[4]

Retrofit ER-32 collet chuck to an Atlas/Craftsman 618

[3]
[10] Like what you see?
Click here to donate to this forum and upgrade your account!

SG51Buss

Active User
Registered
Joined
May 20, 2014
Messages
191
I've wanted an accurate collet system for my Atlas/Craftsman 618 lathe for a long time. Acquired an MT2 drawbar collet set several years ago, but wasn't impressed. Too much runout, plus limited depth. Looked into the 5C collet system, but wanted to avoid the expensive/complicated/heavy chuck, and the long protrusion. Decided to look into the ER collet system. Accurate, simple, compact system, which allows through-passage of workpiece. Inexpensive collet chucks of numerous types everywhere. Except for the older Atlas/Craftsman 618 type 1"-8tpi sleeve-bearing spindle.

Yes, there's collet chuck plans out there, and I was heading in that direction, until I found this PSI woodworking ER32 collet kit, on sale for $68 on Amazon. So, let's see if we can make this thing work. Warning: Even though it's possible to cruise thru this post in a few minutes, the actual project took me two months. I must be out of my mind.

Squaring-up on the collet nut threads (because I wanted the collet nut to run true), I shortened and re-registered the 1"-8 mount end. Mounted it on the spindle, and the collet nut threads ran true.

Tried a 1/2" reference rod, it took almost 3/4 turn to snug the collet (something wrong here), and the runout was horrible. Checked runout of the tapered bore, 0.006" TIR. Well, that is to be expected. So, made a 16.0° gauge plate and checked the collet holder opening. Sure enuff, wrong taper, quite a bit less than 16°. Decided to check the included collets, and found 2 of the 5 collets were a bit off taper, 15.85° and 16.35°.

My near-term project will be using 5/16" and 8mm stock, so ordered precision collets in those sizes. Their tapers measured right at 16°.

ER32-Collet-01.jpg

So, now the challege is to cut a precise 16.0° taper, using the compound slide set at an even more precise 8.00° +/- .01° angle (relative to the spindle centerline). Using the 4-jaw, chucked-up, centered and trued a 1/2" reference bar. Then set the compound slide angle using a dial gauge against that reference bar to get within 0.0005" of 0.240" of forward travel on 1.725" of compound slide travel (1.725" times sin(8°) is 0.240"). Max travel on this compound is about 1.8", so this is as good as it gets.

While cranking the compound, noticed significant operator-induced wobble during the compound travel. Even with the gibs snugged, carriage and crossfeed clamped, it still wobbled too much. Then realized that I'd have to carefully and slowly hand feed that thing for 1.3" worth of cut.

So, this project had to be put on hold until implementation of another bucket-list project, a powered compound feed system, which was done in this thread:

http://www.hobby-machinist.com/showthread.php?t=24766

With that done, repeated the arduous 8.00° compound slide setup. Now, we're ready for some cutting on that ER32 collet chuck.

ER32-Collet-02.jpg

Now, this is some strange metal. You can hear the cutting in this short video:

https://youtu.be/qxwERYLQFmA

Yes, there's cutting oil in there. Never cut anything like this before. Noticed its strange behaviour during the cutting of the mount base. With very sharp and honed bits, both carbide and HSS, using various face and rake angles, this stuff takes a bit more pressure to start a cut, and it doesn't cut clean. More like a tearing. Imagine trying to carve a matress with a shovel. What you get are tears and micro-gouges. It doesn't flake/chip/powder like cast, doesn't curl or string like steels. With 0.001" depth of cut, the chips are various micro-sized C-shaped chiplets. It's not that hard, you can file it, but even the file marks are rough. Very odd.

So, figured that toolpost grinding would have to be used. I have a couple of dremel toolpost mounts, but decided to use a Foredom-type cable drive with a long/slim handle to reach into that tapered bore. Doing only 0.001" finish passes, my various fine-grit stones didn't much care for that mystery metal. It just chewed them up. The surface finish looked more like a skin disease than the normal finely-ground surface.

