[cncvirgin]

hello,

it's been several weeks now that i'm trying to solve this problem:

When i'm doing 3d finishing passes (Parallel Finishing from Rhinocam) at the moment when the Z axis retracts after finishing one part and before transfering to the next. The z-axis 'grinds' and i'm losing steps so the Z0 is not at the same height anymore and therefter the tool digs in deeper than it should making the parts useless.

Here's what i tried until now:

• bought a power conditioner
• reinstalled Windows from scratch on an SSD and optimised it
• upgraded Mach3 to the latest version
• cleaned and greased the z-axis parts
• tripple-checked and tried out all possible and imaginable settings/combinations in Mach3 and Rhinocam
• changed the Step and Dir lengths
• inverted the Step and Dir low/high active
• read all related threads in various forums and tried out any remotely related suggestions
• edited the Gcodes to insert a G4 Dwell before the problematic retract movement
• tried with or without Rapid movement
• tried adding a radial exit of the toolpath before retract
• tried all possible and imaginable combinations of speed/velocity

I think the PC is not at fault, Mach3 driver test shows 'System Excellent' for all kernel speeds including 100kHz

I also think it's not a speed/acceleration problem since i tried out everything from speed: 100mm/min to 3500mm/min with Acceleration of 10-2500mm/sec2

the problem persits, here is a video of it (with a 10sec G04 just before the problematic retract): Untitled Document

in this video you see the Z go up at 200mm/min .. when i test in motor tuning or jog i can get up to 2000mm/min without any problems and really smooth and speedy motion, even when i'm operation all 3 axes at the same time

most related forum threads i found pointed towards a speed/acceleration problem that's why i spent several days testing out all possible combinations. the problem is always the same, always only on the Z and always when retracting after a parallel finishing.

what's really strange is that the finishing pass itself runs without any problems, even at 2500mm/min, it's only the retract that blocks ... in the specs of the machine the max speed is quoted as 4500mm/min .. i tried everything .. slow speed, slow acceleration, slow speed, fast acceleration, fast speed slow acceleration, fast with fast and at least a hundred setings combinations in between

what's even stranger is that a horizontal roughing pass where there is much more radical movement and alsop a lot of Z retracts, it runs without any problems .. even at 2000mm/min


from what a friend who is an expert in mechatronics the machine itself is in good mechanical condition .. and he also can't figure out whats wrong

the machine is a Gravograph VX2 connected via an octocupler to the parallel port of a pretty high-end (for the time) IBM Thinkcentre. I'm using the original motors and onboard drivers and the machine has done me good service for almost 2 years now. I've tried opening up old GCodes which i know did not exhibit this problem (since i managed to finish the products without problems) and the problem is also present in those files which used to run fine

i'm really completely lost .. possibly more than the steps on my Z

[SWATH]

I had the same problem with my mill. It drove me nuts. I swapped the axis drive with a different axis and the problem followed the drive to the new axis. I determined it was something wrong with the driver and replaced it. Haven't had the problem again.

[ger21]

I think the PC is not at fault, Mach3 driver test shows 'System Excellent' for all kernel speeds including 100kHz

The System Excellent message in the driver test is meaningless.
What kernel speed are you running at? Is the line in driver test perfectly flat, or are there large steps?

What kind of drives are you using?

[cncvirgin]

yes the line in the driver test is completely flat, one or two tiny spikes when it takes over but then totally flat ... i have been running for years at 25k kernel speed .. i just tested the other kernel speeds to see if it changes anything but it doesn't .. the line is also completely flat for all kernel speeds

the drivers are the original Gravograph drivers built into the machine, i don't have more info than this: http://www.e-engraving.com/machines/...h_VX2_Spec.jpg

This is the driver board, but it's much cleaner and dustfree now it's a photo from when i first got it:

Attachment 245302

i also forgot to mention in my initial post that when the Z grinds and loses steps and i stop the machine, re-zero the axis and then scroll back a couple of hundred lines before the critical retract point and do a 'Run from here' the problem doesn't happen .. the Z axis retracts without grinding or loosing steps ... even at speeds of 2300mm/min .. if i then do a rewind and run the code from start the problem is back .. totally weird

i'm also starting to think that the driver is at fault somehow ... it's just weird that it only happens under these precise conditions (retract before transfer on a parallel finishing), in motor tuning or jog everything is hunky dory

[awerby]

You might try doing what Swath did - turn everything off and wait a while for the capacitors to discharge, then swap the Z axis to another driver and see if the problem goes away, or transfers to the other axis. I doubt you can still get replacement drivers for that 1986-vintage machine, but you can always contact Gravograph Engraving material and supplies - Gravograph - Laser Engraving Systems, engraving software - New Hermes engraving machine and see what they recommend. If worst comes to worst, you can probably just replace the whole thing with new drivers and breakout board, maybe a G-540, which is pretty popular here.

