"Jim Wilkins" wrote in message
...
On Nov 28, 11:19 pm, "Ed Huntress" wrote:
"Jim Wilkins" wrote in message
...
I'm sure there are many possible solutions. ...
Ed Huntress

The problem is making the first headstock spindle without another
lathe. Once you have it you can machine a better one.

In my case someone would likely offer me a good lathe cheap *after*
seeing the one I struggled to make.

jsw

I wouldn't attempt this without access to another lathe, unless someone made
it a group or club project and made the necessary machined parts available
for a reasonable price.

--
Ed Huntress

On Nov 29, 9:20*am, "Ed Huntress" wrote:
"Jim Wilkins" wrote in message

...
On Nov 28, 11:19 pm, "Ed Huntress" wrote:

"Jim Wilkins" wrote in message
...
I'm sure there are many possible solutions. ...
Ed Huntress
The problem is making the first headstock spindle without another
lathe. Once you have it you can machine a better one.

In my case someone would likely offer me a good lathe cheap *after*
seeing the one I struggled to make.

jsw

I wouldn't attempt this without access to another lathe, unless someone made
it a group or club project and made the necessary machined parts available
for a reasonable price.

--
Ed Huntress

I think it could be achieved starting with a simple machine that's
adjusted into place. For example my dead-center lathe could bore the
headstock pipe for your cast concrete one. Slide the headstock down
the ways over a long fixed boring bar.

The practical application is temporary oversized equipment to
recondition old worn machines. I'd like to rig up a milling head with
enough X and Y travel to clean up the ways of my surface grinder. It
only needs to slide along one axis while cutting, and can rest on
parallels for the other axis.

jsw

On Sat, 28 Nov 2009 23:08:15 -0500, "Ed Huntress"
wrote:



The HLVH layout is extreme, but some space between the front bearing
pair is not unusual. I'm looking at a cross section of a 10EE
headstock and it appears the spacers are about 2-1/2" long. The
support at the tail is an unspaced pair of angular contact bearings.
Top speed of an EE is about 1000RPM higher than an HLVH.

The bearings at the nose of a Bridgeport spindle are separated perhaps
1-1/2". In this case there's a single deep groove bearing at the top
of the quill. A BP spindle running at top speed gets much hotter than
an HLVH.

Grinder spindles typically have the bearings pairs mounted directly
back-to-back.

--
Ned Simmons

That generally agrees with what I've seen, although I haven't had any
spindles apart for a few decades. Thirty degrees F produces about 0.001 in.
of growth in about 5 inches of length.

One thing that we've neglected here is that this is, at least
potentially, a self compensating system. As the spindle warms up and
expands the preload drops, reducing the heat generated in the bearing.
The problem is coming up with a design that will settle at a
reasonable equilibrium under the normal range of operating speeds and
loadings. It seems Hardinge has been able to do this.


That shouldn't be a problem for
ordinary bearings, which are less that perfect all around; there's a little
room for elastic compression.

As it's been explained to me, the problem becomes more critical as the
bearing class goes up. The Class 9 bearings in a Hardinge HLVH must be very
touchy in terms of the growth they'll tolerate.

I've never heard that and find it hard to swallow. If a lower class
bearing has imperfections that allow it to deflect more easily, that
implies there are areas of high stress that would be more sensitive to
damage. I can see where a bearing with more accurate geometry might be
stiffer as a result of better stress distribution, but I'd expect that
would make it more robust, not less.

--
Ned Simmons

"Ned Simmons" wrote in message
...
On Sat, 28 Nov 2009 23:08:15 -0500, "Ed Huntress"
wrote:



The HLVH layout is extreme, but some space between the front bearing
pair is not unusual. I'm looking at a cross section of a 10EE
headstock and it appears the spacers are about 2-1/2" long. The
support at the tail is an unspaced pair of angular contact bearings.
Top speed of an EE is about 1000RPM higher than an HLVH.

The bearings at the nose of a Bridgeport spindle are separated perhaps
1-1/2". In this case there's a single deep groove bearing at the top
of the quill. A BP spindle running at top speed gets much hotter than
an HLVH.

Grinder spindles typically have the bearings pairs mounted directly
back-to-back.

--
Ned Simmons

That generally agrees with what I've seen, although I haven't had any
spindles apart for a few decades. Thirty degrees F produces about 0.001
in.
of growth in about 5 inches of length.

One thing that we've neglected here is that this is, at least
potentially, a self compensating system. As the spindle warms up and
expands the preload drops, reducing the heat generated in the bearing.
The problem is coming up with a design that will settle at a
reasonable equilibrium under the normal range of operating speeds and
loadings. It seems Hardinge has been able to do this.


That shouldn't be a problem for
ordinary bearings, which are less that perfect all around; there's a
little
room for elastic compression.

As it's been explained to me, the problem becomes more critical as the
bearing class goes up. The Class 9 bearings in a Hardinge HLVH must be
very
touchy in terms of the growth they'll tolerate.

