accuracy of the 2 bores Alignment is very importand as well as roundness For tolerances I would check with the fabricators of those bearings I use SKF They used to have a book on precision bearings and their tolerances
Peter

 

[JST]

................................

I'm converting an ABEC 1 taper roller bearing machine to use ABEC 7 angular contact bearings.

I've never done this before and so have been a bit uncertain about how to do it. The OEM bearing runout is unacceptable, ABEC 7 direct replacements are obscenely expensive and the spindle loads don't need them. Expected jobs are 1/2" stock extending 1/2" - 1" from a collet with tenth tolerances on the work. Primarily <10 GHz electronics connector gauges.

..........................................

Just one small question here.......

The parts of your machine are made to be as good as they need to be for an ABEC 1 tolerance.... i.e. all errors about the same magnitude, spindle manufacturing for concentricity, housing parallelism, general stiffness and construction, dial accuracy, screw accuracy, etc, etc.

OK, so into this you put much better bearings. How does that fix all the other reasonably-certain-to-be-present errors?

If it does not, then all you have done is something similar to putting 250mph rated racing tires on a Honda Civic. It won't go faster as a result, it won't corner better, etc, etc.

If you DID fix all those other issues, you'd have to just "jack up the logo and put a new car underneath".

As with the car, so with the lathe.

If you want the machine to do better, the first thing you need to do is to quantify the problem.... "how bad is it?". If it is no worse than the general tolerances of the parts, then you have nothing to improve. (unless you improve everything, example; getting a Schaublin)

If it IS worse, you may have stackup of tolerances in one direction, and making positional changes on some parts may fix it. But you need to then test portions of the machine to find out where the excess error is coming in, to know what to do.

Perfectly possible that removing one bearing and shifting it relative to the other bearing before replacing might actually make an improvement. We do not know that.

You cannot know what to do until you have all the information about how it works now. And not just total runout, but runout at each end of the spindle, clocked vs some common reference, as well as several other appropriate measurements.

The solution comes from a careful assessment of the problem. You get nowhere fast, at large expense, by "shotgunning" the problem with new and expensive parts (that's for auto dealer mechanics). Usually you merely chase your tail, while the real problem goes unrecognized.

Would using a 4 jaw chuck and an indicator not solve many of your problems?

 

[Bakafish]

Should any of us take this seriously? He's never going to do any of this. In the other thread he admitted "I don't have a proper reference to test the squareness", so he apparently doesn't even possess a trustworthy square. And he wants us to believe he's capable of delving into reengineering the bearings of a spindle, swapping the existing conical rollers for some unknown Chinese angular contact bearings he got a "deal' on? So far he has refused to disclose the part numbers of any of these bearings, I wonder why?

He's now claiming he's going to relap this to higher tolerances, but even ignoring the difficulty of doing that correctly, that's a subtractive process that would require the current bearing seats to be tighter than the tolerance required. What are the odds of that?

Part numbers, measurements and methods or GTFO.

 

[TGTool]

Should any of us take this seriously? He's never going to do any of this. In the other thread he admitted "I don't have a proper reference to test the squareness", so he apparently doesn't even possess a trustworthy square. <snip>

If he has squares, he can verify squareness without a second reference but if he doesn't know how to do that it doesn't bode well for other measuring either

He's now claiming he's going to relap this to higher tolerances, but even ignoring the difficulty of doing that correctly, that's a subtractive process that would require the current bearing seats to be tighter than the tolerance required. What are the odds of that?

There's always Loctite.

 

[memphisjed]

... that's a subtractive process that would require the current bearing seats to be tighter than the tolerance required. What are the odds of that?

Jb weld that ish like hot sauce.