[cncvirgin]

so .. the plot thickens .. once again

i did as you suggested and swapped the cables for Z and Y ... unfortunately the problem persists

as for Gravograph, i contacted the european headquarters when i first got the machine and they weren't very helpfull, they just said it wasn't a supported machine anymore and that it was impossible to make this machine work easily with a modern computer (there was an old DOS PC which controlled it originally). They said that they still had a control unit for sale which 'might' work .. the price was 13k - and all it really needed was a cable and it could be directly connected to a parallel port PC with Mach3 .. there was another guy in the UK who had such a box for around 1.5k but at least he was so honest to tell me that all it needed was a custom cable and even gave me advice and some of the pinouts to do it

[awerby]

So the problem remained with the Z axis? That's actually good news, since it means that the problem is probably a mechanical one, wiring, noise, or a settings issue and not the fault of your controller or driver. If it's just faulting when you do a rapid upward move and you're sure your velocity and acceleration settings are correct, that points to stiction of some kind - check again to make sure there's no binding at the top of the stroke. Check to see if any wires are stressed in these upward moves, or if they could be losing their ground connections. There could also be some kind of noise coming from your spindle. To check that, try running the problematic part of your program with the spindle turned off.

[cncvirgin]

check again to make sure there's no binding at the top of the stroke. Check to see if any wires are stressed in these upward moves, or if they could be losing their ground connections. There could also be some kind of noise coming from your spindle. To check that, try running the problematic part of your program with the spindle turned off.

sorry english is not my native language .. what do you mean by 'binding' ?

i'm already doing all the tests with the spindle off

[awerby]

Binding is when the axis gets stuck for mechanical reasons. It could be misalignment in the rails, or between the rails and the screws, or failure of lubrication, etc.

[probinson]

The first thing I would do is try to narrow down problem. I would start by making sure it has nothing to do with the over heating.

Replace the dwell you inserted with a stop. Run the program and when it stops, give the controller a good 15 minutes to cool down. Then check that everything did cool down and resume.

If the problem goes away you still need to run it again, this time restarting it immediately to see if it was the cool-down or if adding the stop fixed it. If adding the stop fixes it then it's probably a very sneaky software bug.

If the problem did not go away then you need see if it is a mechanical problem. For this you need to monitor the drive signals being applied to the stepper motor. This requires some test equipment to monitor the drive signals going to the motor. An oscilloscope with isolated differential probes is ideal if you have one but a set of headphones and a resistor will work just fine.

All you need to find out is whether or not the motor is stopping because some step signals were missing. If you have the oscilloscope you just connect to the motor drive leads and see if the pulses become erratic causing the motor to skips steps. With the headphones you just listen to the pulses to accomplish the same thing.

To use headphone to monitor the signals you need to connect a resistor to cut down the signal level. The value of the resistor isn't critical. We just would rather not blow out either the headset or our eardrums. Some ballpark math to come up with a reasonable resistor value:

A 1 volt signal is pretty loud in a headset and the stepper drive is probably around 50 volts so we want to cut the volume about 50 times.

A typical headphone element is 40 ohms. 40 x 50 = 2000 so a 2K Ohm resistor in series is about right.

The headphone plug has 3 contacts:
Tip - Left channel
Ring - Right channel
Sleeve - Common ground for both channels

I find it easiest to wrap one lead of the resistor around the sleeve ground and poke the other lead into the housing, touching one of the pins. Then use the tip like a probe to connect to another pin. With this you will be able to hear if missing step signals are causing the problem.

After that you will at least have some idea where to look.

Pete

[jeffserv]

As before was mentioned... the problem did not move, with the switching the drive... thus the problem is either in the cabling... when it is in a certain position. Or as seems more likely, something in the bearings, and sliding up and down.

[graffix11]

This may help, Where did I put those steps - YouTube

[jeffserv]

This may help

Great video, this should help. Also don't change 3 things at once. Only do one change at a time. Then test. If that don't work. Reset to original, and try the next change. This way you will discover exactly what the problem is.

Kind of like fixing a car... "I replaced the battery, the alternater, and the cables... and its fixed!" ? = What fixed it, the battery, the alternater, or the cables?? Answer: "I don't know...". If you replace one thing, and it don't fix it, then put it back, and try something else... till you discover exactly what it is.

