How much is too much mini lathe spindle run out?

[WobblyHand]

Understood that one would like a very small number.

Background - bought an ER32 chuck mounted on a plate from LMS. I've rotated it several times (only have 3 possibilities due to the holes) and the best TIR on a ground rod is 0.0021". (Rotating by hand, at far end of spindle.) I thought it would be better. Of course the TIR can come from many causes. If I understand correctly, some of the causes might be:

1. The 3/8" rod isn't round
2. The 3/8" collet isn't true
3. The ER32 chuck assembly is machined incorrectly
4. The spindle plate locating feature is machined incorrectly
5. The spindle has run out

Are there other causes? Have I over simplified this? While rotating the spindle by hand I noticed if I push or pull the far end of the spindle (at the gear end) I measure 0.001" difference at the chuck end. (Radial push, not axial.) Is this too much run out? Is there an adjustment to reduce this? Increase the pre-load on the bearing? I've never adjusted the bearing so it wouldn't surprise me if it needed tweaking.

Any guidance? These are garden variety ER32 collets but they shouldn't be this far off, that is if they actually meet their spec.

I have to say, even with 2 thousandths run out, the collet chuck is very nice to use.

 

[cathead]

Maybe put the ER 32 chuck in a four jaw chuck and zero it out.

 

[macardoso]

Understood that one would like a very small number.

Background - bought an ER32 chuck mounted on a plate from LMS. I've rotated it several times (only have 3 possibilities due to the holes) and the best TIR on a ground rod is 0.0021". (Rotating by hand, at far end of spindle.) I thought it would be better. Of course the TIR can come from many causes. If I understand correctly, some of the causes might be:

1. The 3/8" rod isn't round
2. The 3/8" collet isn't true
3. The ER32 chuck assembly is machined incorrectly
4. The spindle plate locating feature is machined incorrectly
5. The spindle has run out

Are there other causes? Have I over simplified this? While rotating the spindle by hand I noticed if I push or pull the far end of the spindle (at the gear end) I measure 0.001" difference at the chuck end. (Radial push, not axial.) Is this too much run out? Is there an adjustment to reduce this? Increase the pre-load on the bearing? I've never adjusted the bearing so it wouldn't surprise me if it needed tweaking.

Any guidance? These are garden variety ER32 collets but they shouldn't be this far off, that is if they actually meet their spec.

I have to say, even with 2 thousandths run out, the collet chuck is very nice to use.

I have a bigger lathe (12x36) so I'm not sure how my numbers should compare to yours. My spindle endplay is about 0.0001 with roughly 20lbf applied to the spindle. Place your indicator on the headstock near the spindle so your pushing doesn't deform the lathe and affect the measurement. My spindle has 0.0002" TIR as measured on the spindle MT5 taper.

My 3 jaw chuck has 0.0035" TIR in the 1" clamping position, it is pretty true in parallelism.

My 4 jaw chuck can be dialed into near zero runout, but suffers from roughly 0.0012"/12" of out of parallelism to the spindle, an issue for longer parts.

My experience with ER systems has been all ER20 stuff. The chuck should be well ground. You can measure chuck runout by dropping a dial test indicator inside the taper. I'd expect you should be able to fiddle with the backplate to get this within a few tenths. Collets make a huge difference. I have purchased China collets with >0.003" TIR (horrible!). Now I only buy Techniks ER collets. They're not too expensive for the quality and they are guaranteed to 0.0004" TIR, which is about as good as you get in an ER system without going to ultra precision single size collets (has zero collapse range, usually only used for tool holding). The collet nut also plays a very significant role in clamping accuracy. For your chuck, I'd invest in a bearing nut, or a Techniks PowerCOAT nut which will have better clamping force and provide better concentricity.

 

[WobblyHand]

I've partially mislead people. CRS. The real amount of spindle play is 0.002" on a static spindle. Just went down and measured it again. If I push and pull the gear end of the spindle in the horizontal plane, I measure 0.002 play in the horizontal plane at the rod end. (The gear end is opposite end of spindle from chuck.) This is as much as the TIR. Does this point to the spindle bearing being the dominant issue?

 

[macardoso]

In all, I think you need to adjust the spindle endplay. Get that as tight as you can without creating bearing heat. Your lathe manual should describe the adjustment procedure. 0.002" is too loose.

EDIT: Spindle endplay is axial play. Once that is minimized, measure radial runout. If your indicator is mounted to the carriage, then you have a ton of flexibility of the spindle to the headstock, headstock to bed, bed to carriage, carriage to crossslide, etc. which will throw your measurements off.

