Machsupport Forum
Mach Discussion => General Mach Discussion => Topic started by: JohnHaine on March 01, 2012, 10:14:36 AM
-
Hi all,
After adventures getting my Denford mill up and running I went back to try and see why I just couldn't get my Super 7 conversion working to the accuracy I would like, particularly on the cross slide. I was getting inconsistent results on final diameter, with errors getting close to 0.02 mm which didn't seem right. Eventually I decided that the problem was due to slight runout of the timing pulleys through which the feedcrew was driven - since both of these had to be bored out to fit the shafts of the stepper and the screw there was scope for error. And doing careful measurement of feed with a digital micrometer (Mitutoyo) that can resolve a micron, and separately measuring runout with a DTI, I finally decided that both measurements were consistent.
I'm now resigned to going to direct drive from the motor to the feedscrew, since I think there is plenty of torque available, but has anyone else had any experience of this problem; and any recommendations on how to accurately bore timing pulleys for future reference please?
John.
-
You seem to be concerned about an error of well under 0.001", which is certainly well BELOW the accuracy of your leadscrew. So, how have you eliminated leadscrew pitch error as the cause of your problem? That would seem to me a FAR more likely source of error.
Regards,
Ray L.
-
Maybe I didn't express myself clearly. Actually the error I measured was 18 microns peak on feed, so that would be .036 mm on diameter or 1.4 thou of an inch. Manually turning I would expect to be able to significantly better that by careful finishing. With CNC you don't get a chance. I decided that it was runout on the basis that the error was periodic and a bit of spreadsheetery showed that it had components at the rotational speeds of the screw and motor pulleys, and the amount estimated correlated with a direct DTI measurement.
-
Does the lathe use steppers? Do you have gearing? Qhat is the microstepping? Do you have ballscrews or is it just leadscrew?
Hood
-
Hi Hood - yes, it uses steppers. Gearing was just via the timing pulleys. 8x microstep. Leadscrew, not ballscrew, 0.05 inch pitch.
-
Does it repeat at a commanded position? ie if you command a dia then move away then again command that dia does it go there?
Hood
-
I have redrilled and reamed a lot of timing pulleys and never had the problem you are describing. The cast ones from Browning are quite good, by that i mean, if you have an accurate three jaw chuck it will center them accurately. If the casting is bad one can always use a mandrel to true up the hub of them before placement in to the chuck. As you said repeatable inaccuracy per revolution can appear. Timing belts to my surprise are really quite accurate but you can loose a few steps per rev depending on the tension used but that is repeatable.
I would think that it's more a problem of the lead screw and backlash / non use of a good ball screw . A small backlash can come from the thrust bearings if not adjusted just so and can
make a big difference. Another item to look at is the coupling ( if say a bellows type ) since they can give some and cause inaccuracy and you may not notice it.
The smaller the error you try to get to the more anal one must get and comes at a price also. Just a fact of CNC.
BTW, 0.02 mm =.0008" on a diameter.... you must have some backlash ( zero is imposssible else the axis wouldn't move). Only you can define how much time, effort and money
you want to put out to get 1/2 the current value?
RICH
-
Hi Hood: yes, quite repeatable in the manner you describe. Always moving the slide the same way to the measurement point to eliminate backlash.
Rich, I'm not sure who the original pulley supplier was, they were ordered through RS Components. I mounted then in a 4-jaw by the hub, centred on the existing small centre hole (DTI), and drilled and reamed out to 0.25 inch for the motor or 8mm for the L/S shaft. However the holes were too small to easily bore so some inaccuracy may have crept in. Surely backlash shouldn't affect CNC turning to a diameter any more than it does manual as the feed direction is always one way? And my measurement was ~.02 on a radius which is .04mm on a diameter = .0016".
OK, I realise I may be trying too hard here, but if for example I want to turn down a shaft to be a running fit in a bored hole of a given diameter using CNC, should I expect to be able to do this in one program or to stop a few thou short and use semi-manual techniques for the final fit?
Thanks, John.
-
With your setup I would say you can not be as accurate as you want, things that are working against you light machine, possible flex, lead screw rather than accurate ballscrew etc, even the steppers may be an issue as microsteps will never be exactly 1/8th of a motor step.
Even tool being not on centre an have an effect, this will show up as an error on smaller dias more than bigger ones. For example if your tool was 0.2mm low and you set you cut a dia and measure it and its 20mm so you set the DRO to 20mm dia, you then move to cut 10mm Dia it will be 10mm to 2 decimals so thats ok but now you cut at 5mm you have now cut at 5.01, cut 4 its 4.02 etc etc, see pic. That is assuming you have exact movement and no flex backlash etc so it is just an example as how every part of your setup needs to be taken into account.
As for the pulley being out I cant really see how it would make that much difference, not thinking too deeply about it but if you cut at a dia then move the pulley exactly one rev then it should be exactly the distance of 1 rev of screw on axis which is 1.27mm in your case I think, so if you then cut you should be exactly 2.54mm less on dia. Thats if I am thinking right ::)
Hood
-
John,
Surely backlash shouldn't affect CNC turning to a diameter any more than it does manual as the feed direction is always one way?
In general would agree with that statement. Backlash may not behave "exactly" every time.
All the the items I mentioned affect accuracy to some degree. So each part of the "system" contributes to the inaccuracy and they add up and in the end you have
a total error. So when you get down to say a small error you need to find what can or can't be adjusted to help reduce the error. You need to consider how the actual machining
is done. Maybe a different way of saying all this is that the best one could expect is equal to the axis resolution but in reality you will find it is something less.
In the Threading on the Lathe write up, which can be found in Members Docs, see page 55 for JIS standard on a lathe accuracy if you want to machine compare.
RICH
-
Just an update here. I have now rebuilt the cross slide drive with direct coupling of the motor to the leadscrew and things look much better, so I hope I'm back in business. Thanks for all the input.
John.
-
how good is the Myford bed .Is it worn .Myfords have a habit of shying away from work with there flat bed design . I used to have one and found that a couple of zero cuts helps . I now run Hercus/Southbend . And find heaps better . give it a go
-
I got the lathe new and it hasn't had much wear. It was cutting undersize, not oversize, so zero cuts wouldn't help? But thanks for the suggestion, another thing to look at if still dodgy.