Notes on Workshop Techniques

4.1 3-Jaw chuck

The modern self-centering or scroll-chuck works by having an accurately machined spirally grooved plate and matching jaws with teeth that engage the spiral. The teeth need to be of a special design because, as the jaws move towards the center, the radius of the groove naturally gets smaller. This design requirement reduces the contact area allowable between tooth and land of the groove to a single edge. Hopefully, when wear eventually occurs it will affect all the teeth on all 3 jaws to the same degree such that the self-centering ability is not seriously impaired. Of perhaps more worry is wear in the radial slots which locate the sliding jaws, wear here is often the result of 'sprung' jaws caused by over-tightening a 3-jaw chuck - particularly on workpieces supported only at the outer jaw tips.

Chucks are attached to the spindle by a 'backplate' to which the chuck itself is bolted. This backplate usually has an internal thread and plain register (or camlock and cone, or one of several other designs), the accuracy of the register is of fundamental importance to the true running of the chuck. It's probably best to buy the backplate (if not supplied with the chuck) ready machined to fit the spindle, the only jobs then remaining being to machine the front register onto which a step in the back of the chuck locates, and to drill the 3 holes for the securing bolts. On some imported chucks it may be that, even if the chuck register is accurately machined to match the backplate, work gripped in the chuck still shows considerable run-out. If this is excessive then there is little to be done and the chuck should be exchanged for a new one, if it's only about 0.005" TIR then it's possible to correct for this by machining the register undersize, reaming the bolt holes slightly oversize, and (with a piece of 1" silver steel held in the chuck jaws) setting the work to run true with a DTI before tightening the securing bolts. This of course will leave the body running eccentric but this can either be ignored (it's irrelevent to the functioning of the chuck) or machined down using a carbide tool (assuming the body is cast iron). In fact , the odd thou or two run-out can often be eliminated by slackening the securing bolts and thumping the chuck body in the appropriate place with a lead- or hide-faced mallet before re-tightening. All the foregoing assumes the chuck has a separate backplate, chucks with internally-threaded bodies cannot be adjusted in this way unless dissassembled and the chuck-body register machined undersize.

At time or writing I had three 3-jaw chucks of varying vintage. The two oldest were manufactured by Pratt Burnerd, the new one of Polish origin which came with the new lathe. Even the oldest is accurate to within a couple of thou run-out (after some remedial action) and it must be about 15 years old. The most accurate is the Polish chuck, probably because it's new, but it appears well made and I would not hesitate to purchase another chuck of Polish origin. The only problem with the latter is that the hole through the center is smaller than the Burnerd chucks, so that the body will not pass a 1" bar.

I have two independent 4-jaw chucks, the small one (4" diameter) is of little practical use but was purchased because it was the only one available at the time to fit my old ML10 lathe and I didn't know better. The other, a standard medium-duty 6" chuck is far more useful. On gap-bed lathes it is as well to get a threaded-body chuck to minimise the overhang from the spindle nose, this not only puts less strain on the bearings it also provides more room in front of the chuck for mounting work. A backplate chuck in the same circumstances will have an inch or so less working space.

Most 4-jaw chucks are provided with the front face of the body engraved with concentric lines to aid centralising work. These rings are frequently far apart though, and it's worthwhile removing the jaws and lightly adding additional grooves with a threading tool set pointing towards the headstock. I've seen one suggested addition in the form of 4 additional keys, linked in pairs by a spring wire, the theory being that it's quicker to set work by twiddling pairs of keys. The keys are removed for actual turning work.

Neither do collets have to be designed only to hold cylindrical work, for those who need them you can get collets to hold hexagonal and square barstock, but these are mainly for use in production machinery where their cost justifies their existence.