I've never heard that and find it hard to swallow. If a lower class
bearing has imperfections that allow it to deflect more easily, that
implies there are areas of high stress that would be more sensitive to
damage.

There are. That's why they don't last as long if both types are properly
applied. Unless it's overloaded, a Class 9 bearing will run until hell
freezes over, while a lesser bearing will eventually spall and fail. That
assumes that they aren't abused and brinelled, or otherwise damaged.

I can see where a bearing with more accurate geometry might be
stiffer as a result of better stress distribution...

Yes.

but I'd expect that
would make it more robust, not less.

It will last longer in proper service. It also is more susceptible to
overloading from thermal growth, misalignment, etc. If you're going to use
Class 9, everything in the setup had better be perfect. If it is, it will
outlast a lesser-quality bearing.

--
Ed Huntress

On Sun, 29 Nov 2009 16:58:29 -0500, "Ed Huntress"
wrote:


"Ned Simmons" wrote in message
.. .
On Sat, 28 Nov 2009 23:08:15 -0500, "Ed Huntress"
wrote:




That shouldn't be a problem for
ordinary bearings, which are less that perfect all around; there's a
little
room for elastic compression.

As it's been explained to me, the problem becomes more critical as the
bearing class goes up. The Class 9 bearings in a Hardinge HLVH must be
very
touchy in terms of the growth they'll tolerate.

I've never heard that and find it hard to swallow. If a lower class
bearing has imperfections that allow it to deflect more easily, that
implies there are areas of high stress that would be more sensitive to
damage.

There are. That's why they don't last as long if both types are properly
applied. Unless it's overloaded, a Class 9 bearing will run until hell
freezes over, while a lesser bearing will eventually spall and fail. That
assumes that they aren't abused and brinelled, or otherwise damaged.

Not quite. Unless very lightly loaded, all bearings will eventually
fatigue and fail by spalling. Very light loads will cause problems
related to skidding of the balls.



I can see where a bearing with more accurate geometry might be
stiffer as a result of better stress distribution...

Yes.

but I'd expect that
would make it more robust, not less.

It will last longer in proper service. It also is more susceptible to
overloading from thermal growth, misalignment, etc. If you're going to use
Class 9, everything in the setup had better be perfect. If it is, it will
outlast a lesser-quality bearing.

This doesn't make sense to me. If a high ABEC class bearing will
outlast a lower class bearing under ideal conditions, what's the
mechanism that will cause it to fail sooner in a less than ideal
installation?

--
Ned Simmons

Ned Simmons wrote:

I can see where a bearing with more accurate geometry might be
stiffer as a result of better stress distribution...
Yes.

but I'd expect that
would make it more robust, not less.
It will last longer in proper service. It also is more susceptible to
overloading from thermal growth, misalignment, etc. If you're going to use
Class 9, everything in the setup had better be perfect. If it is, it will
outlast a lesser-quality bearing.

This doesn't make sense to me. If a high ABEC class bearing will
outlast a lower class bearing under ideal conditions, what's the
mechanism that will cause it to fail sooner in a less than ideal
installation?

Well, he didn't say fail catastrophically. It could be
that it degrades in performance and doesn't meet class
9 specs any longer.

On Sun, 29 Nov 2009 21:17:00 -0800, Jim Stewart
wrote:

Ned Simmons wrote:

I can see where a bearing with more accurate geometry might be
stiffer as a result of better stress distribution...
Yes.

but I'd expect that
would make it more robust, not less.
It will last longer in proper service. It also is more susceptible to
overloading from thermal growth, misalignment, etc. If you're going to use
Class 9, everything in the setup had better be perfect. If it is, it will
outlast a lesser-quality bearing.

This doesn't make sense to me. If a high ABEC class bearing will
outlast a lower class bearing under ideal conditions, what's the
mechanism that will cause it to fail sooner in a less than ideal
installation?

Well, he didn't say fail catastrophically. It could be
that it degrades in performance and doesn't meet class
9 specs any longer.

Perhaps, but that's not how I read it. Clearly it doesn't make sense
economically to stick an expensive bearing in an inferior gadget, but
I can't come up with any reason it wouldn't last as long as an ABEC 1
bearing.

--
Ned Simmons

"Ned Simmons" wrote in message
...
On Sun, 29 Nov 2009 16:58:29 -0500, "Ed Huntress"
wrote:


"Ned Simmons" wrote in message
. ..
On Sat, 28 Nov 2009 23:08:15 -0500, "Ed Huntress"
wrote:




That shouldn't be a problem for
ordinary bearings, which are less that perfect all around; there's a
little
room for elastic compression.

As it's been explained to me, the problem becomes more critical as the
bearing class goes up. The Class 9 bearings in a Hardinge HLVH must be
very
touchy in terms of the growth they'll tolerate.

I've never heard that and find it hard to swallow. If a lower class
bearing has imperfections that allow it to deflect more easily, that
implies there are areas of high stress that would be more sensitive to
damage.

