Problems with axis accuracy - MachTurn

[JohnHaine]

I'm slowly getting to grips with my Myford Super 7 CNC conversion.  Initial results were promising but I have got caught up in a "problem loop" with the accuracy of the X-axis (i.e. cross slide) feed.

I have fitted a new leadscrew which is 20 tpi though I actually work in metric.  For complicated reasons this is driven through a timing belt with ratio 4/7 from a standard stepper with 8 times microstepping on the driver.  "Steps per mm" then works out to 2204.724.  The problem I have found is inconsistent diameter turning using the turn wizard.  The diameter ends up up to 0.01 mm or so different from what it should, sometimes larger and sometimes smaller than nominal.

I have set about calibrating the leadscrew pitch using a 123 block measured by micrometer to ~25.41 mm (actually measured to a micron) and using the procedure recommended in the setup instructions.  I am setting the slide to a zero with a DTI against the block, the latter mounted against an angle plate on the cross-slide, then removing the block and moving the slide under manual control to zero again and looking at the movement on the DRO.  The DTI is a very nice surplus indicator that has a travel of +/- 25 microns and indicates to 0.5 micron.

Apart from a small indicated error of about 0.08 mm over the travel, I get relatively inconsistent repeatability if I move the slide back 25 mm and then bring it back to zero, of up to about 0.01 mm.  I can reduce this by firmly pushing the slide back against the screw thread before reversing the feed to come back to zero - then I can get to a few microns.  So I was wondering if anyone could help with the following questions please?

Are my expectations of accuracy too ambitious?  If turning down to a specified diameter using the wizard, what is a reasonable accuracy to expect?

Moving the slide repeatedly backwards and forwards to the same nominal position, what repeatability can be expected?

The leadscrew is lubricated with a molybdenum grease, but it seems to me that this may be forming quite a viscous film between the screw threads and the nut, the thickness of which will depend on the force on the screw - would I be better using a light oil?

I don't think I am losing steps, but I was wondering about just how repeatable microstepping is - has anyone any experience of problems from this source?

Any answers to these or other suggestions would be most welcome!  Thanks in advance,

John.

[Hood]

I think you have answered the question yourself when you say

I can reduce this by firmly pushing the slide back against the screw thread before reversing the feed to come back to zero - then I can get to a few microns.

I think you are seeing a fairly good repeatability from the setup you have, microstepping will have some effect, cant give you specific numbers but I would say it safer to think of a uS could be anywhere between one or the next. Your gearing may also introduce inacuracies. Light machine with flexing, stiction etc will introduce innacuracies but the biggest issue is likely to be the screw.
Having said that have a look at the repeatability of some   industrial machines and you may be surprised, their positioning will likely be something like +- 0.005mm, repeatability slightly better, probably 0.005mm.

Hood

[Hood]

Just had a look at HASS, theire positioning and repeatabilty seem to be standard accross their range, it is positioning of +- 0.005mm and repeatability of 0.003mm.
Their machines are by no means the best, in fact they could probably be classed as fairly light in indutrial terms but compared to your myford they are huge rigid mosters, so I would say you are doing well
Hood

[stirling]

I don't think I am losing steps, but I was wondering about just how repeatable microstepping is - has anyone any experience of problems from this source?

Hi John. Microstepping should NOT be included in the "theoretical" calculation of accuracy. Purely from the motor accuracy perspective, full steps are usually stated as being within +/-5% accurate. Microsteps are less than that and are unpredictable. (Note the difference between accuaracy and resolution). Whilst full step accuarcy is purely down to motor manufacture, microstep accuracy is down to a combination of the motor/driver/PS/mechanics etc. As just one example, sometimes, a motor will not even move until it's had 3 or 4 microsteps, then it will "jump". It will do this sometimes and not at others - maybe.

Microstepping's purpose is purely to get smooth(ER) motion than full stepping.

Ian

[Overloaded]

This is fron the Gecko site regarding the linearity of microstepping.

"Another factor affecting accuracy is motor linearity. Motor linearity refers to how the motor
behaves between its ordinal step locations. Ideally a 1.8 degree per step motor should move exactly
0.18 degrees for every step pulse sent to a 10 microstep drive. In reality all step motors exhibit some
non-linearity, meaning the microsteps bunch together rather than being spread evenly over the span of
a full step. This has two effects: Statically the motor position is not optimum and dynamically low speed
resonances occur because of the cyclic acceleration where the microsteps are spread apart and
deceleration where they bunch up. Figure 17 shows a motor with excellent linearity and one with
horrible linearity."

[dresda]

How do you measure .5 micron? And you said you are using a leadscrew..
Ray.

[JohnHaine]

All, thanks for the good advice, I have now at least re-calibrated the "steps per" for the X-slide and things are much more understandable now.  I can see that microstepping is not working exactly as one might naively expect and I'm seeing an effective "average" single step size of around a micron which is double what it "ought" to be.  I have also reduced the axis acceleration - I wonder if the slide was slightly over-running when the motor slowed.

Dresda, as I mentioned, the indicator has a full-scale range of +/- 25 micron and resolves to better than 0.5 micron.  And it cost a few pounds from a surplus stall - a good buy (though this is the first time I've had a use for it).

Could I (re)ask a couple more questions please?

Is my surmise about leadscrew lubrication correct - is it better to use a light oil rather than thick grease to ensure that one doesn't get a variable thickness layer of grease which could affect the position?  And what is people's experience of the accuracy achievable with Mach?

Thanks again,

John.

[Hood]

I would say a light oil is probably better.
Regards accuracy of Mach well that is really a non issue I would say, I very much doubt anyone has hardware capable of the accuracy that Mach is capable of

Hood

[stirling]

I can see that microstepping is not working exactly as one might naively expect and I'm seeing an effective "average" single step size of around a micron which is double what it "ought" to be.

Lost me there John. Microstep errors are non-accumulative so in your case with a microstepping factor of 8, the average should be an 1/8th of a full step.

Ian

[JohnHaine]

Hood - thanks, will go with the oil option.  I should have chosen my words more carefully - Mach is capable of very high accuracy, I was really asking what people had achieved in practice with typical lathe conversions.

Stirling, yes, it should be but doesn't seem to be!