So, next up was my Craytex wheels:

ER32-Collet-03.jpg

After several passes, the coarse Craytex finally gave an acceptable finish. Strangely, I never got sparks with the stones, but the Craytex would occasionally throw a short/tiny yellow spark. Sounds like maybe low-carbon? Exotic chinese rare/precious smelted battleship pig-iron, alloyed with carefully-selected Pabst beer cans and IBM disc drive platters?

After full cleanup, it was time for a test fitting. With the spindle turning, slowly ran a dial test indicator down the bore, and got zero deviation and runout. Chucked-up the 1/2" reference bar (only took 1/8 turn to tighten the collet this time), spun it up slowly, and ran the indicator along the bar using the carriage leadscrew feed.

ER32-Collet-04.jpg

Got 0.0002" TIR at the chuck, 0.0004" TIR at the 6" distance. I can live with that...

ER32-Collet-05.jpg

ER32-Collet-01.jpg ER32-Collet-02.jpg ER32-Collet-03.jpg ER32-Collet-04.jpg ER32-Collet-05.jpg
 

David Kirtley

Active User
Registered
Joined
Dec 17, 2012
Messages
630
Sometimes you buy a tool, sometimes it is a kit. Nice rework. I never have done anything to mine as I only use it on my wood lathe.
 

Bill C.

Active User
Registered
Joined
Mar 11, 2013
Messages
1,390
I've wanted an accurate collet system for my Atlas/Craftsman 618 lathe for a long time. Acquired an MT2 drawbar collet set several years ago, but wasn't impressed. Too much runout, plus limited depth. Looked into the 5C collet system, but wanted to avoid the expensive/complicated/heavy chuck, and the long protrusion. Decided to look into the ER collet system. Accurate, simple, compact system, which allows through-passage of workpiece. Inexpensive collet chucks of numerous types everywhere. Except for the older Atlas/Craftsman 618 type 1"-8tpi sleeve-bearing spindle.

Yes, there's collet chuck plans out there, and I was heading in that direction, until I found this PSI woodworking ER32 collet kit, on sale for $68 on Amazon. So, let's see if we can make this thing work. Warning: Even though it's possible to cruise thru this post in a few minutes, the actual project took me two months. I must be out of my mind.

Squaring-up on the collet nut threads (because I wanted the collet nut to run true), I shortened and re-registered the 1"-8 mount end. Mounted it on the spindle, and the collet nut threads ran true.

Tried a 1/2" reference rod, it took almost 3/4 turn to snug the collet (something wrong here), and the runout was horrible. Checked runout of the tapered bore, 0.006" TIR. Well, that is to be expected. So, made a 16.0° gauge plate and checked the collet holder opening. Sure enuff, wrong taper, quite a bit less than 16°. Decided to check the included collets, and found 2 of the 5 collets were a bit off taper, 15.85° and 16.35°.

My near-term project will be using 5/16" and 8mm stock, so ordered precision collets in those sizes. Their tapers measured right at 16°.

View attachment 82919

So, now the challege is to cut a precise 16.0° taper, using the compound slide set at an even more precise 8.00° +/- .01° angle (relative to the spindle centerline). Using the 4-jaw, chucked-up, centered and trued a 1/2" reference bar. Then set the compound slide angle using a dial gauge against that reference bar to get within 0.0005" of 0.240" of forward travel on 1.725" of compound slide travel (1.725" times sin(8°) is 0.240"). Max travel on this compound is about 1.8", so this is as good as it gets.

While cranking the compound, noticed significant operator-induced wobble during the compound travel. Even with the gibs snugged, carriage and crossfeed clamped, it still wobbled too much. Then realized that I'd have to carefully and slowly hand feed that thing for 1.3" worth of cut.

So, this project had to be put on hold until implementation of another bucket-list project, a powered compound feed system, which was done in this thread:

http://www.hobby-machinist.com/showthread.php?t=24766

With that done, repeated the arduous 8.00° compound slide setup. Now, we're ready for some cutting on that ER32 collet chuck.