There are. That's why they don't last as long if both types are properly
applied. Unless it's overloaded, a Class 9 bearing will run until hell
freezes over, while a lesser bearing will eventually spall and fail. That
assumes that they aren't abused and brinelled, or otherwise damaged.

Not quite. Unless very lightly loaded, all bearings will eventually
fatigue and fail by spalling. Very light loads will cause problems
related to skidding of the balls.



I can see where a bearing with more accurate geometry might be
stiffer as a result of better stress distribution...

Yes.

but I'd expect that
would make it more robust, not less.

It will last longer in proper service. It also is more susceptible to
overloading from thermal growth, misalignment, etc. If you're going to use
Class 9, everything in the setup had better be perfect. If it is, it will
outlast a lesser-quality bearing.

This doesn't make sense to me. If a high ABEC class bearing will
outlast a lower class bearing under ideal conditions, what's the
mechanism that will cause it to fail sooner in a less than ideal
installation?

--
Ned Simmons

There's nothing much to "crush." A given amount of displacement of one race
relative to the other can produce a substantially higher preload in a
higher-class bearing. The difference may be slight, but small variations in
the percentage of yield strength that a bearing is subject to will produce
large variations in its fatigue life -- in other words, the time it takes
for the bearing balls or the race to spall.

--
Ed Huntress

"Jim Wilkins" wrote in message
...
On Nov 29, 9:20 am, "Ed Huntress" wrote:
"Jim Wilkins" wrote in message

...
On Nov 28, 11:19 pm, "Ed Huntress" wrote:

"Jim Wilkins" wrote in message
...
I'm sure there are many possible solutions. ...
Ed Huntress
The problem is making the first headstock spindle without another
lathe. Once you have it you can machine a better one.

In my case someone would likely offer me a good lathe cheap *after*
seeing the one I struggled to make.

jsw

I wouldn't attempt this without access to another lathe, unless someone
made
it a group or club project and made the necessary machined parts available
for a reasonable price.

--
Ed Huntress

I think it could be achieved starting with a simple machine that's
adjusted into place. For example my dead-center lathe could bore the
headstock pipe for your cast concrete one. Slide the headstock down
the ways over a long fixed boring bar.

The practical application is temporary oversized equipment to
recondition old worn machines. I'd like to rig up a milling head with
enough X and Y travel to clean up the ways of my surface grinder. It
only needs to slide along one axis while cutting, and can rest on
parallels for the other axis.

jsw

Maybe. I've thought about some of this in the past, and I don't see being
able to bore the headstock from the bed without a pretty fancy, and strong,
temporary boring rig. If it was just a straight bore, maybe. But I've
changed my thinking on that to include a tube bored on another lathe; a
careful setup to cast it in place when the headstock is cast; and a
honing/lapping rig mounted on the new lathe's bedways, rather than boring,
to finish it off. The loads will be much less and you won't need a
controlled feedrate. The whole affair would be simpler.

But I'm not saying I have all the answers for this. It's just a lot of
thinking and speculating on my part. Nor would I want to discourage anyone
else who wants to give it a try. Viva the experiments.

As for finish-machining bedways, you might like to see my idea for a
right-angle grinding head that moves on ordinary, low-accuracy ways
(Thompson round ways), with two axes of stepping motors that respond to
optical drives that follow a laser beam -- or maybe you wouldn't want to see
it, come to think of it. It was pretty rough and crude, but there's an idea
there. (No, I don't have any CAD files of it. I sketched it around 30 years
ago.)

--
Ed Huntress

On Mon, 30 Nov 2009 09:59:17 -0500, "Ed Huntress"
wrote:


"Ned Simmons" wrote in message
.. .



This doesn't make sense to me. If a high ABEC class bearing will
outlast a lower class bearing under ideal conditions, what's the
mechanism that will cause it to fail sooner in a less than ideal
installation?

--
Ned Simmons

There's nothing much to "crush." A given amount of displacement of one race
relative to the other can produce a substantially higher preload in a
higher-class bearing. The difference may be slight, but small variations in
the percentage of yield strength that a bearing is subject to will produce
large variations in its fatigue life -- in other words, the time it takes
for the bearing balls or the race to spall.

Right. But I understood you to say that the imperfections on the
contact surfaces of a low class bearing will cause it to fail earlier
than a high class bearing when they are both properly mounted. On the
other hand, you also seem to be saying that those imperfections are
*protecting* the low class bearing in a poor mounting.

If you imagine looking at only a very small patch on the bearing race,
there's no way to tell whether the bearing is mounted properly or not,
all you can determine is the contact pressure as a ball passes. If
there are imperfections in the surface of the low class bearing's
race, there will be local peaks in the contact stress, regardless of
the how the bearing is mounted.

Re the temperature compensation business, I was looking thru some of
my references and found this:
http://tinyurl.com/ya7hvm4

There's about a half page missing, but the jist of it is there.

