When grinding a spindle taper, how do you inspect it?

I have a 50 taper CNC toolroom mill. The horizontal spindle was a little battered from careless operators letting the pull studs hit the ID of the taper when changing tools. Didn’t look terrible, but when I put a test arbor in it I could get a slight runout on the arbor and I could wiggle it slightly, so I believed there was room to improve it.

We set up to grind the taper with a toolpost grinder and a Kurt slide. We’ve done this before on our VMCs, this job was easy because the spindle is horizontal. Now that the taper is ground, the test results with the arbor are not significantly better than before and I’m puzzled.

The arbor is a Parlec, looks new, is 2” diameter and 18” long from the large end of the taper. When Installed, I get .0004” to .0006” TIR at the far end depending on its’ position in the spindle. When repositioning the arbor, the high spot turns with the arbor. With 75 to 100 lbs of effort on the end, I can move it about .0025” each way from relaxed, it always returns to the same place. Before grinding, I had about .0006 to .0012 TIR at the end, and the same amount of flex.

There’s no measurable movement of the spindle at the face, or of the arbor right next to the spindle face. Runout next to the face is not measurable, and movement when pulling on the arbor is not measurable. If I put the indicator holder magnet on the spindle face to test flex I get the exact same readings as I do when mounting it on the table, so the arbor seems to move in the taper.


Drawbar pull is measured at 4400 lbs, right on the mfgr’s spec. When I put a very light coat of Canode blue on the arbor taper with a foam roller and rub it in the spindle taper about 1/8” turn, I get a light but very even mark all over the ID of the taper, and what’s left on the arbor is very even, no points of high contact. I’ve attached a photo. The marks left in the taper are about .001/.002” deep and I don’t really want to grind far enough to remove them all.

[FONT=&quot]So – What results should I expect? I was expecting zero runout on the arbor, and no movement or dramatically reduced movement when pulling on the arbor. Is that too much to expect? I don’t think I’m flexing the arbor that far with only hand pressure. Will a self releasing taper like 3.5”/ ft of a Cat taper hold without movement? Do they seat in tighter under cutting load? The arbor does release with a pop now when extracted, and didn’t before. Should the contact be even from large end to small end, or would you make it slightly tighter on the large end anticipating more wear and load on the large end? How would you inspect the taper angle? I’d love to have some precision balls large enough to use but don’t have any.

[/FONT][FONT=&quot][/FONT]

I think you expect too much (that is, too little) in the deflection department. I don't think you're pulling with 75 lb. For an 18-in-long, 2-in-dia cantilever beam, I get ~0.006 in deflection with 75 lb lateral force at the end. Cantilever deflection is force x length^3/(3 x Young's modulus x I).

Check your Parlec on a set of vee blocks and see if there is any runout.

I know of an incident years ago where they were using a test bar like this one to realign the axis of the machine.
After fighting with the problem for nearly a year, someone decided to check the test bar. It was out .030" out on the end!
Reground the test bar, machine problems went away.

We just had 3 of our K&T #6 horizontal mill tapers reground by Gunn machine out of michigan. His tolerance was .0002 at 10" from spindle. And 80% contact to his checking arbor. I guess you have to figure that .0001 error in the bearing only exaggerates as you project out further. By the way I highly recommend him for anyone needing this work, he travels anywhere and does a great job.
Chris German

With the high spot following the arbor it does sound like you should check it for accuracy, perhaps both on centers and by v-block, see if anything looks off. I'm also curious what the accuracy of the slide is, even with reasonably even coating of the blue indicating it's pretty good. Perhaps you can set up a tenths test indicator at the end of the slide, then run it up and down the taper and see if it moves. Rotate the taper a few times and recheck, then move the test indicator up or down the slide and retest (this prevents having a slide error match the ground surface). If everything looks good it points to the arbor having an issue.

I agree with what's been about checking out you test bar. If the high spot stays at the same point on the test bar, that points to runout in the bar. As far as checking the taper, I prefer to measure a taper over gaging it with a mating part. This isn't always easy, especially with internal tapers. One way is with different size gage balls, you can depth mic down from the face and trig out the angle pretty precisely, if you have two balls that will work.(yeah, I know) I've used a shallow diameter gage on larger diameters. Take one measurement flat against the face and then compare it to a measurement taken with 2.000" of gage blocks under the rail of the gage. The reason I don't like gaging with a plug only is because the results can sometimes be deceiving. Its a qualitative test and the results depend more upon the technique of the person doing it than the amount of error actually present.
I can't believe how bad the taper was dinged up to start with. I've run plenty of well beaten machines but WOW! Is it that the spindle is dead soft or was somebody throwing tools in from across the room or something? .001-.002 deep dents, that takes some doing!

