What's new
What's new

Drill chucks accuracy obsession.

Ratz

Aluminum
Joined
Feb 4, 2014
Location
Canada
I have around 8 various chucks in the shop with different mileage. Taiwanese Vertex, Jacobs, old Rohm, and other Chinese versions.

My best is keyless VERTEX capable to take 11/16 stock. Measuring run out of a ½” solid carbide LMT ONSRUD (mill with basically zero run out) and measure about 1” below the jaws of the chuck, I get 0.002 run out. Not bad.

One problem with this chuck is, it is unable to take small drill bit. Less then 1/8”
Old Rohm is out 0.005 measured within same location.

Worst is a Chinese junk at 0.015 sitting at the edge of the rubbish bin.

Since it made me sick to watch a small drill wobble (stubby machines drill is not so bad)
with my second best 0.005, I decided to go all out and spent $220 to purchase a brand new Rohm from MSC. As even the Jacobs are now made in China. Read a 4 star review where it stated run out was basically immeasurable.

Today I installed the new chuck to my milling machine.
67437657 CAP. 1/32-1/2 J3 MOUNT ROHM DRILL CHUCK KEY TYPE

Spindle run out is 0.0001 to 0.0002
R8 arbor JT3 run out 0.0004
Top of the drill chuck 0.0005
Body of drill chuck just below the chuck key holes 0.0008
½” carbide stock known to be basically perfect, ½” below jaws 0.0012”.

I thought if it was 0.0005” I would be OK with that. Being a German precision tool,
purchased from reputable supplier, is this acceptable? Only thing I can see is R8 arbor JT3 run out 0.0004”. When measured 2.5 inches lower, that may be the culprit. Maybe new arbor is in order.
 
I have around 8 various chucks in the shop with different mileage. Taiwanese Vertex, Jacobs, old Rohm, and other Chinese versions.

My best is keyless VERTEX capable to take 11/16 stock. Measuring run out of a ½” solid carbide LMT ONSRUD (mill with basically zero run out) and measure about 1” below the jaws of the chuck, I get 0.002 run out. Not bad.

One problem with this chuck is, it is unable to take small drill bit. Less then 1/8”
Old Rohm is out 0.005 measured within same location.

Worst is a Chinese junk at 0.015 sitting at the edge of the rubbish bin.

Since it made me sick to watch a small drill wobble (stubby machines drill is not so bad)
with my second best 0.005, I decided to go all out and spent $220 to purchase a brand new Rohm from MSC. As even the Jacobs are now made in China. Read a 4 star review where it stated run out was basically immeasurable.

Today I installed the new chuck to my milling machine.
67437657 CAP. 1/32-1/2 J3 MOUNT ROHM DRILL CHUCK KEY TYPE

Spindle run out is 0.0001 to 0.0002
R8 arbor JT3 run out 0.0004
Top of the drill chuck 0.0005
Body of drill chuck just below the chuck key holes 0.0008
½” carbide stock known to be basically perfect, ½” below jaws 0.0012”.

I thought if it was 0.0005” I would be OK with that. Being a German precision tool,
purchased from reputable supplier, is this acceptable? Only thing I can see is R8 arbor JT3 run out 0.0004”. When measured 2.5 inches lower, that may be the culprit. Maybe new arbor is in order.

Doesn't sound bad to me. If you factor in that regular drill bits can have a little bow to them, you might try doing a few chuckings and measure the runout with each attempt. But if you're really looking for concentricity you're better off using a collet system anyway.

I'm mostly Albrecht with some quality Asian chucks mixed in. I work within the limits of the chucks, when I want better accuracy it's collets and a spot drill.
 
.0012 is as good as you can expect on a Jacobs type chuck, you think you are going to get less than a half thou ? thats good collet territory.
if your spindle taper, your arbor, and your chuck are all within .0002, and you extend that down below the chuck, well, it would be more than that wouldn't it? (depending on how the tolerances stack up)
even if you re-ground your internal spindle taper on new bearings, then ground the jt3 end of the adapter in the machine, there is no guaranty it would improve, it could as easily get worse too.

how many different diameters did you check it on? is your spindle new or newly rebuilt to high specs? did you side load while checking runout?
 