--
Ned Simmons

"Ned Simmons" wrote in message
news
On Mon, 30 Nov 2009 09:59:17 -0500, "Ed Huntress"
wrote:


"Ned Simmons" wrote in message
. ..



This doesn't make sense to me. If a high ABEC class bearing will
outlast a lower class bearing under ideal conditions, what's the
mechanism that will cause it to fail sooner in a less than ideal
installation?

--
Ned Simmons

There's nothing much to "crush." A given amount of displacement of one
race
relative to the other can produce a substantially higher preload in a
higher-class bearing. The difference may be slight, but small variations
in
the percentage of yield strength that a bearing is subject to will produce
large variations in its fatigue life -- in other words, the time it takes
for the bearing balls or the race to spall.

Right. But I understood you to say that the imperfections on the
contact surfaces of a low class bearing will cause it to fail earlier
than a high class bearing when they are both properly mounted. On the
other hand, you also seem to be saying that those imperfections are
*protecting* the low class bearing in a poor mounting.

Right. Both. Local overloads are what cause a lower-class bearing to fail
sooner than a high-class bearing, given a good mounting. But the presence of
anomalous bumps and so on are also what give it some "crush" room.

A low-grade bearing will fail sooner than a high-class one in a good
mounting. It will fail even sooner in a bad mounting. Depending on the
nature of the "bad" mounting, however, it can last longer than a high-class
bearing in the bad mounting.

Some of the anomalies in a lower-class bearing don't result in premature
spalling of the balls or races; they may just displace locally. That's the
"crush" room.


If you imagine looking at only a very small patch on the bearing race,
there's no way to tell whether the bearing is mounted properly or not,
all you can determine is the contact pressure as a ball passes. If
there are imperfections in the surface of the low class bearing's
race, there will be local peaks in the contact stress, regardless of
the how the bearing is mounted.

Right. But even when they fail, the spalling may not spread around. If the
load on these "points" is very high, they may not spall at all: the points
will just displace plastically, or fracture off from local overload. Then
some of the preload is relieved; the bearing runs loose; but the mean load
on the bearings is reduced.

A high-class bearing, poorly mounted, will subject the balls and/or spots on
the races to continual overload at certain points in their rotation. There's
nothing much to "crush," so nothing will relieve the overload; they'll spall
sooner, depending again on how the mounting is "bad," than the lower-class
bearings.

You know, all of this is from memory, based on explanations derived from
practice and perhaps from theory as well, passed along to me by bearing
specialists many years ago. I had the job of dealing with lubrication issues
when I was at _American Machinist_, and in those days, that meant listening
to some really boring stuff in interviews. g The guys at Timken were
great; they spent hours explaining things to me about bearings in the real
world. That's where I got all this stuff. My memory for these things usually
is Ok, but I'm reconstructing it.

In general, the idea that good bearings can go to pot quicker than
poorer-class bearings in a bad setup, with a poor mounting, is clear in my
memory.


Re the temperature compensation business, I was looking thru some of
my references and found this:
http://tinyurl.com/ya7hvm4

There's about a half page missing, but the jist of it is there.

Well, right through page 542, there's most of the story. I didn't follow the
relative radii between balls and races in detail, but I see the picture.

Finite-element analysis and bench-testing a prototype sound like a good
idea. g I've never touched thermal FIA so I don't know how predictable the
bearing and spindle temperatures are, but Slocum does identify some problems
that have to be tested.

Very interesting, Ned. Thanks.

--
Ed Huntress

On Tue, 1 Dec 2009 01:01:12 -0500, "Ed Huntress"
wrote:


"Ned Simmons" wrote in message
news
On Mon, 30 Nov 2009 09:59:17 -0500, "Ed Huntress"
wrote:


"Ned Simmons" wrote in message
...



This doesn't make sense to me. If a high ABEC class bearing will
outlast a lower class bearing under ideal conditions, what's the
mechanism that will cause it to fail sooner in a less than ideal
installation?

--
Ned Simmons

There's nothing much to "crush." A given amount of displacement of one
race
relative to the other can produce a substantially higher preload in a
higher-class bearing. The difference may be slight, but small variations
in
the percentage of yield strength that a bearing is subject to will produce
large variations in its fatigue life -- in other words, the time it takes
for the bearing balls or the race to spall.

Right. But I understood you to say that the imperfections on the
contact surfaces of a low class bearing will cause it to fail earlier
than a high class bearing when they are both properly mounted. On the
other hand, you also seem to be saying that those imperfections are
*protecting* the low class bearing in a poor mounting.

Right. Both. Local overloads are what cause a lower-class bearing to fail
sooner than a high-class bearing, given a good mounting. But the presence of
anomalous bumps and so on are also what give it some "crush" room.

A low-grade bearing will fail sooner than a high-class one in a good
mounting. It will fail even sooner in a bad mounting. Depending on the
nature of the "bad" mounting, however, it can last longer than a high-class
bearing in the bad mounting.

