First, I apologize for hijacking this thread... but it is so close to what I'm going through, it didn't seem worth starting a fresh thread.

Also a Chinese machine converted from factory pendant/controller to Mach3, but using PMDX boards... Also at about 90% (spindle speed is not linear).

Trying to cut two sided 3D item out of wood blocks, using code generated by MeshCam Art V5. Code proofs fine in CutViewer.

Problem has been Y-axis losing position, so when other side is machined, it is out of register. Now the Z-axis is losing position, and has both gone above and below reference zero.

Have systematically eliminated heay load electric equipment in the shop. upgraded line filtering power strip for the computer. cleaned and lubed slideways.

Stepper motor tuning steps calculated using a 1" dial indicator. Velocity originally set to manufacture's listed "working speed", but slowed down to 90% as part of troubleshooting. Acceleration... adjusted for 45 degree ramps (I read somewhere years ago). Z-axis motor specs are way different than XY.

Anyway, have been making a lot of chips without a useable sample so far. Any nudge toward identifying the source would be greatly appreciated.

http://www.alibaba.com/product-gs/628157563/cnc_tombstone_engraving_machine.html

Hi Paul - I know you apologized for hijacking the topic you posted this on but that doesn't make it ok to do. These issues are hard enough to keep track of without trying to do it for two people at once. I've split this into a thread of your very own. Thanks for your cooperation.

Anyway - onto your issues. What motors/drivers/power supply are you using? as much info as you can get please - links to data sheets would be ideal.

Ian

PaulWC,
Is the position loss repeatable or randon ie; same line or section of gcode?

Could be due to a number of things, but, when doing 3D the axis must be able to provide required acceleration.
There are a lot of small movements.
Had similar problem that drove me bonkers and only fix was to change out the axis motor since modified feedrates or
motor tuning didn't solve the problem.

RICH

I mistakenly re-posted after getting up... old-timer's desease... sorry.

Attached are pics of the items. The machine is a 20 mile trip away from here. The only data I have access to is the vendor site posted in first message. The three ring binder with the machine only covers the pendant controller... since replaced with the PMDX boards.

Rich,

The behavior is completely random... that's why I'm here. Trying to troubleshoot with no ability to produce errors "on demand" has made this a very frustrating venture.

The machine owner is non-technical, and was convinced by the Chinese vendor that all he had to do was plug in a thumb drive with the stl file, and press the "go" button... presto, carved part... using the added 4th axis no less! Only had a 5mm ER11 collet with a 1/8" reducer insert, and two wing nut stamped toe clamps. Convincing him to invest in additional tooling, measuring, and workholding (the basics) has been a major battle. Replacing the motors on a "let's see if this works" basis is an argument I really don't want to add to my stress level. 

I need to make an effort at resolving this with what is on the floor. Was called out of retirement to help out, but payment is based on how the owner gets paid for his work (sculpture)... on delivery of product. 

Thanks, Paul

I has a similar  problem on another small CNC.  It Turned out that the power cord to my spindle was not shielded.  It was causing noise on my steppers.  I installed a well shielded power cable to the spindle and that cured the problem. 
rog

Well... I checked the pics I have of the machine, and it appears the spindle motor cable is  shielded.. kind of. 

I'll contact the owner and have him get his electronics guru on it!

And here's an updated copy of the xml...

Just got a call from the electronics guru, at the machine site...

The shielding in the pic is for the spindle motor, and it is severed at the exposed location. It is not grounded at either the motor or controller end. He is going to try a splice/jumper  repair of the braid, and chassis ground the shielding.

I'm waiting for a call-back to see if the patient survives surgery! 

Paul

Ran first side of the two-sided 3D project with no sign of axis shift after repairing and grounding spindle motor cable shielding... fingers crossed for the flip side later this evening.

I wrote a simple looping code to exercise all three axis, going up and down around an imaginary six inch square, and stopping at reference zero. Thirty iterations, and no shift.

Paul

All seems to be working after a day's running...

In addition to repairing the spindle motor cable shielding, the owner purchased a high end interference filtering power strip for the Mach3 computer, to replace the economy department store variety he had... as per one of my earlier recommendations.

So, which remedy cured the glitch is a local debate. Owner is happy, and going to bless me with a check Monday. 

Everybody is happy here!

Thanks to all,
Paul