.0012 is as good as you can expect on a Jacobs type chuck, you think you are going to get less than a half thou ? thats good collet territory.
if your spindle taper, your arbor, and your chuck are all within .0002, and you extend that down below the chuck, well, it would be more than that wouldn't it? (depending on how the tolerances stack up)
even if you re-ground your internal spindle taper on new bearings, then ground the jt3 end of the adapter in the machine, there is no guaranty it would improve, it could as easily get worse too.

how many different diameters did you check it on? is your spindle new or newly rebuilt to high specs? did you side load while checking runout?

I checked 1/2, 3/8, and 1/4 with pretty well same results. I did not preform any side loading.
I may not been clear as this is 1/2" Rohm chuck made in Germany. So I am not going to do any mods. but I will likely keep it after reading last two posts.

Yes I use collets when I can, the trouble is all my collets ER25, ER40 and R8 are in 1/16" increments, so if I use numbered or letter drills fit is just not that good. But I guess I could pre-drill with 1/16" bit and then use the chuck with required final drill (number or letter).
 
I checked 1/2, 3/8, and 1/4 with pretty well same results. I did not preform any side loading.
I may not been clear as this is 1/2" Rohm chuck made in Germany. So I am not going to do any mods. but I will likely keep it after reading last two posts.

Yes I use collets when I can, the trouble is all my collets ER25, ER40 and R8 are in 1/16" increments, so if I use numbered or letter drills fit is just not that good. But I guess I could pre-drill with 1/16" bit and then use the chuck with required final drill (number or letter).

Why use a 1/16 drill? Center drills are much better for locating due to the body geometry designed to guide a regular drill into the center, they are nice and short and stiff, and best of all, they will fit in collets!

Once you have the hole started, I don't think there would be much value to using such a small drill to pilot, in fact it would probably be more likely to flex and wander, cancelling out the benefits of the reduced runout.
 
Spindle run out is 0.0001 to 0.0002
R8 arbor JT3 run out 0.0004

Top of the drill chuck 0.0005
Body of drill chuck just below the chuck key holes 0.0008
½” carbide stock known to be basically perfect, ½” below jaws 0.0012”.

WTF ARE you playing at, anyway?

You don't even have the spindle that would objectively support reaching the limits you think you can chase. Short by an order of magnitude, and "maybe" not even pointing on-axis?

And then you stack an R8/JT four times WORSE?

And then a(ny) adjustable Chuck?

See to why it is you start at 2 tenths and get to 12, "right away" and best-case, first.

Chuck(s) out of spec? Or the mountup of them was? And "Oh, BTW", just how tight a spec was it you expected, anyway?

Cheap or dear, all these goods are built to a price-range spec, expect to be rebuilt or replaced as they accumulate wear and damage. Certificated laboratory instruments or measurement standards they are not.

Meanwhile.. "Drill chucks" on a MILL are a quick and flexible time-saver. A CONVENIENCE.

- Drillpresses are meant for plain-old holes.

- Jig bores or mills, traditionally, decent CNC goods more recently, for use where that is not GOOD enough.

Chucks prioritize swap-speed and wide size range. Flexibility of USE is the prime directive. Wise to adjust your expectations accordingly.

If you seek the sort of accuracy the Swiss - and others - actually NEED for making highly precise and precisely located holes, all day, every day, all year?

Buy the equipment made for that purpose.

What sits above and behind the toolholder, plus precise positioning, under. Not just their collets.

Cheap, it was never.

Sorting and cherry-picking ordinary goods - whether 8 or 80 - will not change that in anything close to a sane amount of time, nor with any predictable longevity, once put to the working stresses of real-world work.
 
Drill chuck and accuracy are an oxymoron

In "absolute" terms, and against easily half-a-dozen collet-system alternatives?

Certainly.

"Useful", nonetheless or they would not sell by the brazillions, powered hand-tools far more numerous than mills or drillpresses, if only 'coz so very few of those sell "naked".

:)

"Reasonable expectations" is all we get, ANY product, any industry, any point in the development of the relevant technology and its needs.

Need more performance? Bring more money. Then wait.