Some of the anomalies in a lower-class bearing don't result in premature
spalling of the balls or races; they may just displace locally. That's the
"crush" room.


If you imagine looking at only a very small patch on the bearing race,
there's no way to tell whether the bearing is mounted properly or not,
all you can determine is the contact pressure as a ball passes. If
there are imperfections in the surface of the low class bearing's
race, there will be local peaks in the contact stress, regardless of
the how the bearing is mounted.

Right. But even when they fail, the spalling may not spread around. If the
load on these "points" is very high, they may not spall at all: the points
will just displace plastically, or fracture off from local overload. Then
some of the preload is relieved; the bearing runs loose; but the mean load
on the bearings is reduced.

A high-class bearing, poorly mounted, will subject the balls and/or spots on
the races to continual overload at certain points in their rotation. There's
nothing much to "crush," so nothing will relieve the overload; they'll spall
sooner, depending again on how the mounting is "bad," than the lower-class
bearings.

OK, I see what you're saying. But to allow that a bearing that has
permanently deformed to the degree necessary to relieve an overload is
still OK is rather generous. But I suppose if it still runs without
making nasty noises...


You know, all of this is from memory, based on explanations derived from
practice and perhaps from theory as well, passed along to me by bearing
specialists many years ago. I had the job of dealing with lubrication issues
when I was at _American Machinist_, and in those days, that meant listening
to some really boring stuff in interviews. g The guys at Timken were
great; they spent hours explaining things to me about bearings in the real
world. That's where I got all this stuff. My memory for these things usually
is Ok, but I'm reconstructing it.

I'm afraid those days are gone, at least for us peons. As recently as
5 years ago I could still speak directly to someone at SKF or Timken
or Fafnir who was willing to answer questions that weren't covered in
the literature. I recently had a very simple question about the
strength of the cast housings of mounted bearings relative to the load
capacity of the inserts. The know-nothing I spoke to at Fafnir said
he'd try to get an answer by emailing the tech guy, who was in a time
zone remote enough that their workdays didn't ovelap. I never did get
my answer.


In general, the idea that good bearings can go to pot quicker than
poorer-class bearings in a bad setup, with a poor mounting, is clear in my
memory.


Re the temperature compensation business, I was looking thru some of
my references and found this:
http://tinyurl.com/ya7hvm4

There's about a half page missing, but the jist of it is there.

Well, right through page 542, there's most of the story. I didn't follow the
relative radii between balls and races in detail, but I see the picture.

Finite-element analysis and bench-testing a prototype sound like a good
idea. g I've never touched thermal FIA so I don't know how predictable the
bearing and spindle temperatures are, but Slocum does identify some problems
that have to be tested.

Very interesting, Ned. Thanks.

Well, thanks for sticking with me on this. It may not have much
practical value -- I'm not planning on installing $300 bearings where
$8 units will do, but it's good exercise.

Now back to the birthers...

--
Ned Simmons

On Tue, 1 Dec 2009 01:01:12 -0500, "Ed Huntress"
wrote:


"Ned Simmons" wrote in message
news
On Mon, 30 Nov 2009 09:59:17 -0500, "Ed Huntress"
wrote:


"Ned Simmons" wrote in message
...



This doesn't make sense to me. If a high ABEC class bearing will
outlast a lower class bearing under ideal conditions, what's the
mechanism that will cause it to fail sooner in a less than ideal
installation?

--
Ned Simmons

There's nothing much to "crush." A given amount of displacement of one
race
relative to the other can produce a substantially higher preload in a
higher-class bearing. The difference may be slight, but small variations
in
the percentage of yield strength that a bearing is subject to will produce
large variations in its fatigue life -- in other words, the time it takes
for the bearing balls or the race to spall.

Right. But I understood you to say that the imperfections on the
contact surfaces of a low class bearing will cause it to fail earlier
than a high class bearing when they are both properly mounted. On the
other hand, you also seem to be saying that those imperfections are
*protecting* the low class bearing in a poor mounting.

Right. Both. Local overloads are what cause a lower-class bearing to fail
sooner than a high-class bearing, given a good mounting. But the presence of
anomalous bumps and so on are also what give it some "crush" room.

A low-grade bearing will fail sooner than a high-class one in a good
mounting. It will fail even sooner in a bad mounting. Depending on the
nature of the "bad" mounting, however, it can last longer than a high-class
bearing in the bad mounting.

Some of the anomalies in a lower-class bearing don't result in premature
spalling of the balls or races; they may just displace locally. That's the
"crush" room.


If you imagine looking at only a very small patch on the bearing race,
there's no way to tell whether the bearing is mounted properly or not,
all you can determine is the contact pressure as a ball passes. If
there are imperfections in the surface of the low class bearing's
race, there will be local peaks in the contact stress, regardless of
the how the bearing is mounted.

Right. But even when they fail, the spalling may not spread around. If the
load on these "points" is very high, they may not spall at all: the points
will just displace plastically, or fracture off from local overload. Then
some of the preload is relieved; the bearing runs loose; but the mean load
on the bearings is reduced.