View attachment 82920

Now, this is some strange metal. You can hear the cutting in this short video:

https://youtu.be/qxwERYLQFmA

Yes, there's cutting oil in there. Never cut anything like this before. Noticed its strange behaviour during the cutting of the mount base. With very sharp and honed bits, both carbide and HSS, using various face and rake angles, this stuff takes a bit more pressure to start a cut, and it doesn't cut clean. More like a tearing. Imagine trying to carve a matress with a shovel. What you get are tears and micro-gouges. It doesn't flake/chip/powder like cast, doesn't curl or string like steels. With 0.001" depth of cut, the chips are various micro-sized C-shaped chiplets. It's not that hard, you can file it, but even the file marks are rough. Very odd.

So, figured that toolpost grinding would have to be used. I have a couple of dremel toolpost mounts, but decided to use a Foredom-type cable drive with a long/slim handle to reach into that tapered bore. Doing only 0.001" finish passes, my various fine-grit stones didn't much care for that mystery metal. It just chewed them up. The surface finish looked more like a skin disease than the normal finely-ground surface.

So, next up was my Craytex wheels:

View attachment 82921

After several passes, the coarse Craytex finally gave an acceptable finish. Strangely, I never got sparks with the stones, but the Craytex would occasionally throw a short/tiny yellow spark. Sounds like maybe low-carbon? Exotic chinese rare/precious smelted battleship pig-iron, alloyed with carefully-selected Pabst beer cans and IBM disc drive platters?

After full cleanup, it was time for a test fitting. With the spindle turning, slowly ran a dial test indicator down the bore, and got zero deviation and runout. Chucked-up the 1/2" reference bar (only took 1/8 turn to tighten the collet this time), spun it up slowly, and ran the indicator along the bar using the carriage leadscrew feed.

View attachment 82922

Got 0.0002" TIR at the chuck, 0.0004" TIR at the 6" distance. I can live with that...

View attachment 82923
Okay I know my hearing is not perfect working with lathes over the years. Doesn't sound bad to me. Some metals just squeal. The only thing might be the boring bar is a little small in diameter. If the finish looks good without a lot of chatter marks then it is fine. I like to grinder approach to finish the taper. Very nice job. I am glad you were able to convert it to your lathe.
 

SG51Buss

Active User
Registered
Joined
May 20, 2014
Messages
191
Okay I know my hearing is not perfect working with lathes over the years. Doesn't sound bad to me. Some metals just squeal. The only thing might be the boring bar is a little small in diameter. If the finish looks good without a lot of chatter marks then it is fine. I like to grinder approach to finish the taper. Very nice job. I am glad you were able to convert it to your lathe.
Thanx, Bill. Yeah, my tinnitus doesn't help me much, either. And stupidphone audio has it's shortcomings. Wish you had been here to experience cutting that stuff. Outer cuts, facing, were all very strange. Agree on the grinder, so I'm now looking for better stones...
 

SG51Buss

Active User
Registered
Joined
May 20, 2014
Messages
191
Following up on this project, I decided that after all the effort to get the compound to the precise 8° angle (actually, within 1 MOA), I wanted to grind the collets to match the collet holder and be centered to the lathe spindle. So, the compound angle has been left undisturbed since the machining of the collet holder.

The plan is to cut a mandrel, in progressively smaller diameters, to fit the collets in the set, one at a time. Each successive diameter will be a couple thou oversize so that the collet will press-fit onto the mandrel. This will ensure that the collet is unsprung and true, plus withstand the mild grinding forces.

The first collet is 3/4", so I turned a mandrel down to 0.752", 0.0000" TIR, for a nice press fit. I really don't expect to use this collet. Something this large, for this lathe, is probably better off being setup in the 4-jaw. So, this one is sort of a proof-of-concept, and will act as a sort of sacrificial piece to get the little 22mm grindstone settled and true.