Then check your ER chuck without a collet or test bar, get it running true.

Finally buy high quality collets, the cheap ones just do not cut it.

 

[WobblyHand]

I have a bigger lathe (12x36) so I'm not sure how my numbers should compare to yours. My spindle endplay is about 0.0001 with roughly 20lbf applied to the spindle. Place your indicator on the headstock near the spindle so your pushing doesn't deform the lathe and affect the measurement. My spindle has 0.0002" TIR as measured on the spindle MT5 taper.

My 3 jaw chuck has 0.0035" TIR in the 1" clamping position, it is pretty true in parallelism.

My 4 jaw chuck can be dialed into near zero runout, but suffers from roughly 0.0012"/12" of out of parallelism to the spindle, an issue for longer parts.

My experience with ER systems has been all ER20 stuff. The chuck should be well ground. You can measure chuck runout by dropping a dial test indicator inside the taper. I'd expect you should be able to fiddle with the backplate to get this within a few tenths. Collets make a huge difference. I have purchased China collets with >0.003" TIR (horrible!). Now I only buy Techniks ER collets. They're not too expensive for the quality and they are guaranteed to 0.0004" TIR, which is about as good as you get in an ER system without going to ultra precision single size collets (has zero collapse range, usually only used for tool holding). The collet nut also plays a very significant role in clamping accuracy. For your chuck, I'd invest in a bearing nut, or a Techniks PowerCOAT nut which will have better clamping force and provide better concentricity.

Very good idea on mounting indicator to headstock! I will do that and report back.

I know about the Techniks collets. Was too cheap to buy them, but I've stared at them a while.

At the moment I'm doubting the spindle. If the spindle isn't true, it's hard to get the rest right.

 

[macardoso]

Very good idea on mounting indicator to headstock! I will do that and report back.

I know about the Techniks collets. Was too cheap to buy them, but I've stared at them a while.

At the moment I'm doubting the spindle. If the spindle isn't true, it's hard to get the rest right.

Yeah, I feel the same pain as you. I settled on ER20 for my mill toolholding and have 24 ER20 chucks. The chucks were $20 each and I have to add a bunch of collets on top of them. Hence why I bought import ones. Unfortunately, the money I spent was just wasted since they were so bad. I can't use most of them and I could have gotten 5-6 really nice collets for the amount I spent.

You have to get the spindle true first, then deal with the chuck and backplate. Only after those are good can you go to the collets.

 

[macardoso]

One more thought. ER collets center best when properly torqued to spec. This torque is wayyyyy more than you'd expect. For an ER32 collet using a standard (non-bearing nut) you need to torque to 100 ft-lbs. I doubt many of us actually do that, but it does improve TIR when using good collets and nuts. My cheap collets deform under much less torque while the Techniks stay true.

 

[WobblyHand]

Yeah, I feel the same pain as you. I settled on ER20 for my mill toolholding and have 24 ER20 chucks. The chucks were $20 each and I have to add a bunch of collets on top of them. Hence why I bought import ones. Unfortunately, the money I spent was just wasted since they were so bad. I can't use most of them and I could have gotten 5-6 really nice collets for the amount I spent.

You have to get the spindle true first, then deal with the chuck and backplate. Only after those are good can you go to the collets.

Well, once you measure things the right way, things do not look so bad. Dial indicator on headstock. Measure spindle runout at chuck taper. A tiny fraction of a tenth. (Less than a 10th of a 10th.) Let's call that good! With the cheap collet in place and a rod of unknown provenance about 0.0009". Guess the chuck is fine. Maybe not the collets or rod. I can figure that out later. At least it's very fixable.

Pushing or pulling the spindle at the gear end in the horizontal or vertical plane resulted in no discernible movement at the chuck taper surface.

That was a lesson in metrology... Don't indicate off the compound! It's too wiggly! @macardoso thank you so much!

Now to get things a little more rigid. Tough with what I have, but maybe there are some improvements to be made. Different thread.

 

[macardoso]

It is amazing how much machines move under load - even really big ones. Usually the best you can do is understand how it moves and compensate with spring passes or other techniques. Think about how hard you pushed on the machine and then compare with the force of actually cutting. That's why people with smaller machines struggle to hold tolerances which are pretty easy to hit on heavier machines.

Another important point, and not to be understated, is the fit and finish of the mating components (bolt flanges, ways, gibs, etc.) play a huge role in machine rigidity. You usually can't do much about the amount of iron you have to work with, but small improvements in the fit of the machine components can have huge increases in rigidity and accuracy.