A high-class bearing, poorly mounted, will subject the balls and/or spots on
the races to continual overload at certain points in their rotation. There's
nothing much to "crush," so nothing will relieve the overload; they'll spall
sooner, depending again on how the mounting is "bad," than the lower-class
bearings.

OK, I see what you're saying. But to allow that a bearing that has
permanently deformed to the degree necessary to relieve an overload is
still OK is rather generous. But I suppose if it still runs without
making nasty noises...


You know, all of this is from memory, based on explanations derived from
practice and perhaps from theory as well, passed along to me by bearing
specialists many years ago. I had the job of dealing with lubrication issues
when I was at _American Machinist_, and in those days, that meant listening
to some really boring stuff in interviews. g The guys at Timken were
great; they spent hours explaining things to me about bearings in the real
world. That's where I got all this stuff. My memory for these things usually
is Ok, but I'm reconstructing it.

I'm afraid those days are gone, at least for us peons. As recently as
5 years ago I could still speak directly to someone at SKF or Timken
or Fafnir who was willing to answer questions that weren't covered in
the literature. I recently had a very simple question about the
strength of the cast housings of mounted bearings relative to the load
capacity of the inserts. The know-nothing I spoke to at Fafnir said
he'd try to get an answer by emailing the tech guy, who was in a time
zone remote enough that their workdays didn't ovelap. I never did get
my answer.


In general, the idea that good bearings can go to pot quicker than
poorer-class bearings in a bad setup, with a poor mounting, is clear in my
memory.


Re the temperature compensation business, I was looking thru some of
my references and found this:
http://tinyurl.com/ya7hvm4

There's about a half page missing, but the jist of it is there.

Well, right through page 542, there's most of the story. I didn't follow the
relative radii between balls and races in detail, but I see the picture.

Finite-element analysis and bench-testing a prototype sound like a good
idea. g I've never touched thermal FIA so I don't know how predictable the
bearing and spindle temperatures are, but Slocum does identify some problems
that have to be tested.

Very interesting, Ned. Thanks.

Well, thanks for sticking with me on this. It may not have much
practical value -- I'm not planning on installing $300 bearings where
$8 units will do, but it's good exercise.

Now back to the birthers...

--
Ned Simmons

On Dec 2, 5:01*am, Ned Simmons wrote:

Well, thanks for sticking with me on this. It may not have much
practical value -- I'm not planning on installing $300 bearings where
$8 units will do, but it's good exercise.

Now back to the birthers...

--
Ned Simmons

It may not have any value at all. The $8 bearing has a minimum
tolerance. There is no guarantee that it will have characteristics
that Ed says will make it last longer in a poor mounting. The $300
bearings are selected after manufacture from the lot that has the $8
bearings. So if the demand for $300 bearings isn't very high, you may
get a high precision bearing from the $8 bin. Also bearing
manufacturing has improved in the last thirty years. So more of the
$8 bearings are closer to meeting the specs of the $300 bearings.

Granted not all manufacturers have tighter control on their
production.

Dan

On Wed, 2 Dec 2009 06:01:43 -0800 (PST), "
wrote:

On Dec 2, 5:01*am, Ned Simmons wrote:

Well, thanks for sticking with me on this. It may not have much
practical value -- I'm not planning on installing $300 bearings where
$8 units will do, but it's good exercise.

Now back to the birthers...

--
Ned Simmons

It may not have any value at all. The $8 bearing has a minimum
tolerance. There is no guarantee that it will have characteristics
that Ed says will make it last longer in a poor mounting. The $300
bearings are selected after manufacture from the lot that has the $8
bearings.

That may be true of resistors, but not bearings. For example, Barden
makes *only* precision bearings. Precision bearings also have
different cages than the bearing you get if you simply ask for a 6206.

So if the demand for $300 bearings isn't very high, you may
get a high precision bearing from the $8 bin. Also bearing
manufacturing has improved in the last thirty years. So more of the
$8 bearings are closer to meeting the specs of the $300 bearings.

Granted not all manufacturers have tighter control on their
production.

This is true. I was told by an SKF engineer that most of their ABEC 1
deep row bearings will meet ABEC 5 standards, and I've verified this
in a couple cases where I built very low speed spindles with sub-tenth
runout using ABEC 1 bearings. The fallback plan was to replace the
bearings with precision units, but it wasn't necessary.

--
Ned Simmons

On Dec 2, 2:31*pm, Ned Simmons wrote:


That may be true of resistors, but not bearings. For example, Barden
makes *only* precision bearings. Precision bearings also have
different cages than the bearing you get if you simply ask for a 6206.

I think it is true for most manufacturers. It has been sometime since
I read whatever I read. I may have been reading about roller bearing
at the time with the emphasis on using the sizes that are relatively
cheap because they are manufactured for use on semi-trucks.