I fitted the 3/4" collet, and ran the DTI down its 8° surface, and found that it was indeed a wider angle, about 8.15° (for a total angle of 16.3°), about what I was expecting. Then I painted the whole 8° surface with a sharpie (poor man's layout blue), setup the toolpost grinder, and made a few very light passes.

ER32-Grind01.jpg

The little grinding wheel is still settling in, but you may be able to see that the collet's taper is indeed off. And sure 'nuff, the sides weren't perfectly straight and it's slightly out-of-round (probably can't see that here, gotta turn the chuck to see the varying pattern).

ER32-Grind02.jpg

ER32-Grind01.jpg ER32-Grind02.jpg
 

francist

Active User
H-M Supporter - Gold Member ($25)
Joined
Sep 5, 2013
Messages
1,411
Very nice Steve. As a fellow 618 owner I am also intrigued with all the little custom trinkets you've added to the machine -- carriage stop (very discreet), large cross feed dial, nifty little cover to stop chips from getting into that stupid spot under the headstock, etc etc. Too cool! Now I just have to figure how you did them all.

Thanks for posting, very interesting.

- frank
 

SG51Buss

Active User
Registered
Joined
May 20, 2014
Messages
191
... As a fellow 618 owner I am also intrigued with all the little custom trinkets you've added to the machine -- carriage stop (very discreet), large cross feed dial, nifty little cover to stop chips from getting into that stupid spot under the headstock, etc etc. Too cool! Now I just have to figure how you did them all...
Hey, Frank, thanx for your interest! Yeah, your wandering eyeballs caught some of those things. Well, I plan to do some more threads on those 'trinkets'. Most of them were done by dear 'ol dad back in the '50s and '60s, when I was just a kid, so they'll be more of a post-mortem expose' than an active project. Since they're Atlas/Craftsman specific, they'll go into that sub-forum. I think you'll be interested in that large-dial crossfeed, it has a special feature. And, to pique your interest, ask yourself what appears different about the far side of the carriage...
 

francist

Active User
H-M Supporter - Gold Member ($25)
Joined
Sep 5, 2013
Messages
1,411
Hey, Frank, thanx for your interest! Yeah, your wandering eyeballs caught some of those things. Well, I plan to do some more threads on those 'trinkets'. Most of them were done by dear 'ol dad back in the '50s and '60s, when I was just a kid, so they'll be more of a post-mortem expose' than an active project. Since they're Atlas/Craftsman specific, they'll go into that sub-forum. I think you'll be interested in that large-dial crossfeed, it has a special feature. And, to pique your interest, ask yourself what appears different about the far side of the carriage...
Well I did notice the hinged cover over the across feed screw, but that was intimated with the "etc, etc". Brilliant idea, but I'll look again just in case that's not it.

-frank
 

SG51Buss

Active User
Registered
Joined
May 20, 2014
Messages
191
Good eye, Frank. Yeah, that hinged cover is longer, protects more of the screw and way, and makes it easy to lube the screw. But, that's not it. The mod is very visible, but not obvious...
 

francist

Active User
H-M Supporter - Gold Member ($25)
Joined
Sep 5, 2013
Messages
1,411
Ok now you're just toying with me. It's the wider carriage itself, right? The back wings (if that's what one could call them) look about 1-1/4" wider to each side, and there's something not quite Atlas about the shape. I'd expect the carriage has less potential to cock to the side when you change travel direction. Brazed on?

-frank
 

Rangemaster1

Gunsmith
Registered
Joined
Sep 9, 2014
Messages
90
Good eye, Frank. Yeah, that hinged cover is longer, protects more of the screw and way, and makes it easy to lube the screw. But, that's not it. The mod is very visible, but not obvious...

Hey Frank, check out the power feed on the compound.

Steve, tell us about that.
 

SG51Buss

Active User
Registered
Joined
May 20, 2014
Messages
191
Haha, Frank, you got it! Yes, a wider carriage, longer gib, and 2 more gib screws. Really stabilizes it.
 

