Inside of my 5C shows 1/2 a thou which does for me.

I fitted a small Chinese collet chuck to my Cowells lathe. I made the backplate myself from scratch, making every effort to minimize runout I ended up with just under 0.02mm. I hoped for less but it's good enough for my needs.

Apart from the limitations of the search engine, the forum isn't structured as a reference source. Wikipedia is best for that, but you can't have a conversation with it! Wikipedia's article on runout doesn't answer the question either – it's an overview without figures. Maybe we should improve it?

There probably isn't "a run out figure that was deemed acceptable by the majority of model engineers", because the answer depends on the nature of the work. Money matters too! A few things I do are crude enough to let me swap ends in a 3-jaw without worrying about the resulting error. Other work demands much more care, usually with a 4-jaw adjusted to the best of my ability. My lathe, chuck, and my Dial Indicator limit the best I can do to about 0.01mm close to the chuck. Long work requires steadies and an accurately adjusted tailstock. For close up only I have an ER collet chuck, which can be carefully centred with the same Dial Indicator, so similar run out to the 4-jaw. My collets are inexpensive, bought from Warco and ArcEuro, and they're all as good as the chuck.

I rarely attempt high accuracy work, instead working within the ordinary limits of my equipment, which is something like 0.02mm or a thou. I don't use collets often, but once in a while I do a job with round shafts being repeatedly taken on and off the lathe: when this happens collets are much faster than the 4-jaw.

You can guess I don't do much watchmaking or instrument work! If I did, something like a Cowells with a complete set of collets would be needed.

Industry have another reason for wanting low TIR collets, especially ER. They spin them very fast, and a slightly unbalanced collet at 30,000 rpm does a poor job. The same collet at 3000rpm would be fine.

So I come at run-out by asking 'what do I need to do to get the required result?' Quite often there are alternatives; just as well – I dislike spending money on tooling that's better than my lathe or mill can make use of.

However, as Model Engineering is a hobby, others enjoy owning the very best, even if there's no practical need for it. Always nice to have capability and reliable tools, especially if someone else paid for them! Most of us own a mix of more-or-less suitable tools obtained by various methods.

Dave

I'd think .001" is still doing pretty good Andrew, but as others have already said. What your using any work or tool holding system for dictates how much is or isn't adequate. There's obviously a bit of a difference between making a new wheel barrow axle and machining something like a piston or as aptly pointed out, watch parts.

A lot of the following I wish I'd understood a lot better and much sooner than I did. But fwiw.

Yes the run out check you did is important, but it's still only verifying the chucks taper in the radial plane. There's a bit more to it than that, and visualizing how ER collets and all the multiple surfaces involved are expected to work in all 3 dimensions is I think helpful. To start with, there's how well that chucks internal collet taper is aligned and sitting exactly square to the lathe spindle and lathe bed in the axial plane. As the work or tool length gets longer, some of the extremely hard to measure and tiny inaccuracy's get magnified to be more than noticeable. ER collets can close and grip the work or tool shank more than enough when there properly torqued. But if the nut threads or the same on the chuck body, the grinding on the nuts internal front taper, collet O.D. and I.D. or there tapers aren't all square to the lathes spindle and to each other, then as the collet is closed, it inevitably gets forced slightly askew. Collets are obviously flexible or they couldn't grip, its that same flex and the minor clearances that allows them to be forced slightly out of square within the chuck recess even when it might be 100% perfect. That drives your axial run out way up no matter how precise your chuck and collets radial run out happens to be. So the chuck or collets exact run out numbers don't always give the full information about what level of accuracy they can produce. But even so and just like even a highly inaccurate 3 jaw, if any shorter work is fully machined at a single chucking, then any inaccuracy becomes almost unimportant. The work will still be as true as your spindle and bearings allow. Or for longer and larger diameter work, there's still really nothing that can equal the accuracy of the old school method and turning between centers. I trust that over even my best chucks or collets.