This is true. I was told by an SKF engineer that most of their ABEC 1
deep row bearings will meet ABEC 5 standards, and I've verified this
in a couple cases where I built very low speed spindles with sub-tenth
runout using ABEC 1 bearings. The fallback plan was to replace the
bearings with precision units, but it wasn't necessary.

So if most meet ABEC 5, what are the chances that some meet ABEC 7 or
9?

Dan
--
Ned Simmons

"Ned Simmons" wrote in message
...
On Tue, 1 Dec 2009 01:01:12 -0500, "Ed Huntress"
wrote:


"Ned Simmons" wrote in message
news
On Mon, 30 Nov 2009 09:59:17 -0500, "Ed Huntress"
wrote:


"Ned Simmons" wrote in message
m...



This doesn't make sense to me. If a high ABEC class bearing will
outlast a lower class bearing under ideal conditions, what's the
mechanism that will cause it to fail sooner in a less than ideal
installation?

--
Ned Simmons

There's nothing much to "crush." A given amount of displacement of one
race
relative to the other can produce a substantially higher preload in a
higher-class bearing. The difference may be slight, but small variations
in
the percentage of yield strength that a bearing is subject to will
produce
large variations in its fatigue life -- in other words, the time it
takes
for the bearing balls or the race to spall.

Right. But I understood you to say that the imperfections on the
contact surfaces of a low class bearing will cause it to fail earlier
than a high class bearing when they are both properly mounted. On the
other hand, you also seem to be saying that those imperfections are
*protecting* the low class bearing in a poor mounting.

Right. Both. Local overloads are what cause a lower-class bearing to fail
sooner than a high-class bearing, given a good mounting. But the presence
of
anomalous bumps and so on are also what give it some "crush" room.

A low-grade bearing will fail sooner than a high-class one in a good
mounting. It will fail even sooner in a bad mounting. Depending on the
nature of the "bad" mounting, however, it can last longer than a
high-class
bearing in the bad mounting.

Some of the anomalies in a lower-class bearing don't result in premature
spalling of the balls or races; they may just displace locally. That's the
"crush" room.


If you imagine looking at only a very small patch on the bearing race,
there's no way to tell whether the bearing is mounted properly or not,
all you can determine is the contact pressure as a ball passes. If
there are imperfections in the surface of the low class bearing's
race, there will be local peaks in the contact stress, regardless of
the how the bearing is mounted.

Right. But even when they fail, the spalling may not spread around. If the
load on these "points" is very high, they may not spall at all: the points
will just displace plastically, or fracture off from local overload. Then
some of the preload is relieved; the bearing runs loose; but the mean load
on the bearings is reduced.

A high-class bearing, poorly mounted, will subject the balls and/or spots
on
the races to continual overload at certain points in their rotation.
There's
nothing much to "crush," so nothing will relieve the overload; they'll
spall
sooner, depending again on how the mounting is "bad," than the lower-class
bearings.

OK, I see what you're saying. But to allow that a bearing that has
permanently deformed to the degree necessary to relieve an overload is
still OK is rather generous. But I suppose if it still runs without
making nasty noises...


You know, all of this is from memory, based on explanations derived from
practice and perhaps from theory as well, passed along to me by bearing
specialists many years ago. I had the job of dealing with lubrication
issues
when I was at _American Machinist_, and in those days, that meant
listening
to some really boring stuff in interviews. g The guys at Timken were
great; they spent hours explaining things to me about bearings in the real
world. That's where I got all this stuff. My memory for these things
usually
is Ok, but I'm reconstructing it.

I'm afraid those days are gone, at least for us peons. As recently as
5 years ago I could still speak directly to someone at SKF or Timken
or Fafnir who was willing to answer questions that weren't covered in
the literature. I recently had a very simple question about the
strength of the cast housings of mounted bearings relative to the load
capacity of the inserts. The know-nothing I spoke to at Fafnir said
he'd try to get an answer by emailing the tech guy, who was in a time
zone remote enough that their workdays didn't ovelap. I never did get
my answer.


In general, the idea that good bearings can go to pot quicker than
poorer-class bearings in a bad setup, with a poor mounting, is clear in my
memory.


Re the temperature compensation business, I was looking thru some of
my references and found this:
http://tinyurl.com/ya7hvm4

There's about a half page missing, but the jist of it is there.

Well, right through page 542, there's most of the story. I didn't follow
the
relative radii between balls and races in detail, but I see the picture.

Finite-element analysis and bench-testing a prototype sound like a good
idea. g I've never touched thermal FIA so I don't know how predictable
the
bearing and spindle temperatures are, but Slocum does identify some
problems
that have to be tested.

Very interesting, Ned. Thanks.

Well, thanks for sticking with me on this. It may not have much
practical value -- I'm not planning on installing $300 bearings where
$8 units will do, but it's good exercise.

I don't know if I've ever touched a $300 bearing. Well, maybe -- I've
handled some all-ceramic bearing sets, and the ones with ceramic races, as
well as ceramic balls, cost like crazy. Or they did.

