Collet Chuck (Part 2)

en8 might well not be round enough - try turning it down a bit, then putting the turned end in the chuck.

Or am I teaching granny?

I'm down to 0.003 after tapping the backplate appropriately as I tighten the bolts on the backplate/spindle end of things - I'll just take abit off and try again...

Thaks

Steve

Turned the EN8 down to a nice finish Peter and now I get a repeatable 0.03mm :-((

I've tried a couple of different sizes of silver steel bar and the result is

0.03mm, which I think rules out a diff collet.

If I rotate the chuck on the backplate and measure the runout when the chuck is mounted in three possible positions there is no discernable difference, which I think rules out the chuck.

Not convinced there isn't something iffy with the backplate.

Steve

First, meassure in the backplate's cone, where the collet fits in. Runout zero? The collet is guilty. Try a different collet.

Have a look into the collet's bore. Ground or just drilled? Black or silvery. Chinese or ... quality?

Did you clean the collet? Sticky anti-corrosion gum won't help for precision.

Nick

Steve

You can't rule out all possible errors with the chuck body with the test you describe. the chuck collet centreline and the chuck register diameter could be out. Not likely but I have seen it. You can check this by fitting the test piece in and note the highpoint (.03mm deflection) position in relation to the backplate. Move the chuck round 120 deg on the backplate and try again. If the highpoint has moved position in relation to the backplate then the chuck body register and collet centreline could be out. If the error is in the same position in relation to the backplate there is something wrong with the backplate to spindle mounting.

Another way of testing or correcting if the chuck body is soft enough is to ensure your centres are absolutly in line and centre drill a test piece at both ends. turn the test bar between centres to a suitable diameter for the largest collet you can manage. Fit this test piece to your collet chuck, I assume that the test bar will go right through without the backplate fitted, and fit the bar back between centres. Ensure the test bar at each end is running totally concentric, it should be as you machined it. Test the run out of the internal register of the chuck body to see if the chuck body is actually causing the problem.

I'm not sure I've described this very well but it is easy to do with the parts in front of you. The fact that you can affect the runout by tapping the backplate as you fit it suggests there is movement between backplate and chuck, if so you should be able to use this to reduce the error. If the backplate to spindle register is in fact allowing movement then I would be concerned about ever getting consistent results with this setup. I've not had a good look at the C6 spindle so I'm not sure if this is of any relevance but it might help to give you a few more ideas to find where the error is.

Regards

Keith

I did this, but paid no attention to where the highspot was - so I'll run through this part again.

Sounds like this is the absolute check on what's going on. I haven't checked how true the lathe is running for over a year so I'd probably need to do a full check on the lathe, before starting off down this route. The biggest collet the chuck will take is 20mm nominal (ER32) so it's probably worth getting one of those to ensure I've maximised the setup.

The best runout is not repeatable and is obtained 120 round from where I machined the register in the first place (I marked it up following your earlier advice - thanks). It might be that both the backplate and the chuck are contributing to the problem - rotating the backplate on the spindle is definitely "tuning out" the run out to a significant degree.

Thanks for the post Keith, very useful. I'll have another run through and post back what I find. Might be away a while!

Best Regards

Steve

If you're fitting a new collet chuck it is wrong to use the runout of a test bar as the control measurement. This runout test cannot distinguish between chuck error, collet error and test bar residual error.

The fundamental test is the concentricity of the chuck internal taper. This needs to be internally clocked at two widely separated diameters - these should both clock within tenths. Two equal (in both amount and angular position) runouts shows parallel displacement. Unequal runouts indicates skewed mounting.

Pallel displacement is the most likely problem and easily fixed by using a loose fitting register. With the fixing bolts lightly tightened, gently tap the chuck into concentricity before final tightening.

For final checks with individual collets, dowel pins are convenient accurate test bars.

Jim

Thanks Jim. The runouts are equalt in an angular sense, but not in size. Taking care to have DTI working at the same spot on its travel the runout difference is less than 0.01mm but I suspect you were speaking of 10ths of a thou.

Repeatability onto the spindle is an issue - at least 0.04mm on re-mounting the backplate.

Steve

Steve

Unfortunately you seem to be chasing your tail with this collet chuck and you have lost me as to what exactly you have done. As Jim says the test on the internal spindle excludes the collet and test bar as an error but very much leaves both the backplate/spindle and chuck body/ backplate as possible errors. If you have machined the backplate register on the machine and obtained zero rainout then that should have eliminated the spindle/backplate interface, but ONLY if that fit is repeatable. One of the issues I had with these things in the past is that you fit the backplate and get it running as accurately as possible, machine the required register and mark where the backplate fits. The problem then is that you have no face on the backplate to use the DTI on when the chuck body is fitted so no way to "tap" to the original position if that is necessary. When I cut the register on the backplate I then also take a light cut off the outer diameter of the backplate that is visible when the chuck body is bolted on. This way I have a face to check on with the chuck bolted up. Without this you have no way of telling if the backplate is exactly as it was when you cut the register. If as you say yours has a loose fit then it is a lottery as to what error you will get.

The figures I posted for my setup in your original question were with the components assembled as supplied. While mine were not bad, after fitting properly to your own machine you should be able to get an apparently zero runout on the collet seating taper. Unless of course you are using a very, very expensive DTI. After fitting on mine, the backplate reads zero runout on the register outer diameter, flat face that the chuck butts against and the outer diameter for checking. As long as I keep the threads and contact faces very clean this is repeatable on backplate removal/refit within the resolution of my Starrett DTI. With the chuck fitted to backplate I again get no apparent runout on the collet taper but with collets fitted I can get a variation on the test pieces of maybe 0.0002-3" depending on the particular collet being used. My idea of no runout is that my 0.0005" Starrett doesn't move more than the needle width so most likely it is within a couple of tenths of a thou.

displacement between backplate and chuck body. While cutting the backplate register/outer diameter I also check on the flat face the chuck body contacts to ensure that it also has zero runout, if not (often the case) I take a light facing cut off this surface. This ensures the backplate rear face (to spindle) and front face (to chuck body) is absolutely parallel at the fitted position. Another possible cause of the angled chuck is again the register, if that is not entering the chuck body fully it can hold it off minutely at an angle. Many people make this fit quite loose so that it can again be tapped into final position before the bolts are fully tightened. I have found on many occasions that the chuck body internal register may have a burr on the outer edge which needs removing.