ER collets also have quite rigid requirements for the minimum tool shank or in your case, the work piece lengths used. Too little on the lower end of the collets closing range and you can permanently deform or even break segments out of the collet. 5C and many other collet designs as an example close using a single front taper, so any short lengths have no real effect. ER's close and grip from both ends, it's a built in design feature so as much grip and rigidity as possible is obtained when tool holding which is what they were really designed for. So I use a minimum of at least 80% work or tool shank length inserted into the collet bore before I close them. I think that's worth a mention because for work holding, most of us are shall we say a bit conservative about cutting off and just wasting excess material.

One of the very best sources about choosing the various ER collet and nut types, using and maintaining them, and why you would do so that I know of would be here, https://www.youtube.com/@REGOFIX

For a bit of extra information and if the full development history of the ER collet design is out there, I haven't found it yet. After a long search I found a virtually new and fairly specialized but now long out of production radial milling head. Luckily it also came with a full set of the optional Schaublin 25 series collets the OEM offered. So I've had a chance that maybe not many others have had to measure and compare them against my set of ER 25 collets. From that and knowing Rego-Fix loosely based there ER collets on the original Schaublin design. It seems that way back in 1972 they kept both of the same collet nose and body tapers, but added a bit over .100" to that series length and mostly at the collets front for that extraction groove, enlarged the chuck body to allow for the larger diameter nut to gain the ability of pulling the collet enough to open.The Schaublin collets use quite a bit smaller chuck body and certainly aren't quite as easy to use, but I'd say Rego – Fix's real innovation was their much more complex and imo very clever nut design.

I just bought a Chinese MT5 ER40 collet holder for a specific job on my lathe and after much cleaning of the spindle taper it is the most accurate collet I have ever had the good fortune to meet.

The shank through hole looks like a rough casting and is as hard as the top of Gods' head but I cannot detect any run out at all on the collet.

So is this sample of one typical?

Are ER40s actually better than ER32s?

Do these things get less accurate as they diminish in size? As I descend into the "hobby" market and start counting the pennies?

Have fun,

Robin

Has anybody mentioned the runout between the bore and OD of the individual collets – there's bound to be some? I've never attempted to measure it on my ER 25 set because if I have a job where concentricity is crucial I either turn between centres or do my utmost to arrange for the relevant diameters to be turned at the same setting.

I have made a lot of collet chucks. The run out of the chuck (RC) is always far below 0.01 mm (0.0005". That is not a surprise because the collet taper is turned using the same lathe the chuck is used on. I measure this run out on the inside of the chuck!

The next to measure is the run out of a bar clamped in the chuck using a collet close to the chuck (RB). The RB run out is most of the time larger than RC but below 0.02 mm (0.001" because my collets are not that precise.

The last thing I measure is the run out at the end of the bar (RE). Generally, this is larger than the RB. It depends on the alignment of the head stock (HA) and the quality of the collet. On my lathe(s), the HA is about 0.015 mm/ 100 mm.

The thing to consider is how run out effects the result. If I can turn the part in one fixation, than it is of "no influence". If I have to reposition the part, and dimensions are critical, I use a 4 jaw chuck.

If I have to turn a long straight shaft, I use a live centre. I turn long shafts CNC and can easily correct for mis alignment of the head/tail stock.

If I need to turn a part between centres, I place a bar of mild steel in the chuck and turn a 60° tip on the end of the bar. Than the run out of the chuck does not matter. I correct any head/tail stock misalignment by turning the bar CNC.

Edited By Huub on 14/09/2023 21:59:41

"Beats me how LBSC and all ever managed to produce their creations without all the NASA and RR standards we expect of todays tooling to enjoy our hobby."

I am so pleased I am not alone in thinking along those lines. I simply don't 'get' all this fretting about the 'nth' degree of accuracy in my back-shed workshop. I take my hat off to those who can achieve it, and no doubt some models like a gas turbine require it, but my philosophy is if it looks right, fits right and works right then it is right!

I have bought junk on occasion but hope springs eternal so it is easy to get caught out. Probably best not to risk buying a new set of ER40 collets then, unless I am feeling flush and can afford high-accuracy

best

Robin