Collets find most use in watch and clock making where there is a great deal of re-chucking fine spindles that have to run true. I would very much like a full set of collets but, given the cost and my particular needs, I have difficulty justifying their purchase. If I ever happen to drop across a used set I'll probably buy it, but until then I'll do without. An alternative for one-off jobs is use the old 'split bush' method where a cylinder of steel is bored and reamed in the 3-jaw to exact fit on the work, the cylinder is then marked so it can be replaced back in the 3-jaw in the same position, then it's split lengthways with a slitting saw. Work gripped in this bush in the 3-jaw runs very close to concentric and is one method of utilising an eccentric chuck (Note: I didn't say worn chuck - a worn chuck will not repeatedly hold a workpiece in the same position so this trick is no help, and anyway, see above for correcting an eccentric 3-jaw). A variant on the split bush theme is to machine a bush so that it accepts the across-corner dimension of square barstock, I have made a selection of these in the past (for 1/4", 3/8" and 1/2"), useful for quick and dirty turning of square bar (where I can't be bothered to set up properly in the 4-jaw).

It's important with standard collets that only work of the correct size be inserted, any attempt to grip work slightly smaller than the correct diameter is going to damage the collet, and it won't run true anyway so it's not worth the effort. Same goes for the flexcollet - stay within the specified range.

Other specialist chucks are related to the common chucks, like the 6-jaw and 4-jaw self centering chucks. The latter is particularly useful for turning square work but understand that the work must be dead square (in cross-section) or only two of the jaws will actually be gripping. The 4-jaw S/C chuck can be used for round or octagonal work of course, but again there is a danger that only 2 of the 4 jaws will be doing any work. There is usually enough 'spring' in the jaws to account for a few thou off-square. There are also "two-jaw" chucks which are really vices designed to be held on the faceplate.

There are many complex chucks which deserve a book all to themselves (some have..). The ornamental chuck is a classic example consisting of layers of geared slides, work mounted on the top surface describes complicated and convoluted orbits and a single-point tool will cut out a very fancy pattern. Bit like the 'Spirograph' toy I suppose, very artistic but not particularly useful. Other complex chucks have a more substantive function, the eccentric chuck for example enables eccentrics to be machined by sliding the work over to one side under control of a micrometer screw.

Once mounted on the angle plate with a datum surface square to the lathe axis the angle plate can be slid around the faceplate and the square alignment will be maintained. If the plate and workpiece are large it's useful to be able to remove the faceplate and lay it on a flat horizontal surface for bolting up, makes it easier as the work is less likely to fall off onto the lathe bed during this procedure. A useful extra is a machined spindle nose mounted on free-running bearings, it's then easy to test for balance of the setup. It's always necessary to balance the plate to prevent serious vibration problems. When satisfied that the workpiece is in the approximate position required, balanced, and the clamps are secure, the faceplate can be mounted on the lathe (use a piece of wood or rubber matting to protect the lathe bed in case of accidents).

I also have a rotary table on which I attached a Myford threaded spindle nose, apart from chucks the faceplate can also be screwed onto this and is very useful method of profiling the outside of model locomotive cylinders after boring.

I happen to have several accessories on which I've machined a replica of the spindle nose (register and thread) and I used one of these during machining as a test gauge. If you do not have such an item then consider machining one especially for this type of job as it makes life much easier.