SG51Buss

Active User
Registered
Joined
May 20, 2014
Messages
191

Rangemaster1

Gunsmith
Registered
Joined
Sep 9, 2014
Messages
90
Hey, Rangemaster1. Welcome to the forum! That powered compound is disclosed in this thread:

http://www.hobby-machinist.com/showthread.php?t=24766

I see you're into gunsmithing. A noble trade.
Now I'm wondering if I should post up some amateur acccuracy mods done on this lathe, in the gunsmithing section.

Thanks for for the welcome, and thanks for the info on the compound feed. I use my compound to turn tapers on muzzle brakes. I taper the rear of the brake down to the barrel diameter for appearance, and I seem to do a lot of them, mostly on the target rifles I build. It gets tedious at best. A power feed would certainly make life easier. I can get a near polish on the brake by hand, but it's a pain. I'm book marking your thread so I can find it when I get home. Thanks again,

Ron
 

SG51Buss

Active User
Registered
Joined
May 20, 2014
Messages
191
Finished grinding the ER32 3/4" collet, made with a few very light passes.
The little grinding wheel settled-in and now has a proper cutting edge.
ER32-Grind03.jpg


Changed over to a medium Craytex wheel to polish-out the grinding.
ER32-Grind04.jpg


Test fitted this collet into the collet holder, got no sloppy feel.
Slipped a 3/4" bar into it, and found that tightening from finger snug to full tight was less than a 1/10 turn of the collet nut.
Much better.

ER32-Grind03.jpg ER32-Grind04.jpg
 

SG51Buss

Active User
Registered
Joined
May 20, 2014
Messages
191
While doing the next collets, came across something that may be of interest.

First, the 5/8" collet had a few small burrs where the ID exits the backside of the collet. These burrs would catch the edge of a test shaft as it was slid thru, and could cause accuracy problems, plus scratch the shaft. The burrs had to be dressed out with a craytex point. Here, you can see the thin, shiny corner edge where one of the burrs was removed.
ER32-ColletBurr.jpg

The second issue is the collet taper. As measured earlier, the 5/8" collet showed a 15.85° total angle. But, when fitted to the 0.627" mandrel, its angle was rechecked using the compound and DTI, and found to be very close to 8.00°.

What this means is that an ER series collet may not present a precise 16° taper while in its relaxed, unsprung state (especially the budget ones). Some ER collet chuck projects that I've read involve using the collet as a master gauge when cutting the chuck inner taper, using layout blue markings as an indicator to adjust or confirm the compound angle. I've noticed, in those project pictures, that this 'master' collet is empty, relaxed, unsprung.

As such, for those folks that choose this 'master-collet / layout-blue' method of cutting the taper, I would humbly recommend fitting a slightly oversized (0.002" - 0.003") shaft into any collet that would be used as a master angle checker .

ER32-ColletBurr.jpg
 

SG51Buss

Active User
Registered
Joined
May 20, 2014
Messages
191
Been grinding the other ER32 collets, have some observations about them.

None of them (so far) have the correct taper, don't have straight sides, and are eccentric (by up to 0.003")
What's worse, is that they looked so nice and shiny, until I took a closer look. It seems that they were rather roughly cut, then someone spent considerable time at the buffing wheel to make them pretty. So, this is what to expect from import tooling, pretty junk. Oh, well. After grinding, they're better than they were.

Now I'm ready to work on my high-dollar, high-precision SYIC Techniks 8mm collet.
It appears to be much better quality, smoother bore finish, proper chamfering of bore edges, no sign of buffing wheel cover-up funny business.
But, under the magnifying glass I found numerous nano-sized metal hairs emerging from the slits. They're stuck there and don't wash away with cleaner. Had to meticulously scrape them out with a fine dental pick.

The mandrel has been turned dead true, then collet fitted and painted. Checked its taper with the dial gauge, looks very close.
ER32-Grind8mm01.jpg


Started with a light 0.0005" cut, and let the powered compound make several passes until spark-out.
There is a very slight wasp-waist shape here, and it's very slightly eccentric, revealed by the residual paint in the circled area.
But, we're only talking a few microns here.
ER32-Grind8mm02.jpg


Advanced the cross slide another 0.0005", and let the powered compound make numerous passes until spark-out.
The collet cleaned-up nicely, no more indicator paint left.