Derek Smalls said:

I can't believe how bad the taper was dinged up to start with. I've run plenty of well beaten machines but WOW! Is it that the spindle is dead soft or was somebody throwing tools in from across the room or something? .001-.002 deep dents, that takes some doing!

Click to expand...

That was my thought too - if I had operators doing that bad a job of seating toolholders they'd be out the door pretty quickly.

Derek Smalls said:

.....I can't believe how bad the taper was dinged up to start with. I've run plenty of well beaten machines but WOW! Is it that the spindle is dead soft or was somebody throwing tools in from across the room or something? .001-.002 deep dents, that takes some doing!

Click to expand...

Those dings look like damage from trash, shavings, etc, from not cleaning the tapers of the tool shanks often.

Just a couple of quick tests to determine if the run-out is due to the spindle/taper or is a test bar problem.
First, mount the test bar and mark the high position(s) with a sharpie on the test bar. Take it out and mount it rotated 180°. If you read the high spot in the same place, the problem is the test bar. If it's in a different spot, the problem is likely in the taper. In this case, I'd suggest mounting and removing the test bar a few times and see where the high reading lands. If it is at a random position, it would rather indicate that the angle of the taper is somehow off and the taper on the bar cannot engage the spindle in a consistent manner (I doubt that this is the case). If the position is consistent and the end of the taper has a flange, I'd bet that you have some raised dent/burr on the face of the spindle contacting the flange of the test bar (a light cleanup of the face of the spindle wouldn't harm anyhow).

Paolo

The high spot moves with the bar. I just wondered that a Parlec bar could be out that far. I'm not certain that I have good enough gaging here to check the taper on the bar. There's a center in the straight end but the tapered end is threaded for a pull stud, I don't have a good way to hold it that doesn't involve reduced accuracy itself. I did chuck the straight bar in a 4 jaw a while back and indicate the length, it appeared perfectly straight with a .00005 indicator. Perhaps I'll send it back to Parlec for exam.

I ground the vertical spindle also just to give it a fresh start. The vertical taper is as nice as the horizontal is rough. When I put the arbor in the vertical and indicate the end, it runs out .0002, and the high moves with the bar. The vertical drawbar pulls 2000 lbs vs the 4400 of the horizontal(both are factory correct specs). I don't see how that can make a difference, but it's about the only difference between the spindles. Here's a photo. You can see a groove about 1/3 in from the big end, the H taper had it too but it got ground out.


We did use the slide to check the taper with a .00005 test indicator, and it looks perfect. The slide travel is only .5" longer than the taper so there isn't much room to vary it's position. I did disassemble the slide for cleaning beforehand and checked all the parts on a granite plate. Then when together I mounted a .00005 indicator on the slide with the slide base on the granite, and checked it's travel by running the indicator along the granite with the slide - no measureable error. I don't like using the tool that created a feature to check the feature so I was looking for another way to verify the surface of the taper. Also it's challenging to get the path of the indicator tangent to the taper which can affect your accuracy.

So we got a nice blue pattern, but the arbor is suspect. I got the same reading on a new 50 NMTB tool shank, I guess the odds are low that 2 tapers from different sources would have identical errors, so I'm going to call the taper finished.

Throwing the tools from across the room is exactly the words we used here to describe the gouges. The hor. spindle is over 6' above the floor on a machine with a broad table, and the pull studs are T shaped with sharp edges. Reaching that high, and that far across with a heavy tool would be tiring and I can easily see even a caring operator making mistakes, but geez - look at the gouges on the face of the spindle where they missed the hole altogether. Was surprising on an otherwise well cared for machine. Apparently that's the consequences of using a machine like this for jobs that have a lot of toolchanges.

I'm bigger than most, but It still takes an effort to get a tool in the spindle without touching it where I shouldn't - trying to figure out a device to guide it in easily.