Hope you find it...

[cncvirgin]

great video indeed !!!

so i did what he suggests, first disabled XY axis - still losing steps
increased the pulse width to 15 for both step/dir - still loosing steps
tried sherline mode - still loosing steps
Tried very slow speed an acceleration - still loosing steps
tried various speed/accel settings - still loosing steps
pulled the Z motor, marked with a sharpie - still loosing steps

here it is:

Attachment 245594

strangely enough it say SERVO on there ... how can i tell if this is a servo or a steper ? There is no info on the web on this motor and the manufacturer does not have any info on this product anymore since it was a long time ago and it was from a different manufacturer just rebranded.

so at least i know now that it's not a mechanical/binding problem on the Z axis
so it means the problem is either with the motor itself or it's in the path from stepper to computer which could mean either:

drivers
power supply
optocoupler
parallel port
software (mach3)
Gcode

i guess the last two are probably not the problem, there doesn't seem to be anything weird in the Gcode and i went over all possible settings in Mach3 several times and wouldn't know what else to try in there

the power supply and the drivers are on this old Gravograph board and since i'm not so good with electronics i wouldn't want to attempt replacing just one of the components so i'm thinking about swapping out the whole thing: new powersupply, new drivers ... and while i should probably change the stepper as well since it's 25 years old. this would also give me a feature upgrade with microstepping but it will cost 3-500€, still hoping to find the problem without needing to spend so much

but i still can attempt what Probinson suggested, listening to the signal in audio seems like a good and fun idea

not quite sure yet where else to go from here ...

BTW .. thanks so much everyone for helping me with this !!!!!

[probinson]

They called it a servo but it is definitely a stepper motor. The sound of a stepper motor if quite distinctive and they labeled it as having 1.8 degree steps.

There are several negatives using steppers v.s.servos, performance wise, but the fact that they don't require feedback to operate is a big advantage when troubleshooting. You can run your program with the belts removed and see if the stepper still looses steps. That way you completely remove everything mechanical after the stepper. Because the motors aren't driving anything, you can also connect any motor to any axis to ensure the problem is not with the motor itself.

Pete

P.S. For that matter, the motor doesn't even have to be one of the motors on the machine. You just plug any similar stepper motor in and see if it tracks properly.

[cncvirgin]

You can run your program with the belts removed and see if the stepper still looses steps.

yes that's what i did and it's still losing steps

BTW here are some detailed pictures of the machine: https://picasaweb.google.com/1142453...eat=directlink

here are the cables that come out of the stepper (or servo) enclosure:

[jeffserv]

As to the test you previously did, you switch cables to different drives. So the problem still stayed with the Z Motor, or Cable.

The Z cable does a lot of back and forth, and larger companies replace them routinely, to make sure they never have this problem, some companies replace them yearly.

I have not on my machines...

Now, if you put a different stepper on the end of the cable, and it did it also. Then the stepper is good.
Now you need to bypass the cable, and with the original motor, wire it in on the opposite side of the cable... and see if it looses steps.

It is common for that cable to go out.

Sincerely,

[cncvirgin]

still not progressed much ...

good news is i bought a new stepper motor and managed to connect it so it works (wasn't easy since i didn't have the pinout of the old motor so i had to determine which lead is which first)
bad news is: it wasn't the stepper, still losing steps with the new one

so it's something downstream from the stepper. I doubt it's the cable since it's really well attached and there is no movement of the cable at all when the machine is running since the stepper transmits through a belt. I will try to double check it but it'll be hard since it's channeled through the underside of the machine.

being a musician i'd quite like to try out listening to the signal in audio
i'm not sure i completely understand the instructions.

To use headphone to monitor the signals you need to connect a resistor to cut down the signal level. The value of the resistor isn't critical. We just would rather not blow out either the headset or our eardrums.

The headphone plug has 3 contacts:
Tip - Left channel
Ring - Right channel
Sleeve - Common ground for both channels

I find it easiest to wrap one lead of the resistor around the sleeve ground and poke the other lead into the housing, touching one of the pins. Then use the tip like a probe to connect to another pin. With this you will be able to hear if missing step signals are causing the problem.
After that you will at least have some idea where to look.
Pete

poke into which housing ? i have all the leads coming from the driver (4 leads) going into clip connectors so it's easy to connect .. but not completely sure what to connect to what ...

in any case ... since the problem seems to be in the old electronics of the machine i think i will stop troubleshooting this at this point and completely eliminate the old electronics and power stage. i'll probably buy 3 new power stages from Leadshine, a good power supply and if i have to an interfaceboard but i have this super cheap chinese 6-axis USB board lying around which i'll try first.

thx everyone for your help with this .. i'll keep you posted when the new electronics are up

[probinson]

Hey cncvirgin,

I was wondering how the troubleshooting was progressing.