Changed over to a medium Craytex wheel to polish-out the grinding.
ER32-Grind8mm03.jpg


Finished, oiled, put back into its wrapper for later testing.
ER32-Grind8mm04.jpg

ER32-Grind8mm01.jpg ER32-Grind8mm02.jpg ER32-Grind8mm03.jpg ER32-Grind8mm04.jpg
 

Rick Leslie

Active User
Registered
Joined
Jul 30, 2012
Messages
480
Hey, Rangemaster1. Welcome to the forum! That powered compound is disclosed in this thread:

http://www.hobby-machinist.com/showthread.php?t=24766

I see you're into gunsmithing. A noble trade.
Now I'm wondering if I should post up some amateur acccuracy mods done on this lathe, in the gunsmithing section.
By all means, please do! Gunsmithing is near and dear to my heart as well. I would love to see some of your work in that arena.
 

SG51Buss

Active User
Registered
Joined
May 20, 2014
Messages
191
Hi, Rick. Okay, great, the votes are in. I'll post some when I get a round tuit...
 

SG51Buss

Active User
Registered
Joined
May 20, 2014
Messages
191
Next up was the ER32 high-dollar, high-precision SYIC Techniks 5/16" collet.
It also is much better quality, smooth bore finish, proper chamfering of bore edges, no buffing wheel cover-ups.
But, it also had numerous nano-sized metal hairs emerging from the slits. So, cleaned them out with a fine dental pick.

After fitting to the mandrel and painted. Checked its taper with the dial test gauge, almost dead-on.
Started with a light 0.0005" cut, and let the powered compound make several passes until spark-out.
There was very slight wasp-waist shape, and it was barely eccentric.
It was so close that it cleaned-up on that first cut setting. Finished it off with the craytex, oiled, and put it away.


Last was the 1/4" collet. This one blew me away. :faint:
First, it wasn't a 1/4" collet, it was a 6mm collet marked 1/4". :headscratch:
Then, after mounting, found its taper was so far off, that I had to stop and double-check my setup.
Sure enuff, its supposedly 16° taper angle worked out to 17.25°.
And worse, it was extremely eccentric (by several thou) and WOBBLY! :wantmommy:

Started the cuts, and they revealed the seriousness of the issue.
In this pic, you can see the severe out-of-taper. About 0.005" of the radius has already been ground out, long way yet to go.
Circle #1 - The sides aren't even straight.
Circle #2 - You can see the rough machining. The polisher guy apparently missed this one.
ER32-Grind6mm.jpg


So, decided to stop here for awhile. Let the machines and me cool-off.
Submit this post, and take a nap...

ER32-Grind6mm.jpg
 

SG51Buss

Active User
Registered
Joined
May 20, 2014
Messages
191
Well, time for an update.

I wasn't getting the accuracy I was expecting after grinding the 16° tapers on the collets.

So, decided to repeat the whole mandrel/regrind procedure on the top 30° taper of the collets. Glad I did. Some were way off.

The 3/4" collet top taper had a severely and strangely stepped offset.
ER32-TopGrind-01.jpg

The 5/8", 1/2" and 3/8" collets were close, not as bad, but still needed cleaning up.
The hi-precision Techniks 8mm collet top taper was dead-on.
The hi-precision Techniks 5/16" collet top taper was extremely close.

The budget 8mm collet top taper had the wrong taper and an offset.
ER32-TopGrind-02.jpg

The 1/4" collet top taper also had the wrong taper and a severe offset.
ER32-TopGrind-03.jpg

Next will be as-fitted accuracy measurements...

ER32-TopGrind-01.jpg ER32-TopGrind-02.jpg ER32-TopGrind-03.jpg
 

SG51Buss

Active User
Registered
Joined
May 20, 2014
Messages
191
More updates.

