mistermikev

55 minutes ago, MiKro said:

Well pulse is most likely to fast now, I would go back to the 2 or 3 setting on that. Set the accel rate to 10% to 15 % max for grins to see what happens on all axis. Bump your microsteps up by the next value for tests at this point. try say 3000 if that is a choice. you may try the other direction as well.

I added a file for you to gen a toolpath with. rename by removing the .zip

MK

test.crv3d.zip 19 kB · 0 downloads

roger that.  thanks for the help.  will try asap.

47 minutes ago, MiKro said:

Mike in this picture I can barely see the dip switches. It appears that on two of them DIP 5 is on and the other it is not? Also what micro steps are you running and what is your accleration? Whan I have had an issue with the type of movement you describe it was usually the acceleration that was a problem. I am still concerned about the dips though? On most drivers DIP 4 or maybe 5  (???  been awhile) is normally used to to half the power at idle.

also would like to see all of the wiring to the drivers? Are you using wiring to the enable connectors on the drivers as well? Most do not use this as the 5VDC for pulse and direction is all that is needed.. Or is the small connector that has the Black, Red, Yellow to the Pulse and direction?

Also what drivers are they specifically? If they are the TB6600 then many have had issues with them.

 

These are the drivers I use.

https://www.automationtechnologiesinc.com/products-page/kl-stepper-drivers/kl5056

 

thanks for the chime in @mikro. 

I believe these are the "crappy little brother" to the 6600 ie the 6560.

the dips: this machine does have a slightly dif (smaller) motor for the z axis.  I think I read that the dip switches are set dif on this drive and possibly the A drive too? 

Can't find these exact drives anywhere altho there are similar 6560s on evilbay... but they don't have the white connector(black/red/yel you mentioned) - no idea what wires are what.

accel - I believe it is 1500, microsteps I believe are 320 on the xya and 400 on z if I'm not mistaken.  pulse is set to 0/o on all??  I have played with all of these settings with zero success.

I took it all apart this morning and was looking at it just b4 work and was planning to label the drives and take a crap ton of pics for future reference -when I get home tonight. 

1 minute ago, curtisa said:

I had forgotten you'd stumped up for the extra axis on your machine. Certainly worth a shot if it's easy to reconfigure A axis driver as X.

I mean I would physically swap them as opposed to reconfigure in mach3 .  

1 minute ago, curtisa said:

What about if write up a quick small program that just moves the X axis in small increments, with a pause between each move?

Psuedo code would read something like:

When the program ends measure the distance travelled and see if the total amount of movement adds up to 1 inch.

I would certainly be willing to try that - you've done so much so far - don't know how I'll ever repay.  I know I know... "someday I might ask you for a little favor..." hehe.  that said, I don't think x movement by itself is the issue.  it seems to be x only in combo with y... but I could be wrong, would love to know.

was thinking... I have a 4th axis... in theory, if I get the gumption... I could swap the drive board between x and a.  doubt I'll ever use the a axis anyway... but I'd be taking a chance at bringing the whole thing down by jacking around with it.  have to think on it some more.  the prospect of being up and running is very tempting.

33 minutes ago, curtisa said:

So it's slow rates of stepping on X axis while Y axis is simultaneously doing fast rates of stepping, but only when X is traveling in the negative direction west beyond X=0? That makes absolutely no sense whatsoever. 

Based on your observations and my interpretation, when you run the code you've just attached you're seeing:

1. the cutter zip west in a straight line
2. gradually head north at a slight easterly angle
3. turn around and head due south
4. turn around and head due north <- incorrect motion at this point?
5. cutter zips back across to the origin and now X is too far east?

Step 4 of the above should be heading north at a slight westerly angle, but instead it only goes due north? And during step 4 are you seeing Mach3 update the X axis readout while it's making the move?

 

The slower the movement of the axis the fewer pulses per second are being sent to it. The voltage/current waveforms being applied to the stepper motors should the same magnitude, but the rate at which they change will be different. They shouldn't be any more or less susceptible to interference whether you're stepping at 1IPM or 100IPM. Noise *could* be an issue perhaps, but it's got a funny way of influencing how the machine is being affected.

