Lathe jumps out of gear

I wittered on about coarse and fine adjustments in an earlier post. I've marked two lock-nutted screws in blue which I'd investigate.

The one at the rear attaches the fork to the actuating rod, and the angle of the fork is probably adjustable. It may be the nut grips the shaft directly and has shifted, or it sits in a notch that's worn, also allowing the fork to shift.

Dave

Thanks for all the suggestions. Here are some answers to the questions.

There is absolutely no play between the knob and the selector fork - the knob is attached to the shaft with a roll-pin and the lock-nuts at either end of the shaft are good and tight. The are are no flats on the shaft. The slightest twist of the knob results in a corresponding movement of the fork and there is no lost motion. The play between the fork and the gear it moves is no more than would be expected.

When the gear drifts to the right, the shaft, fork, and knob move in unison. There is no lost motion. It certainly appears that the fork moves the gear.

Although the lathe is more than 20 years old the the gears are very seldom used for speed changes 'cos I use a VFD just about all the time.

Hello Gary,

That is all very helpful information and it takes the cause of your problem straight back to the indent balls and their fit in the dimples in the housing.

One of the pair that are close together is the dominant one and I think the only way to determine which it is will be to remove the springs and balls in those two positions, fit the lever control and make some careful observations with a good light down the drillings when the gear is shifted into the two positions it takes.

Then you can act knowing what needs correcting. With the new information you have added on the lack of use the mechanical gears have had in the time you have owned the lathe, it all points to a fault built in during manufacture, just as you suspected. 

Regards Brian

Edited By Brian Wood on 30/12/2022 09:50:12

I suspect that the issue might be that that the '3 O'Clock' (looking at the lever) hole has wallowed out a bit, and shouldn't be able to align with the detent ball in this position - there is visible wear in this hole in the original photos (the hole is at 9 O'Clock when looking at the lathe). If the detent is '1/2 a hole' out, it shouldn't engage.

How well do the balls fit in the holes in the handle? If they have room to wander, they will be able to engage in the detent holes sooner than they should. (If they are loose, the holes in the handle may be able to be reamed out to take a slightly larger ball, or a small, round ended dowel - made to suit.)

My 2p, anyway.

Edited By Andy_G on 30/12/2022 10:28:56

I've come late to the party but the pic of the dimples in the head show noticeable signs of wear and could well lift the ball out of the detente and stop the train. Hard balls on a soft iron casting ? Poor tooth contact due to a failure to fully engage won't help. Noel.

Are both the layshaft bearings still good?

It's hard to see why a gear should start drifting when the gear lever has hardly been moved in 20 years even if the detents are a bit 'off'.

Good grief! 3 days and this trivial problem not fixed. You are as bad as me. I often say it takes me an hour to boil a 3 minute egg (HSE warning the recommendations are probably an hour boiling now to avoid listeria).
Just move the lever vertical, re-drill the roll pin hole, choose a single detent and make 3 new dimples in the clear area.

I suspect the designer's original intention (on the German designed lathe they copied) was to have 2 detents holding the mid position to give a very firm hold when the compound gear is out of mesh to avoid it being accidentally nudged when running and grinding the gears. Some twat in the Chinese drawing office also started with the left hand hole and then found the right hand hole clashed with the lever so shifted it a bit instead of redrawing the first hole. The manufacturing process involved a bloke on piece rate with a pistol drill who didn't care if the detents were misaligned.