Interesting. Let's suspect the measurement system first – it's easier to play about with it, and metrology holds many traps for the unwary. I suggest that you replace the milling cutter with a rod which has a small rounded end (ideally), like a conventional DTI probe. Ensure the tip is as co-axial with the spindle as possible. Repeat the experiment. You could also do it under power, with no risk of damage to your indicator's contact pad.

The FB-2 quill is pretty simple. IIRC, thrust is reacted by a pair of deep-groove ball bearings (they might be angular contact), with the outer race of the pack retained in the quill by a screwed ring, and the inner by a collar screwed to the spindle shaft. I don't think there is any provision for adjustment, so the bearings ought to be a matched pair. Anyway, your test shows no significant axial free movement, so I suspect the measurement set-up, rather than the machine – at this stage.

It almost looks like something is unscrewing.

Can you isolate the measurement to just the components of the quill?

Martin.

The two matched deep groove ball bearing races were specially made for Emco. They have markings on to ensure that they are assembled correctly.

In the past I have replaced these with angular contact races, but with no real gain as regards machine performance.

Any preloading of these bearings can stall the machine on the highest speed.

The movement on the clock from applied external pressure is to be expected, this is a bench top milling machine remember.

Any checks for spindle play should be done with the clock fixed to the head of the machine, or with the Quill removed from the head, and held in the machine vice, with suitable protection. I doubt any play will be found.

There are a number of factors at play with this set up.

Is the top surface of the measuring device perfectly parallel to the table surface?

Is the end of the Cutter perfectly parallel to the table surface?

Is there any run-out on the collet?

All of these three defects can give spurious results.

Personally if I swing my clock over a 200 mm long parallel on the table, (tramming) and get zero deflection, then that is good enough for me.

Regards

Gray,

The test tells you nothing about the mill, as you don't know if the corners of the endmill are ground symmetrically.

Andrew

Is the spindle spring pushing the spindle down or up? Up is usual for drilling machines but it should be down for milling.

Martin C

Thanks for the second video. It's impressive and pleasing when someone acts so swiftly on advice given. Now I'm getting worried, because your measurement set-up does seem blameless. It would be even better to remove the possibility that the collet chuck is the culprit, but that seems a very long shot. Perhaps you should try again, turning the spindle from its upper end. It's certainly a puzzle.

It was probably back in the '80s when I was interested in obtaining a replacement spindle, that I discovered that Emco would not supply the component parts of the quill/spindle assembly separately. That probably explains why the exploded diagram, to which you provided a link, shows the quill/spindle as a complete assembly. The parts diagram I have in front of me is probably an earlier version, which shows the internals of the quill. The thrust bearing arrangement, which controls the spindle's axial position (z-axis) is as I described in my previous post. Unfortunately, it's not shown in your diagram.

I'm not sure what the spring (25) does. I think it acts between gear 12 and part 19 ('pressure washer' ), to reduce possible rattling of the gear on the spindle splines. Hopefully Graham Meek will be along to tell us more – he knows vastly more about these machines than I do. I think the gear is helically cut. This means that it will try to move in one direction along the spindle when torque is applied, and in the opposite direction when torque is applied in the opposite direction. Perhaps the gear's axial float is transmitted by the spring to the spindle, but it still doesn't explain why the spindle is free to move axially.

I have noticed that, in some gears, the spindle can be turned in one direction more easily than the other. I assume this is because of end-float in one set of helical gears, but I haven't gone hunting for an answer… Having seen your videos, I'll go and check my FB-2, when time allows. Perhaps, like the garage mechanics say, 'Oh, they all do that, Sir.'

It's beginning to look like one of the screwed bearing-locating collars is loose, but why you can't move the spindle by applying axial force is a puzzle.

Edited By Kiwi Bloke on 07/02/2022 10:08:58

Edited By Kiwi Bloke on 07/02/2022 10:09:39

Interesting..
 

Edited By Chris Mate on 07/02/2022 12:18:13

Edited By Chris Mate on 07/02/2022 12:19:01

It seems to me that the (preloaded ?) ball thrust bearings locating the shaft in the z direction have some slight assymetry in their races causing the balls to track in a slightly different position depending on direction of rotation. Should perform OK though as the spindle usually only rotates in the same direction.

To isolate the components of the quill, you can attach the DTI base to the head casting.

Then locate the measuring finger on various parts of the quill, including those that rotate and in addition, also to those that should remain fixed.

Hopefully, this might help pinpoint what is happening.

Martin.

Further to your last set-up I have tried the same set-up on my FB2, which has a brand new spindle and angular contact bearings instead of the Ball races.

Using a clock similar to the one shown in the last video, I had 0.03 mm deviation but in the opposite sense. Using my Mitutoyo clock in exactly the same way I had 0.02 mm deviation, but again in the opposite sense.

During these test I had a DTI resting on the spindle nose flange. While this face is not ground after heat treatment of the spindle. I had a consistent 0.01 mm deflection which ever way the spindle was rotated.

The next test was to use a Dial Gauge or "Drop Clock" to check this out. This was set up in my boring head to get the ball stylus running absolutely true on the machine centre-line. With this test I had 0.005 mm variation during the two opposite rotations.

As a last check I used the Dial Gauge held in the Machine vice and indicating on the end of the Spindle. I had the same 0.01 mm deflection I had earlier irrespective of which way the spindle is rotated.

All test were carried out using a long Allen key in the draw bar socket.

One problem I have learnt over the years in my profession is that the lever type DTI can give two different readings when presented to the work in different directions. It is due most probably to the clock construction.

This I believe is what is at play here.

Lastly, while the Lever clock has a good range of adjustment on the stylus. Its accuracy as a measuring device is limited. Most Manufacturers used to stipulate a maximum angle either side of the centre-line, or in-line position, after which the scale on the dial was meaning less.

Regards

Gray,

Hmm. Following Graham Meek's words of wisdom, I think I was premature to consider the measurement set-up blameless. My theory about helical cut gears and end-thrust can also be safely ignored. I'd still like to understand what the spring is for, though.

I have just set a 0.01mm per div. plunger-type DTI against the end of the spindle, with the mag base on the milling head, quill locked. A quick and dirty test. Turning the spindle in each direction gives about 0.01 mm TIR, and I can't induce axial movement of the spindle by gorilla tactics.

I'm not familiar with tool-setting indicators, as used in the first video, so can't suggest how one might contribute direction-specific error. I don't buy the suggestions about indicator probes bearing on a surface that is not accurately at right angles to the spindle axis, in both orthogonal directions. An inclined plane would show up as a cyclic indicator deflection, in either direction of rotation, not a fixed deflection, as demonstrated. However, it's common experience to see a stable indicator deflection dependent on the direction of travel of the surface wrt the probe tip, for what should be fairly obvious reasons.

I suppose the pragmatic answer is to stop worrying, and register it as a curiosity, so long as you continue to have no problems with work produced on the machine.

Going off at a tangent… In an application like this, where the needle roller bearing and spindle are selected for desired (lack of) clearance, presumably a degree of radial preload is desirable. How then is the spindle assembled into the bearing? Will it slide in, or has it to be 'screwed' in? And if the needles tilt a little in service, perhaps because the shaft is not precisely parallel, will this induce axial deflection?