Meanwhile... relatively common collets in the "middle" - SK's with about one-quarter the collapse range of the ER's full millimeter:

SK16 Collets - MariTool
 
Last edited:
depending on how the tolerances stack up

Please don’t take it personally, of course you will because I’m pointing out something—

it’s not tolerances that add but deviations, errors. A tolerance cannot be measured, it’s given by somebody in conjunction with a measure. I hear this so many times even from time-tested mates. It’s wrong. Having made a fool of myself now I beg your pardon for lecturing around.
 
...it’s not tolerances that add but deviations, errors.

"Too soon we forget".. when the goods were more worn or had been manufactured, ground or re-ground imperfectly, more folks made use of "countervailing errors" more often.

Holder was out "x" thou one direction, tool was out "y" thou as well, one sought the "O'Clock" mate-up that offset the two errors as best could be.

That often cut the cumulative error down, sometimes to near-zero, flaws notwithstanding.

Economical goods quality has improved the past 50 years, but so have the goals for the work.

The concept is still worth chasing when it matters enough to invest the time.

Read "Seldom", but most especially if the goods are already about as low-error as can realistically be afforded, yet not QUITE good enough for a touchy tasking.

Case in point, right here.

Did the OP try "clocking" his so-called perfect test item or not?

Might mic "nearly perfect" as to OD, but how straight is it, really? it isn't in the factory any longer. It is out in the rude world.

What is the TIR at the furthest point from the face of the chuck, any given clamping of many, not just the closest, and not just the ONE reading? More need to be recorded for valid statistics.

What was the cleaning procedure?

What was clamping torque, each go?

One wants to chase tiny numbers one has to put in the work. The "problem" might not be where he thinks it is?

:)

NB: Few drill chucks claim any better than a thou and a half runout over their full range as an average, in any case. Even that is for the "premium" grades.

Collets with wider grip ranges, such as ER, are better than chucks, but even with more than one "grade" are not that close to perfect, either, are also spec'd as to their "average" over the range.

And then we hit the wall on spindle taper, dings, debris..then bearings & perhaps their preload?

"Run what you have."

What you WISH to have is .... still only a dream?

:D
 
What is the runout spec? Lemme guess, not specified anywhere since this is speciality product only for *Murican market?

Röhm Spiro is their best keyless chuck and its specified for 0.05mm or almost twice as much what(0.03mm) you are measuring... so yes, you are obsessed.

"CNC drill chucks" might be bit better, many of them specified at 0.02 to 0.03mm runout.
 
Yes I use collets when I can, the trouble is all my collets ER25, ER40 and R8 are in 1/16" increments, so if I use numbered or letter drills fit is just not that good. But I guess I could pre-drill with 1/16" bit and then use the chuck with required final drill (number or letter).

You should get at least a few ER16 holders and a couple full sets of metric collets. Even the cheap imports are good enough for drill use, and will give you a ton of flexibility when using drills that you don't have right now.
 
Please don’t take it personally, of course you will because I’m pointing out something—

it’s not tolerances that add but deviations, errors. A tolerance cannot be measured, it’s given by somebody in conjunction with a measure. I hear this so many times even from time-tested mates. It’s wrong. Having made a fool of myself now I beg your pardon for lecturing around.

This is interesting.

In my world, tolerance stacking is a term used to quantify the limits of uncertainty, i.e. it's usually discussed before a part is made, not after. Usually in the context of two or more toleranced dimensions that when summed at their limits contradict one other toleranced dimension.

Deviation stacking (as you describe it, it's not a term I've heard used before) is a physical attribute, a resultant total deviation from the nominal of a sum of two or more toleranced dimensions, as measured on a physical object.
 
Please don’t take it personally, of course you will because I’m pointing out something—

it’s not tolerances that add but deviations, errors. A tolerance cannot be measured, it’s given by somebody in conjunction with a measure. I hear this so many times even from time-tested mates. It’s wrong. Having made a fool of myself now I beg your pardon for lecturing around.

Yup, you are correct, no offense taken, that is the commonly MIS-applied term, for the “potential” of cumulative deviation;)
 
What is the runout spec? Lemme guess, not specified anywhere since this is speciality product only for *Murican market?