Still not getting the accuracy I was expecting after grinding the 30° top tapers on the collets.

I was thinking that the 16° chuck cavity would be doing the bulk of the centering, and that the nut (with its slight looseness) would naturally center on the collet's top taper, and simply squeeze the collet into the body. It appears that the nut has a greater influence on collet centering than I originally thought. This is turning into a real challenge and learning experience.

So, I decided to chase the metric 40x1.5mm collet nut threads on the chuck body, concentrating on the important thrust side of the threads. Cut a thread relief area behind the threads, did the changegear approximation for cutting metric 1.5mm threads, and synchronized the threading bit to the existing threads.

A couple of 0.0005" skim passes on the thrust side of the threads got them concentric.

ER-ColletChuckThreading.jpg

Fitted a plastic crush/loading spacer inside the collet nut, so that the nut could be align/tightened to the chuck, and checked the runout of the as-loaded nut's 30° taper.

It, too, was a little off, about 0.0010" TIR. So, did a light skim to recenter it.

ER32-ColletNut.jpg

We'll see how this works...

ER-ColletChuckThreading.jpg ER32-ColletNut.jpg
 

SG51Buss

Active User
Registered
Joined
May 20, 2014
Messages
191
Well, this project has been an exasperating learning experience. After all that work to precision grind the collet chuck and collets, testing the runouts were horrible and inconsistent, anywhere from 0.001" to 0.003".

So, suspecting the clamping nut, I fitted a 3/8" test bar by pressing in a collet without the nut. The 16° taper is narrow enough to solidly hold the bar as-is.

ER32-Nut-Minus.jpg

Well, whattya know. I got runout of ZERO at the collet face, 0.0003" runout at the 3" position. THIS is what I wanted.

Sure 'nuff, isolated the problem to the collet nut. Long story short, the clamping nut was made to the wrong dimensions. Its extractor ring was contacting the rear collot groove face when the nut was tightened. The distance from the clamping nut face to the backside of the extraction ring was over 9.5mm, and inconsistent. I believe that for the ER32 clamping nut this dimension is supposed to be 8mm, which would allow for the collet to cleanly penetrate into the nut's 30° taper for the full clamping range. From what I can tell from measuring the inner dimensions of this clamping nut, the feature depths were made to the specs of the ER40 collet system. I shaved 0.030" off the backside of the extraction ring, leaving just enough for it to work, the runouts improved markedly, but now I had no collapse clearance.

Hopefully, this pic identifies the issue:

ER-ColletNutWarning2.jpg

Dang it, I liked that nut. Knurled and large enough to hand tighten, with a handy tommybar hole. So, decided to order (2) ER32 nuts off eBay (The $8 versions).
If you study this pic, comparing the original nut on the left with the new ones on the right, you may be able to see the difference in the width of the 30° contact face, as though it were an ER40 nut.

ER32-Nut-New.jpg

The new nuts were rough inside, with curls of metal dangling from the ends of the threads, and coarse finishing of the extractor ring edges, making collet insertion difficult. Fortunately, the threads and clamping face were like finish ground. Cleaned them up and chamfer/polished the rough extractor ring edges, collets fit much easier now. They also didn't have the telltale notch/mark for the extraction ring, so ground little dimples in the nut body adjacent to the wide part of the extraction ring, per Rego-Fix.

Tried both new clamping nuts, on 1/2", 3/8" and 5/16" test bars. The first one gave 0.0005" to 0.0008" runouts, the second one produced 0.0001" (at collet face) to 0.0003" (at 3" out), much better.

ER32-Nut-SelectNew.jpg

It appears that the clamping nuts have a much stronger influence on accuracy than what I was expecting. Lesson learned.

I'll settle for this second one. For now.

ER32-Nut-Minus.jpg ER-ColletNutWarning2.jpg ER32-Nut-New.jpg ER32-Nut-SelectNew.jpg
 
[5] [7]
Top