Steve, I'm sorry about the length of this post but I'm not good at elegant prose. My main point would be that you have to sort this out methodically and eliminate each cause of possible error in sequence. If you cannot get the backplate to spindle fit to repeat accurately (even if you need to tap and test with the DTI on every fit) then the runout at the collet is going to be a complete lottery. Apologies if this is "teaching granny".

Best regards

Keith

SNIP

SNIP

Keith has given you lots of useful advice. A few extra comments may help.

I was brought up on Imperial units so tenths of a thou is my natural measure - it somehow seems a lot less clumsy than 1/400 of a mm!

With two unequal runouts at the same angular position.

You have both parallel displacement and cockeyed mounting (i.e the backplate mounting is not quite square to the spindle axis).

Tap the backplate until one runout is zero. The runout on the other test plane shows the amount and direction of the angular error.

Repeatability of re-mounting the backplate to the spindle nose is of prime importance. Until you've reduced this to a satisfactory level you'll be chasing your own tail in correcting the chuck position. At this sort of level absolute cleanliness and burrfree mating surfaces is vital

Jim

1/1,000th of a mm, or a micron, is a useful measure too.

Of course it doesn't look as good when you measure something and it's only accurate to three or four (or twenty) of them rather than one, but still ...

Good electronic micrometers will measure to a micron, and even if you are using a 1/100th mm device like a digital caliper, you just have to add a zero on the end.

And "Accurate to 3 microns" does sound okay :)

Much less clumsy than 0.000118110236 inch too ;)

Tim

Is that larger or smaller than a Gnat's left bollock ?

.

-- Regards,

John Stevenson Nottingham, England.

Visit the new Model Engineering adverts page at:-

Depends which side the gnat dresses

Regards, Tony

I suppose it's inevitable that my preference for Imperial measure would not be supported by the metric minded. I can well understand that, brought up on the metric system, Imperial units can seem to be outdated and irrelevant.

However there are a few occasions where Imperial numbers just fit better.

With the average home workshop tools it's not too difficult to achieve one thou accuracy. One tenth thou marks a different region. It's close to the limits of standard mics and small lathes and beginning to call for the use of gauge blocks and grinding/lapping techniques.

On this basis "a few tenths" is a conveniently vague term for an accuracy that's reasonably achievable in the home environment.

"Accurate to 3 microns" implies an accuracy that does not exist.

"Within a few times 3 microns" echoes the clumbsiness that gave rise to my original comment!

Jim

P.S. For the professional community I of course agree that the the micron is the natural small unit - 1 thou is far to big!

Hello All

I have a friend who is a professional turner and is young enough to have been educated entirely in metric.

I find "a fortieth of a millimetre" translates well.

Russell

Clean sheet!

Got a new back plate.

Mounted back plate onto spindle and took a lick off the perimeter of the back plate so I had a smooth concnetric place to run the DTI (£20 one not lab standard)

Took a lick off the face of the back plate to ensure everything was square. So now the perimeter and the face are known to be true.

I have a pop mark on the spindle and pop marked the backplate to line up.

Took the backplate off the spindle, and put it straight back on pop marks lined up and tightened it up - runout 0.03mm

The register as supplied was smaller than required by the Collet Chuck spec, which was an easy way to pick up Jim's suggestion.

I took the backplate off, mounted the chuck and put the whole lot onto the spindle and tapped the backplate gently to run true and tightened it up. Then I tapped the collet chuck into place and tightened everything up. It all runs true.

This proves that there is a problem with the backplate to spindle fit but nothing about the collet chuck runout.

Thanks for the advice I now have a logical method for chasing the problem through.

Regards

Steve

Well done Steve, good to hear that you are making progress. I don't know the C6 spindle mount system but judging from the pictures here

it would appear that concentricity relies purely on the fit between the parallel spindle register and the backplate internal register. If so then I have experienced similar problems with some pre-machined backplates I've bought for my Myford. Some suppliers seem to machine the backplate register with a little too generous a clearance. So much so that I now regularly measure the diameter before I buy. You might like to measure the internal diameter of your backplates and see if they vary at all. If they are all a little over size you might need to bite the bullet and make one yourself just for the collet chuck. I suspect that with a normal 3 jaw most people wouldn't notice a 0.03mm variation. As has been said before absolutly clean surfaces is essential.

One way I rescued a Boxford backplate with an oversize regsiter was to machine the internal diameter 4mm oversize and I shrink fitted a sleeve in place, although a push fit and loctite would do just as well. this allowded me to machine the internal register diameter again to a much closer fit on the spindle. Sounds a bit of a bodge but it has been in use on my Boxford for the last two years with no problem at all.

Now you have the collet chuck firmly mounted to the backplate it might be easier just to "clock out" the small error each time you fit it.

Best regards

Keith

OK, so why not fix the backplate? Put it in a 3 jaw chuck, align, face off a tad, lick over the centering cone. DO NOT unchuck, but take the whole setup (chuck and backplate) off the spindle and reverse it to see how well the backplate fits. Repeat if necessary.

That should work, unless I misunderstood you.

Nick