Röhm Spiro is their best keyless chuck and its specified for 0.05mm or almost twice as much what(0.03mm) you are measuring... so yes, you are obsessed.

"CNC drill chucks" might be bit better, many of them specified at 0.02 to 0.03mm runout.

Not sure they are inherently "better" as to the working parts and min-possible error if lovingly assembled to a tail, but...

They CAN gain the advantage of greater consistency with less of that hassle, being sold with tail already integral, decent factory QA.

One example... NPU or HEXA:

Products : Llambrich
 
This is interesting.

In my world, tolerance stacking is a term used to quantify the limits of uncertainty, i.e. it's usually discussed before a part is made, not after. Usually in the context of two or more toleranced dimensions that when summed at their limits contradict one other toleranced dimension.

Deviation stacking (as you describe it, it's not a term I've heard used before) is a physical attribute, a resultant total deviation from the nominal of a sum of two or more toleranced dimensions, as measured on a physical object.

Added bold to the part where you answer your own question. I've never heard about deviation stacking either because I always just called it total deviation, or total error.

I spent a lot of time at my old job poring over pages and pages of inspection reports for plastic parts first articles coming off the tools looking for just that. Some dimensions would be measured out of spec, but as long as the total deviation was within the tolerance stack limit, I would sign off on the report. This is of course operating under the assumption that every part coming off that tool was gonna be the same, of course. :crazy:

But it made me feel good about my life, especially in contrast to one of the other engineers who would just shrug and say "it's close" when signing off. Huge pet peeve of mine. You put the damn tolerance there to begin with! YOU COULD HAVE MADE IT ±0.5mm IF YOU WERE GOING TO ACCEPT PARTS THAT WERE OFF BY MORE THAN 0.2mm :angry: Sorry for the digression.
 
Added bold to the part where you answer your own question. I've never heard about deviation stacking either because I always just called it total deviation, or total error.

I spent a lot of time at my old job poring over pages and pages of inspection reports for plastic parts first articles coming off the tools looking for just that. Some dimensions would be measured out of spec, but as long as the total deviation was within the tolerance stack limit, I would sign off on the report. This is of course operating under the assumption that every part coming off that tool was gonna be the same, of course. :crazy:

But it made me feel good about my life, especially in contrast to one of the other engineers who would just shrug and say "it's close" when signing off. Huge pet peeve of mine. You put the damn tolerance there to begin with! YOU COULD HAVE MADE IT ±0.5mm IF YOU WERE GOING TO ACCEPT PARTS THAT WERE OFF BY MORE THAN 0.2mm :angry: Sorry for the digression.

In my experience it's not uncommon to find tolerance spec'd tighter than necessary. Often this is done to avoid rejected parts - if the tolerance is tighter than necessary but the vendor misses by a tiny bit it's no big deal. Plus it looks like you're doing the guy a favor when you let it slide. Though really you're not since it's teaching him that it's okay to blow tolerance.

So yeah, I agree, poor practice.
 
In my experience it's not uncommon to find tolerance spec'd tighter than necessary. Often this is done to avoid rejected parts - if the tolerance is tighter than necessary but the vendor misses by a tiny bit it's no big deal. Plus it looks like you're doing the guy a favor when you let it slide. Though really you're not since it's teaching him that it's okay to blow tolerance.

So yeah, I agree, poor practice.

Of course, getting tolerance stacks right up to the limits of function, but not beyond, is hellishly difficult. I've made the tolerances too tight before, but then I always specify "OK per mating part assembly" or some such, I've never just shrugged my shoulders and said "you did your best" when looking at parts out of spec. It's just a mentality that in my opinion is incompatible with acceptable engineering practice.
 
Cheap ass Jet "Mill"/Drill made in taiwan 1979
Cheap ass Shars keyless chuck made in China
Cheap ass Chinese end mill
Mitutoyo indicator 0.0005"
TIR is 0.001" at end mill

stir_the_pot.jpg


YouTube
 
I had to drill some very long, straight holes in Ti a few years back and spent big bucks on a Hardinge "special accuracy" collet and fiddled around quite a bit to get the run out to a couple tenths.

I couldn't imagine any drill chuck doing this...
 








 
Back
Top