But anything I'm likely to work on will get along with cheap. In fact, I
kind of like plain bearings...g


Now back to the birthers...


Yes, on to the birthers...

--
Ed Huntress

On Dec 1, 1:01*am, "Ed Huntress" wrote:
...
Right. But even when they fail, the spalling may not spread around. If the
load on these "points" is very high, they may not spall at all: the points
will just displace plastically, or fracture off from local overload. Then
some of the preload is relieved; the bearing runs loose; but the mean load
on the bearings is reduced....
Ed Huntress

Are you suggesting that lower grade bearing components have less
stringent heat treatment?

jsw

Ned Simmons wrote:

This is true. I was told by an SKF engineer that most of their ABEC 1
deep row bearings will meet ABEC 5 standards, and I've verified this
in a couple cases where I built very low speed spindles with sub-tenth
runout using ABEC 1 bearings. The fallback plan was to replace the
bearings with precision units, but it wasn't necessary.

Don't worry. It's probably only a matter of
time before they start relabeling and selling
genuine ABEC 1 bearings from China...

On Wed, 2 Dec 2009 08:02:47 -0800 (PST), "
wrote:

On Dec 2, 2:31*pm, Ned Simmons wrote:


That may be true of resistors, but not bearings. For example, Barden
makes *only* precision bearings. Precision bearings also have
different cages than the bearing you get if you simply ask for a 6206.

I think it is true for most manufacturers. It has been sometime since
I read whatever I read.

I don't know for sure one way or the other. But besides the matter of
the cages I mentioned before, it seems marking would also be a
problem. Marking the part number would certainly have to be done
before final grinding of the assembled bearing, and more likely even
before heat treat of the components, which I assume is done after
rough turning but before the races are ground. That's pretty early in
the process to be selecting components.

I may have been reading about roller bearing
at the time with the emphasis on using the sizes that are relatively
cheap because they are manufactured for use on semi-trucks.


This is true. I was told by an SKF engineer that most of their ABEC 1
deep row bearings will meet ABEC 5 standards, and I've verified this
in a couple cases where I built very low speed spindles with sub-tenth
runout using ABEC 1 bearings. The fallback plan was to replace the
bearings with precision units, but it wasn't necessary.

So if most meet ABEC 5, what are the chances that some meet ABEC 7 or
9?

I wouldn't be surprised if the occasional bearing meets at least some
of the specs. My understanding is that the geometry and finish
requirements to meet electric motor bearing standards, which are for
the most part quietness specs, are the same things required to make a
true running bearing. On the other hand, EM quality does not require
especially tight tolerances on the overall dimensions of the bearing,
so standard bearings may be less likely to exceed their class in that
regard.

Like Ed, I've formed these impressions based on conversations I've had
with various tech support persons over the years, so take them with a
grain of salt.

--
Ned Simmons

On Wed, 2 Dec 2009 11:50:25 -0500, "Ed Huntress"
wrote:


"Ned Simmons" wrote in message



Well, thanks for sticking with me on this. It may not have much
practical value -- I'm not planning on installing $300 bearings where
$8 units will do, but it's good exercise.

I don't know if I've ever touched a $300 bearing. Well, maybe -- I've
handled some all-ceramic bearing sets, and the ones with ceramic races, as
well as ceramic balls, cost like crazy. Or they did.

They still do. But even a pair of Bridgeport spindle bearings costs
close to $300.

--
Ned Simmons

"Ned Simmons" wrote in message
...
On Wed, 2 Dec 2009 11:50:25 -0500, "Ed Huntress"
wrote:


"Ned Simmons" wrote in message



Well, thanks for sticking with me on this. It may not have much
practical value -- I'm not planning on installing $300 bearings where
$8 units will do, but it's good exercise.

I don't know if I've ever touched a $300 bearing. Well, maybe -- I've
handled some all-ceramic bearing sets, and the ones with ceramic races, as
well as ceramic balls, cost like crazy. Or they did.

They still do. But even a pair of Bridgeport spindle bearings costs
close to $300.


Jeez. I'm glad I don't own anything newer than about...oh, 1950. g I need
to buy a new set for my Walker-Turner drill press. You're making me nervous.

--
Ed Huntress

"Jim Wilkins" wrote in message
...
On Dec 1, 1:01 am, "Ed Huntress" wrote:
...
Right. But even when they fail, the spalling may not spread around. If the
load on these "points" is very high, they may not spall at all: the points
will just displace plastically, or fracture off from local overload. Then
some of the preload is relieved; the bearing runs loose; but the mean load
on the bearings is reduced....
Ed Huntress

Are you suggesting that lower grade bearing components have less
stringent heat treatment?

jsw

No, I wasn't suggesting that, Jim. I don't recall the ABEC specifications on
heat treatment, but the point here is that imperfect balls and races will
have high points, where the specific load will exceed the compressive
strength of the metal, and will get squashed down or broken off.

--
Ed Huntress