I'm more concerned that perhaps you've got a buggy copy of Mach3.

between steps 3 and 4 there is a very slight angle... and I think it is here that is causing the issue... but admittedly I hadn't watched the coordinates while it was running.  Might have to try that.  that said... I ran this 3 or 4 times in a row, then did goToZero and we were off by about 1/4".  It does seem to move at an angle at step4.  It is hard to tell if it's moving at an angle for step 3 because it's so slight.

on my old machine I was using the copy of mach3 I got with the machine, but had licensed it legit thru mach3.  On the new machine (the one I tested this on) I had downloaded the latest version of mach3 and then licensed it.  So I doubt it's an issue with the version... and I suspect many people have been able to cut neck pockets using mach3.  I would think I would have heard others with this issue.

Maybe these stepper drivers have issues with very small movements?  Then it only seems like it's the x.  could be that that one driver has a bad chip and is missing steps.  I have heard of this issue for others.  I really need to pull the thing apart and figure out what drivers these are.  If I can get one cheap it'd be worth trying to replace it.  That said... I'm aware I could switch x/y axis and do another test... but I'm certain I get the sm result.  

so... had set my machine coords, calibrated my x and y and saved the fixture... often when i do this and close out and come back to mach3 mill... my machine is rendered unable to move.  just gives a terrible whine from the motors when you try to jog it.  it's infuriating.  Then you have to replace the .xml file to get things back working... and this resets all the calibration and machine coords I just set.  Mach3 works fine after these procedures and only bugs out when you re-open it.  I'm sure that's a feature.  The "if you want your machine to just scream and not move" feature.

that said... I wrote a program to test slight angles and sure enough... breaks it every time.  just a simple one pass around a triangle with slight angles and x is off.  y seems to maintain.  if my understanding is correct... those little commands that would keep the spindle moving at an angle probably are very small bursts of electricity.  Those would be more susceptible to emi.  I'm hoping anyway.  Perhaps the replacing of the cables will fix some of that.

as I understand, I will want to run an actual ground wire to the spindle.  currently, there are only 3 wires going to it... I think I can wire up the 4th prong to ground, and then on the spindle side I can wire that prong to a lug washer and place it on the mounting screws.  Should be getting the wire sometime this week... and a few remaining connectors I need. 

Next up: adventures in rewiring.

TEST_ANGLE_PRECISION.rtf

further, last night I ran a new prog to create a shelf for my pc to sit on with a bunch of slots throughout for airflow... ran it like a champ.  no staggered edges... everything is perfect.  again, that is all square.  I used the sm post process that I used for the neck pocket.  No issues.

well, ran them 5 times each and go to z after 5 times each... looks pretty spot on to me.  I have a very fine crosshair indicating my zero and I set zero by eye before running (I removed my workpiece finally)... if it is off by any amount it is def not discernable by eye!

right on.  I like your enthusiasm.  I have a saying that this reminds me of: "I have all my code written... now I just need to get it to work"!!  here's to an uneventful first build!!

17 hours ago, curtisa said:

Another couple of tests to try.

Run the 'Diagonal test 0_5 inch' file 4 or 5 times and verify the cutter always ends at X0, Y0. Repeat the same process with the 'Diagonal test 0_04 inch' file and check again.

Diagonal test 0_04 inch.rtf 504 B · 0 downloads Diagonal test 0_5 inch.rtf 502 B · 0 downloads

roger that.  will do.  was thinking of creating a test... where I try different angles... because I suspect it's the very fine movements that are the problem.  45 deg doesn't seem to pose an issue... my guess is I'd have to try some 5, 2.5 and 1 deg angles and see what shakes.  anywho, I will go run these asap.  thanks again for the continued help!!