My belief is that there nothing wrong with the wiring or anything mechanical as the skipping only occurs after a particular sequence is executed and the problem repeats exactly the same each time. You just don't get that kind of selective and precision repeatability from faulty hardware. Exact repeatability requiring a specific sequence to activate is a symptom of some sort of software glitch rather than a hardware issue. But, I'm not the one troubleshooting the machine so my understanding of the symptoms is rather limited and could be completely wrong.

Anyway, back to the task of proving to yourself whether or not the motor is skipping steps because it was told to. For that, you can use headphones and simply listen to see if step pulses become absent or chaotic when the motor skips.

The axis is supposed to be retracting at a constant rate and therefore the controller should be sending a steady and continuous stream of step signals to the motor. If you hear steady tones on all 4 driver lines when the motor skips then the problem is on the machine side. You will, likewise, be able to hear if the drive signals are missing or corrupt and know that the problem is somewhere from the controller back.

(Note that the motor will be both loading the lines and generating signals so some experience interpreting the sound is required if the tone changes but doesn't completely stop during the stepper skips. I can provide instruction on that if the need arises.)

You posted a picture of the driver and across the front are the stepper motor connectors. That's a good place to access the motor signals because the contacts are easily identified and the connectors are large enough to insert some sort of home-made probe along side the wires to connect your test equipment.

The headphone has 2 independent elements but they are connected with a 3 wire cable so there are 2 options in preparing you headphones:

1) If you don't want to mess up the headphones then you can create a 2 wire probe but you will have to listen to the motor phases in separate passes.

2) If don't mind possibly sacrificing them for the cause then you can create a 4 lead probe and listen to both stepper drive phases at once (all 4 lines). This is the best way but you have to remove the original wires and connect your own directly to the left and right elements. If the elements are not easily accessible, you may not be able to restore the headphones to original condition.

The construction is straight forward. One probe that you can put into the motor drive connection is wired directly to the headphones. A second probe is wired though a resistor to reduce the signal to the other headphone contact. If you are making the 2 probe setup then the headphone connections would be the sleeve and tip (or ring and tip to have sound in both ears) on the headphone plug. If you are making the 4 probe setup then you connect make your connections directly to one earpiece element and then make another probe, probe/resistor set for the other element.

This is a bit of a digression but you'd be surprised how useful this simple tool can be.

Say you want to make your rapid moves as fast as the machine can go. What do you do? You set up a program that moves the axis back and forth and ramps up the speed. You put a little extra load on the motor for good measure, run it until it starts skipping steps, then back off a little from the last speed that worked.

Pretty straight forward but how do you know that real usage doesn't exceed the requirements of that little test program? Well, if you were listening to the motor drive when you performed your axis speed test, then you know what it sounds like when you get close to or exceed the limits. You simply listen the drive signals when the machine is doing actual work and you can hear when the motor is getting too close to the limit. You become that guy in the machine shop that can put a stethoscope on your engine and tell you what's wrong.

Your brain is an amazing pattern recognition machine. Your eyes can decipher lots of information at once but your ears can decipher information that happens fast. A stethoscope is every bit as useful as a magnifying glass.

Pete

[probinson]

poke into which housing ? i have all the leads coming from the driver (4 leads) going into clip connectors so it's easy to connect .. but not completely sure what to connect to what ...

Guess I got sidetracked and didn't actually answer the question. Sorry.

Your phases don't seem to be labeled on the PCB so you need to pull the plug and use an ohm meter to find out which wires connect to the motor coils. You don't need to know polarity, just which pair of wires are used for each motor coil.

The motor plugs are all wired the same so just choose the easiest one to work with. Measure between the plug contacts (the cable end that goes to the motor not the PCB pins) to find out which pair of wires are connected to each coil. The wires that read a low resistance to each other are on the same motor coil so that is where you will connect your probes. You want the speaker element in parallel with the motor coil (with a resistor in series with the element to limit the power, of course).

Pete

Oh, the 2K resistor value limits the power to the headphone element to around 25mW. The power dissipated by the resistor, however, could be on the order of 1.5W if it is connected to a drive with no motor loading the output. A 1/4 Watt resistor should be fine with a motor connected but you need higher Wattage and/or resistance it you want to listen to the unloaded driver output.