Are you saying that some of your moves are 20 thou of an inch out, which is how I am reading it. We have jet engines over on this side of the pond, a local firm in Barnsley (of all places) markets them - and they are looking for accuracy down to 1 thou.
I think - instead of messing about with bit and pieces, if you are to get down to the accuracy required, you will have to set out a plan and stick to it, otherwise you will do one bit, and it will upset another.
The first thing, as has been said, is to check the gibs on your table. Depending on where the table is pushed or pulled, if the gibs are slack then the table will "lurch" as it changes direction. On my lathe, it lurched on the Zaxis, and affected the X axis diameter it was cutting. If you have not done gibs before, by all means read up on them, but they are relatively simple - a series of screws pushes a metal plate in, against the running surface of the table, taking up the slack. Unfasten the locknuts, and (mine have an allen key) gently tighten the inner screw up - not tight, just take up the slack and a fraction more. Tighten the locking nut again (without allowing the inner screw to move).
It is a bit difficult to measure the effect of this, but if all is nicely "gibbed up" movement of one axis should not alter the setting of the other, particularly on a change of direction, but you should still retain smooth, even movement. Be careful on the table extremeties, where there is less wear, becasue it is easy to overtighten these and cause the table to stick or at least loose steps.
The next thing is to set your steps per unit correctly. This does not mean measuring anything - it means calculating (sorry BobShop - I do not use the Mach 3 function - it is open to inaccuracy). The calculation is simple - (motor steps per rev) multiplied by (microsteps - see drive) multiplied by (gearing - direct drive = 1) multiplied by (pitch of leadscrew). This should round off at a simple number (unless your leadscrew is something you picked up on a second hand stall) - mine is 60,000 - 200(motor) x 10(Gecko microsteps) x 3(gearing) x 10 (leadscrew 10 turns per inch), You can see from this I have 60 steps to 1 thou of movement - which is good for accuracy. You may not have the gearing so you might be down to 20,000 - still 20 per thou, (although you cannot guarantee the accuracy of microsteps)
Once you have done this you can test your movement. With a digital caliper, you can test the movement of you table over say 6 ins, using the MDI. move to the right, zero your DRO's, then with the MDI go in 1 inch steps to the right and check each distance. They should be accurate to within say 2 thou. The purists will say this is not good enough - but, quite frankly, I defy them to measure any more accurately with a caliper. The "error" if any should be consistant (+ - 2 thou) throughout the length. If it is consistantly accruing or loosing thous as it goes along - then there is a fault - and this is probably in your leadscew - all other things being simple numbers.
If this is the case, then (and only then) can you try making small adjustments to your steps per inch to correct the matter. This is extremly rare, and really I do not like even mentioning it, becasue it confuses the issue.
The above measurements can be taken with the "backlash compensation" on or off, but to adjust for backlash you need the compensation OFF.
It is very simple. Starting at 0.0 - on the MDI go to X1 (to get rid of backlash). Set up your caliper and zero the DRO's. Do a G1 on a low Feed Rate say 4 inch per min and go X2 then X1. In theory the table will move right then move left and be at the starting point. It will not be and you can read the distance on your calipers. The missing distance is the number of steps that the gearing and leadscrew took to get their shoulders ready to push the table the other way. You can, if you wish repeat this and take an average - since you are measuring - but if you are careful one will do. This is the backlash compensation and should be entered in the backlash table and compensation switched on. If you do averages - do not forget the initial move to get rid of backlash each time.
The only things about backlash is that it will not be consistant over the length of the table because of mechanical wear. If you job entails you using one particular part of the table, then measure the backlash over that part for better accuracy for that job.
You should find now that you can move your table up, down, in and out to a good accuracy - and it is worth taking the time to do this, becasue you can sit down and machine something with a bit of confidence.
This has nearly taken longer to write than to do it - except maybe for the gibs. My table is accurate to a thou (and you want to see my backlash) and I regularly machine steel to within a thousanth of size - i.e. fitting a wheel on an axle requires a two thou gap for the glue.
The other "trick" when machining parts to a fine accuracy is to, as we used to do manually before a computer did it for us, always do the cutting move in the same direction - i,e, run back a little over, then come in again from the same direction every time, not just run back to the spot and start machining. This makes sure that any innacuracies still in the system always happen (or dont happen) on every cut. It dosen't matter which, as long as you are consistant.
Somebody also put in another post that mills are not as accurate as lathes - and I do not know the Taig machine so you may have to be especially considerate to good machining practise to get the accuracy you seek, but good luck.If you already know most of this I apologise - once I get typing I don't know when to stop.
If the household authorities were more forthcoming then I would love to have a go at a little 7 H.P. turbo prop engine - coupled to a 175 amp 24 volt generator - which would make a fine basis for a gas turbine locomotive - and be totally authentic. I have the generator - now..............................
Guest
