No announcement yet.

Accuracy of Chucks, collets etc.

  • Filter
  • Time
  • Show
Clear All
new posts

  • Accuracy of Chucks, collets etc.

    I did not feel like making anything this morning but in view of the problems Brian seemed to be having with his 3 jaw I made a few tests on my 60 yr old Southbend 9 inch lathe and the accessories I use.
    The dial gauges I usually use are marked in thous, For the tests I used ones marked in half thous.
    Mandrel register- od - no perceptible movement of needle, Bore of spindle again no perceptible movement. Face chucks screw against again no perceptible movement.
    3 jaw chuck surprisingly good,using marked keyhole 0.0005 inch tir measuring at chuck face, at 2 inches out the same.
    Using other keyholes0.003 inch and 0.0025 tir at face, at 2 inches out 0.005 tir both.
    4 jaw I got the face to 0 -0 ie no perceptible movement of needle. at 2 inches out 0.002 Tir.
    3 C collet, running on adapter in mandrel nose at face ran at less than half a division movement, probably less than 0.0003 inch at 2 inches out 0.001 Tir.
    5 C collet running in screw on collet chuck, at face ran at 0.0005 Tir, at 2 inches out 0.002 Tir.
    All tests were done using a seven sixteenth dowel pin. Mandrel was turned by pulling on the primary belt.
    My collets are all secondhand, the 3 C s are in literally mint condition, whole set to half inch cost me 300$. The 5 Cs are old and they and their chuck have quite a history, from England to South Africa to Canada.
    I hope this is of interest and perhaps help to someone.
    Regards David Powell.

  • #2
    You have some pretty good kit, better than average for the 3 jaw chuck. Try a range of sizes in it if you have some ground stock, it will highlight any scroll errors. I have a couple of MT sleeves for capstan lathes which make good 1 and 2 inch test pieces. The best performing keyholes should have a little mark for future use.


    • #3
      Believe me, I started out with junk, learned how to get reasonable results with lots of effort, and just kept on moving ( Usually) forwards to try to be able to get better results with less effort. Many Home Shop Machinists do better work than I do. I do not usually seek perfection, havent the time to build all I want in my lifetime, but I do want to leave behind some good quality working models.
      Regards David Powell.


      • #4
        i like this post. it highlights what i said earlier: you have to know your chuck. it also shows that collets are not as accurate as most people assume/would like them to be.


        • #5
          In my case I think that most of the errors which I see in the 5c collets have one of two causes, firstly many of the collets themselves are elderly, they have seen hard industrial use and secondly they are held in a heavy un named collet chuck which fits on the screwed nose of the 9 inch Southbend, thus there is a lot of overhang. If I test and then unscrew the chuck and then refit it I get slight changes in the errors. But compared with what I started out with even the very worst of this lot is wonderful.
          Regards David Powell.


          • #6
            I can usually get my old worn 3-jaw chuck to within 1-2 hundreths of a mm by loosening the screws a little and tapping. Jaws are a bit bellmouthed however, 4 jaw I can get to basically no noticeable runout. Don't got any collet chucks for the lathe, yet. Checking runout on my ER32 taper is less than a hundreth of a mm in the mill, don't have a good ground shaft right now so I can check the runout of a mounted part.

            I really like 3 jaws and being able to adjust them. Don't have a true adjusta-thru-chuck or whatever they're called. And my new 3-jaw chuck that still sits in the box isn't one either, but I am wondering if I can modify it, or make the backing plate so I can have 4 adjusting points.


            • #7
              At the museum, for the Smart & Brown model A, we have a large choice of chucks.

              100mm PB three jaw in new condition.
              2 off 5" PB three jaw, with all of their jaws ground true.
              125mm Chinese all jaws ground (one of the good ones made by Huhhot)
              160 Chinese new, Bison copy.
              6 3/4" serrated jaw Pratt in excellent condition with a huge quantity od soft jaws plus two sets of hard jaws ground.

              All except the serrated jaw chuck have their backplate to chuck registers made purposely loose to enable fine adjustment if required. They are held on their backplates by six screws, not the usual three, an easy modification. Before grinding any jaws, the main body of the chuck is set to run true, and one of the keys is selected to be the master.
              We also have a nice Toolmex 6" lightweight four jaw independent in new condition and a 160mm Chinese one which is so so for quality. They don't need a loose register for obvious reasons.
              An er25 plate type collet holder on a backplate also has a loose register for fine adjustment if required.

              The lathe also has a large quantity of collets which fit directly in the spindle. Most are well worn, so I selected candidates for refurbishing. The collets were held open with shims and bored with Picco solid carbide for the 1/4" and 6mm, and with a 6mm solid carbide Garant boring bar for the 5/16", 8mm, 3/8", 10mm, 1/2", 12mm, 5/8" and 16mm sizes. Despite the interrupted cuts, the bores are very good and run true. There are also stepped inside and outside collets, I have not checked them for accuracy.


              • #8
                Originally posted by dian View Post
                i like this post. it highlights what i said earlier: you have to know your chuck. it also shows that collets are not as accurate as most people assume/would like them to be.
                As the estimable Mr Powell pointed out, the goodness of the collet is dependent on the device holding it, and the machine it is on. I would not look down on collets as inaccurate, it may be the machine that has the issue.

                Some machines take the 5C (or other collet) natively in the spindle. Those are generally higher class machines, and you can expect to get rather close to the inherent accuracy of the collet when using them in such a machine (given any issues of wear and distortion from abuse).

                You can usually do OK with a "collet closer" that fits into the native taper of the spindle. The closers do vary, and the best is one you make yourself on the machine, especially if you always put it into the spindle in the same orientation.

                One place where you CANNOT assume accuracy is when you have an add-on "collet chuck", perhaps screwed onto a threaded nose, and sticking out a third of a mile from the spindle nose. Tn that situation, the sky is the limit.... it might be worse than a 3 jaw, or it might be essentially spot-on concentric. Or anywhere in between. And it might shift around while in use.
                CNC machines only go through the motions


                • #9
                  well, i while ago i wanted to equip the fehlman p50 with er 25 collets. i sent back three sets of german provenience (whatever that means today) because most of the collets were way out from the 0.01 mm advertised tir. some were as bad as 0.1 mm. i settled on regofix 5ยต "super precision" collets and i can tell you they are in a range of 0.02 mm at best. my measurements took many hours and were done very dilligently.

                  i own one 5c collet, so i cant talk about those, but what i noticed is that most of the sellerst dont even advertise tir.

                  (i mentioned this several times, but here goes again: there is no set standart for specifying tir of collets, like e.g. 10 x diameter from chuck for drill chucks according to din. some get measured at 30 mm, which has to be accounted for when comparing. when a drill chuck shows 0.05 mm for a 10 mm drill, that can be translated to 0.015 mm for a 10 mm a collet.)

                  edit: i also have a set of ebay collets for under $3 a piece. they are not even ground on the inside and i never really bothered to measure runout. but i noticed that several are around zero 30 mm out, although the endmill is probably wobbling in there quite a bit.
                  Last edited by dian; 11-26-2020, 08:44 AM.