Y axis backlash story continues...

Made up a circular spacer from a piece of stainless TIG wire. In fact, I used 1.2mm wire in the end so that it was easier to locate in the space available. Then jacked up the collar again. 

The backlash was better now - 50um instead of around 130um ie 2 thou instead of 5 thou. But that's still crap enough to require more attention. 

Playing about with the DTI, it seems that the backlash measured directly at the ballnut mounting relative to the knee bearing is quite a bit less. And it varies if you lock the saddle and measure at different distances from the ballscrew. So there are clearly several things going on, including some significant contribution resulting from loose gib adjustment.

With the saddle (ie Y axis) firmly locked, the drive belt tightens and loosens as load is applied, so clearly some lost movement is happening there, despite what I thought. Clearly I hadn't tightened it up as much as I should have when reassembling it.

I then started to look more closely at the Y axis gib adjuster. This has a large slotted screw at each end, so you can drive it in or out by tightening the appropriate screw, taking care to loosen off the opposite screw first.

I was quite surprised how little movement of the gib was necessary to completely lock the saddle. I managed to achieve a complete lockup after only driving the gib in by a matter of several mm and had to knock it back out with a hammer and brass drift in the end. During that state of solidity, the backlash measured at the ballscrew pulley was negligible - any movement dominated by springiness rather than hysteresis (lost movement). You'd have to say that's a good sign.

I completely removed the gib strip to see why it was so reluctant to move. In fact, once the preload was removed, it slipped out very easily. Quite simply, the ways are cast iron on chrome and the wear seems to be almost non existent. And despite the very modest taper angle, when the gib bites, it locks the saddle quite suddenly. I suppose that's a good thing to find. I'll measure the taper and the movement range to get an estimate of how much it actually tightens against the ways. That should indicate how little the ways have worn over the years, given that the gib appears to be original and shows no sign of merging from the end of the adjustment slot, which is what would eventually happen. It can't have been adjusted more than a few mm from its original position.

It seems that the taper is 3.7mm over ~280mm. So 1mm of movement of the adjuster will take up a mere 13um of play. 

On the Blidgeport, I had to remove a short length (perhaps a couple of cm) from the end of the X axis gib to stop it poking out of the end of the table due to the excessive wear on the ways. I don't seem to be anywhere near that territory here.

The gib itself is beautifully made, like the rest of this machine. It's been hand scraped in and there is no sign of any wear, damage or gunk buildup. The oilway (through hole) was clear and the oil channel didn't require much cleaning.

I'm not an expert on these matters but I assume the green coating is some form of low friction finish. 

I had to remove the wipers from the front of the saddle to allow removal of the adjuster screw so I could withdraw the gib. They are composed of a Nylon carrier and moulded elastomer seal.

So:

• Replace the gib, adjusting it carefully to be at the point of biting. This is a bit more difficult to judge on a CNC machine without handles than on a manual machine but needs to be done carefully as it clearly contributes to the backlash.
• Tighten up the belt to eliminate visible movement under load.
• Retest.

Ideally I'd get the backlash down around the 25um / 1 thou mark. The original machine spec claimed 10um accuracy, although I've no idea what that actually looked like in practice.