First, mount the cast iron backplate casting in the 4-jaw held by the boss with the flange set to run as true as possible. Take a cleaning cut across it's face deep enough to get beneath the skin of the casting in one bite to avoid blunting the tool, and also across the edge to leave it true and circular. Reverse the casting in the chuck and set it to run true again (use the flange as reference - but no need to take a DTI to it as it doesn't need to be exact). Turn the boss and rear face of the plate to size based on the measurements of your current chuck. Center and drill right through about 1/4" diameter, followed by a larger drill to give acces to a substantial boring bar (I used a 3/4" machine drill with No.2 Morse shank). Measure the inside diameter of the threaded portion of your chuck and bore through the casting to produce this size of hole. Thread the hole all the way through (1-1/8" x 12TPI for the Myford) using the lathe's screw-cutting capabilities. Take care that your threading tool is ground to the correct included angle, that it is sharp, and that it has a small 0.010" flat (or radius) on it's tip. Use a thread gauge to set it in the tool-holder. This is a substantial thread to cut and if you haven't cut one this deep before it's worthwhile reviewing my notes on the subject. The objective is to get the threading tool to cut on face only (the advancing edge), and provide this cutting edge with top relief so it cuts more cleanly. Continue cutting the thread (2-5 thou per pass) until the gauge is a firm screw fit. If you are confident of the accuracy of your gauge you can consider the screw-cutting process complete. I was not so confident so I unscrewed the chuck, turned it around, and tried the thread on the spindle before removing it from the chuck (in this way, it can be refitted to the lathe spindle and, if it's still too tight, another scrape taken off the thread). This particular thread should not be too tight, it needs to turn freely by hand pressure alone and stop positively at the back face of the spindle nose register. The thread's function is not to align the chuck (the plain register does that), but to secure it to the spindle nose. If the thread is too tight and the register too slack then it's even possible the chuck will not seat truely every time it's fitted.

The next job for the Myford fitting is to machine the plain register. Measure the depth required for the plain portion and add about 1/16" to allow clearance where the thread ends. Bore the thread out of the end of the boss to this depth, and to about 50 thou under the final bore size. This dimension needs to be very accurate, so approach the required bore size using small incremental cuts and aim to produce a very fine finish. Use a plug gauge to ensure you don't go oversize - remember, you'll only get one chance at this, and an oversize hole will scrap the part. I aimed for 0.001" undersize and removed the remainder with emery cloth, repeatedly reversing the chuck and trying the job on my actual spindle until I was happy it was a perfect fit. Chamfer the corners well so that future dings do not interfere with it's fitting, and take a final skim across the end of the bore to be certain it's square to the bore.

All further turning will be done with the backplate screwed onto the lathe's spindle nose. The chuck/body register (i.e., the interface between the chuck and the backplate) is around the periphery of the chuck's rear face, and also the inner edge of the large, shallow recess. These are the datum surfaces which should be concentric with the chuck jaws. The backplate is usually held to the chuck using 3 or more bolts. Normally, one would machine a step on the flange to fit inside the recess in the rear of the chuck, and this dimension would be a tight fit thus ensuring work held in the chuck jaws runs truely. With premium quality chucks you can rest assured that this rule holds true, with cheaper imported chucks (as this one was - a Polish TOS chuck) one cannot be so sure that the body and recess are concentric with the jaws. I decided to make the step 0.020" undersize (to allow +/- 0.010" radial movement) to enable fine alignment. The bolt holes drilled in the backplate were similarly oversize for the 8mm securing bolts. Also, ensure there are no burrs on the rim of the chuck's rear face where the 3 bolt holes are - my chuck had surprisingly large burrs caused by (I assume) an automatic tapping machine. These would definitely interfere with correct registration so I took time to remove them with a fine file.

The chuck body was lightly bolted to the backplate (I know 'lightly' is an obscure definition - it should be tight enough to register correctly yet loose enough that it can be moved around with taps from a 2lb lead-faced hammer). A piece of 1" diameter silver steel (ground drill-rod) was clamped in the jaws. For this sort of precision measuring job always tighten the chuck using only the bevel gear marked No.1 (or the one closest the jaw No.1) as this was the one that would (should) have been used in the factory during the finish grinding process. It does make a difference. A DTI was used to indicate runout of the rod, and the lead hammer to tap the chuck body until it indicated true. The rod was then removed and replaced to check for repeatability, and if a random error proved evident a mean position chosen. This particular chuck ran within 0.002" which is pretty good for a 4-Jaw self-centering type.

With the chuck positioned correctly on it's backplate some Loctite #601 was run around the edge, this seeped some way into the gap between the two, and the bolts were finally torqued up to their full weight. The Loctite in combination with the bolts will prevent any chance of slippage under any conceivable normal turning operation. If it's necessary to remove the backplate sometime in the future a combination of gentle heating and a thump with said lead hammer will release it.