Hello, new here, first here is what about 2-3 days of machining looks like for botched work:

I've been running this machine for several months without issue, far as I know no missing steps, but today it seemed like the X was suddenly off by .1 inch twice during 3 hour cuts.


I have rotary encoders from Digikey and 4 axis mill (2 slave drives for the Y) and was thinking of getting the Encoder Interface Board

http://romaxxcncrouters.com/store#!/Sound-Logic-Encoder-Interface/p/10142577/category=2363055

Some of these jobs are expensive taking a few days and expensive wood. I am just hoping to take out 'insurance' on the part by specifying an allowable deviation, say of .005", on any given axis before Mach 3 or 4 shuts down the machine waiting for me. Is anything like this possible?

Since I just bought the G540 about 4 months ago, with the great rep it has would like to avoid servos!! Thanks

Shawn

Hi,
given that you have steppers then encoders are not going to help much. Stepper motor drives
are typically open loop. With a drive of that description and I believe the 540 is an example
then it cannot correct missing steps even if you had an encoder to detect it.

At best with the drive that you have you might be able using an arduino or similar is detect
when missing steps occur and enter some fault state so that you could correct it.

I understand there is a Chinese manufacturer, Longs maybe, that offer a closed loop stepper
drive. While cheaper than a genuine servo still quite a premium over open loop drives like the 540.
Have heard that Tormach have adopted them as standard on their new equipment.

Finding out why your steppers are getting out of whack maybe a better bet. The only few times
I've had issues is when I was pushing too hard, usually overfast rapids with a heavy workpeice
and then they stall and whine, no guessing required as to the cause!

Given your description I would guess that you are suffering from intermittent noise where a noise
impulse surreptitiously adds or subtracts a pulse. It can happen in bursts. If it occurs on the direction
pin during a rapid it usually produces a stall. I would be looking at the step lines. It not unreasonable
to pass the control lines thru a ferrite ring, one or two turns, supresses some RF without affecting
the pulses much.

Craig

Hi,
just had a look at the board in your link. Don't think its anything like smart enough to
'close the loop'.

The Chinese manufacturer I was thinking of was Leadshine. They look to be the real deal.

I use Vexta 5-phase steppers and low backlash planetary gear drives, consequently can't
use Geko stuff. If I were using 2-phase motors I would and from the feedback on the forum
don't think I would use anything else, they are very good.
The Vexta drivers I use are like most Japanese stuff, expensive but superb.

Craig

Thanks for the advice, I will be cleaning up the wiring and using ferrite core rings. I assume the wire gets wound around the ring like an induction coil. Other than that the only 'solution' seems to be to run a very long stepping program to see if motors are losing steps. This gigantic job was the first time I noticed a problem, it missed steps twice. You can see where the entire program shifted and the amount of shift seems to be about .1, the amount of a missing step.

0.1inch is equal to 1 step? Seems quite course but if so then the encoder interface board is definitely not going to help with that size of step.

Hood

Hi,
yes you do use ferrite rings as a coil. The ones I'm thinking of are commonly seen on the input wiring
of switch mode power supplies, computer power supplies and similar. They are primarily for EMI suppression.

I work repairing welders and see lots of them. Wouldn't mind betting you could find a whole bunch surplus
at just about any electronic repair outfit.

Commonly the wires are passed thru the hole a couple of times. The inductance goes up as the square of the no.
of turns so don't get too carried away, 3-4 turns will offer max suppression without distorting the pulses too much.
It is in fact better to have several cores along the length of cable with one or two turns than one core with 6 turns
say. What you're trying to achieve is to 'break' the cable into a number of shorter pieces that each piece becomes
a much less efficient anttena.

After I replied last night to your post I started thinking more about the board in the link you posted. I realise now
that it might work better than I had guessed. I assume it allows an encoder to be interfaced and monitored by Mach.
Mach could then signal an error should steps be lost. There have been a number of topics that discussed whether
Mach could 'close the loop' and auto correct. The majority of experienced users cnocluded no. That does not preclude
Mach from detecting an error or growing error from occurring. I have been a bit hasty in dismissing the board and the
idea on which it is based. My apologies.

Notwithstanding that it still seems better to find the reason for the error rather than just detecting it. From your description
the machine has worked well for some time so whats changed?. It could be as simple as a loose coupling or something.
Does the error occur on one axis only? If it were electrical noise you might expect it to affect more than one axis unless
one axis is more sensitive to noise, maybe proximity to anther cable.

Let us know how you get on.

Craig

Is this the first time that you have experienced missed steps when doing 3D machining?
If more than once can you duplicate the fault?
Are the missed steps random or consistant at some section of the gcode.

You need to find what is causing the problem.

The problem could be due to inadequate motor torque for the commanded move at current
acceleration and max velocity setting. 

 RICH

BTW,

I would never run a code that requires "days" to machine something.
Suggest you have a hard  look at  how you are setting up the project in the software such that rough and  finish cuts are
efficiently done to minimize the time.

RICH

Days is not an accurate description. These tool paths are usually 2+ hours at a time, 4 max. It may take a few days to finish one piece.

Not sure about duplicating the fault, they both occurred towards the end of the last tool path. The only thing I can think of to nail it at this point is to repetitive stepping with the motor against a dial indicator for say 4 hours to the range of the axis.

My motors are bipolar stepper 400 oz in with a 5 start ACME thread 1/2-10", and I run at 100-125 IPM max rapids/feed.

Not sure about duplicating the fault, they both occurred towards the end of the last tool path.

By the last pass I take it that it was the finish pass.
What you can do is insert some code such that there would be a move to a check location
along with a short dwell so you can inspect position to see if skipping occurred.Strategicaly place it
in the program ( can be in multipe places ) to see if yo can repeat the problem. If it's random then it is difficult and may not yield any results. I have a seperate digital read out on my mill so can compare against the DRO.Poor mans readout can be a a simple mark / something on the motor shaft so you can check position at some point in time. Skipping can occur and you won't even hear it and result can be that the axis continues on with a loss of position or the axis just runs away.

What is important is how much torque you have while machining at feedrate. In the machining you have a lot of very small combination moves and thus the accel is almost instant in time. You shold minimize the torque required  to just move the axis and have ample torque available for required acceleration of the code
and headroom for machining of the material.

As replied you need to find the cause of the problem. It could be a number of things or combination of
things. Start with basics, be meticulous in what and how you are testing such that you eliminate
factors.

RICH