I think the mounting face of the collet chuck is not square to the axis of the collet. As you have a test bar,which I assume has a morse taper and a long parallel portion, fit the test bar into the lathe spindle taper,then part the chuck from its backplate fit a collet to the chuck that will grip on the test bar,no need to pull the collet up dead tight. set a dial indicator to measure runout of the back FACE of the chuck,if it runs true it will at least prove that the axis of the collet is true to the mounting face, I have assumed that you have examined the taper in the chuck where the collet fits and there are no burrs or damage to the taper or the collet,does the runout fault occur with all the collets?,the collet that you first used may be faulty,Should the mounting face prove to be at fault,either return the chuck or mount a piece of steel bar preferably greater than the size of the largest collet available,turn the bar to suit the collet,also the bar will be running true,mount the chuck on the bar via the collet a lightly skim the back face, this action may solve the problem but also invalidate the guarantee.

I had similar inaccuracies when I made my collet chuck back-plate: Insufficient tightening was the main reason for my measured errors: I tightened the collet closer more fully and the errors all but vanished.

James.

If you cannot improve things with the above, then you need to further prove out the ER chuck.

Then need to make a tapered mandrel, same taper as the collet location, drilled and tapped to enable clamping back. Initially set compound over by clocking the internal taper of the ER chuck.

Carry out fine tweaking if needed to get a perfect bed using Engineers blue. Clamp back and clock face and recess for true. Any errors are then further magnified at the outer recess diameter and adjacent face.

When using ER collets on a CNC mill at work we had a bench mounted tightening fixture to ensure they could be pulled up tight. Also do not assume 12mm silver steel is going to be round and straight, it very rarely is.

Too late to edit, but for any collet that I want to be running good, I have been buying the high precision collets from Regofix. In saying that, I just tested a cheap set of collets that is going to be used to hold parts for milling hexe's etc, so does not need to be all that concentric, just has to hold the part. To my surprise , the 4mm,6mm, and 8mm were better than 0.01mm at the collet face and at 30mm out from the collet face. This is a cheapo ER20 set. For my cnc wood router I use only regifix and it is ER20 also . I tried the Regofix nuts and the regular nuts on the 6mm collet. Of all the nuts we have, the regofix and 3 of the no brand were very good. 3 of the no brand made the collet not run true at the 30mm mark and next to the collet face was showing nearly 0.02 run out and .06mm run out at the 30mm . This is the worst of the no brand ER20 nuts.I have now identified those nuts and will use them on the milling fixture for Non important things. While at it, I checked with an indicator on a known bad collet for concentricity. As I suspected, it did not run concentric at the front to the back of the collet. It has a reading difference in the front to the back of about 0.03mm. I used a red marker for the front high side and a blue for the rear high side. They are almost opposite, which makes sense and explains why the test piece with that known bad collet is no good. What I will do is set up and regrind the id concentric and see if I can save it. It will become then a special sized collet effectively. I will put that onto the oneday list.

Hope this info may help some people out there. I have read about the high torque required to hold stuff with the ER40 collets, and for what I do with them, have not found a need to tighten any more than what I can on the spanner lengths that are supplied by Regofix or the other spanner suppliers for the various ER sized systems. Granted ,most of the stock I hold is turned finished bar in most cases, or is Carbide cutters.

Neil

Micheal,

Why do you need a spanner in each hand? The S7 has a spindle lock. Undoing the collet may be more difficult!

A pair of spanners can allow a zero torque on the spindle and machine tool. Although they have the huge toque value for the ER40, I have only used about 30ftlbs and it is fine for every thing I have been doing. The smaller ER 20 I use less than that. Never had a cutter come out from ER40 or the other smaller ones. I also don't go trying to take the max possible cuts etc either , when holding a cutter or a work piece.

Neil

.

My advice would be to send both the chuck and collets back. – I initially purchased not knowing any better quite a bit from that company. – Everything (except for a vertex milling vice) has had to be replaced or needs to be.

Then purchase a chuck and collets from somewhere that does quality control to a reasonable standard themselves and does not rely upon the end user (the customer) to do it for them.

I have an ER collet chuck for my lathe purchased from ARC and it is a excellent bit of kit by any standards. Especially for the money.

Nick

.

Cheap test bars are available from 'Oxymoron Enterprises'

[ best known for their Chocolate FireGuard ]

MichaelG.