3 minutes ago, MiKro said:

Yep mine does the same most times as it has to do with the steps being off so slightly it is the best I can come up with. LOL

well I can def live with that kind of accuracy!!  Honestly, I'd like to get on to running some other things I have planned... but at this point I just don't trust this machine.  

23 minutes ago, curtisa said:

It is physically returning to the same spot though, not just what Mach3 is reporting?

yessir... 

7 minutes ago, MiKro said:

Mike, you do realize that 0.0001 is one tenth of a thousand. It is nothing to concern about. or one tenth of 0.001" most machines have 0.005" or more backlash in them.

Most things done in wood will change more than 0.020" from day to day due to humidity.

MK

yes, I do realize that .0001 is one tenth of .001 (hehe, couldn't help myself!)  wasn't expressing concern about this... that's what it always says when it returns to zero.  

.0001 on x and y each time.  just faster!!

3 minutes ago, curtisa said:

What about if you take that same code and change the line that reads:

To:

And run again?

roger that.  will do right nao.

grbl... i imagine that has all sorts of niceties for doing holes and such for pcb layout?  I've oft wondered why couldn't one just use mach3/aspire and load in a pcb shape... but I guess the answer is going to be that aspire is going to try to trace the outline of it... as opposed to using the fill.  anywho, congrats.

50 minutes ago, curtisa said:

One more code to try. This just does a handful of moves around a square pattern 2" x 2". At the end of each group of moves it should return to X0, Y0 and pause for 5 seconds for you to check if any drift is occuring. I've annotated the lines of code for you to check as they run.

Square and 45deg.rtf 897 B · 0 downloads

ok, so rebuilt my machine... tried it - sm result so... can rule out any issues with the pc!

ran you code... every time it went to xo/yo it was at .0001 which is typically what I get when I go to zero.  some interesting code there in the sense that it's commented well and I've learned some interesting things from it... like how to pause!  saving that for later.  

well thank you for that.

1 hour ago, MiKro said:

Well,I think that you should stop worrying about the guitar. You are trying to run before you crawl. It is time for basics. Establish an X0 Y0 that you will use from now on that is your machine coordinates. From there you always have a reference point. You need to make circles, squares and pockets first. these will help you understand your machine and machining. Without this knowledge you are wasting your time in my opinion. First get that down as this will also help troubleshoot machine problems. Too much is being said about possible Gcode problems as well as this, that, and chicken fat. Get back to basics. 

I also do not understand the fascination with Gecko? Yes it is used by many but it is not the endall of problem solution. .

Just my 0.02cents,

Peace,

Mk

well, I have done lots of preliminary cuts that were successful.  did a dust boot -a few times actually, did a spoil board 2x, did some circles in the form of coasters... did some engraving.  planed my materials using it, did my mounting holes - all went swell.  honestly each part of this guitar is broken down into simple steps... and even the body shape went fine.  

the reason gecko is of focus is that it is very much recommended for this machine.  lots of folks have had success with it.  further the implementation is very simple.  it has the drivers right on the control board which simplifies things a bit.  as I understand I can always add a motion controller to it later and go out via usb or ethernet.  It is quite popular and as such has a lot of support.  That is also the reason I chose mach3 - there is lots of support/threads regarding it.  In my mind these are all good places to start.

I realize there are lots of other solutions out there ucnc, etc.  My thoughts are that if I could start with something simple... I can continue to learn and perhaps learn why one might go the ucnc rout other than just to get usb or ethernet.  

I know your advice was to not use center-line - but I can't help it - that's what makes the most sense to me.  Everything I've done apart from the neck pocket has worked great using that.  It would seem that using ctr line is not an option when tiling but other than that... I don't see a down side to it.  that is indeed what I'm picking for my x0/yo.  then we come to machine coordinates... perhaps I need to learn something there but machine coordinates seem irrelevant in that they don't really matter if you just disregard them once you have workpiece coordinates (and assuming you don't have limit switches setup).  

Either way... I'm gonna keep throwing things at this... build a new pc today, rewire the steppers/vfd next week, perhaps build a new controller - until it makes sense and runs the way I expect.  Perhaps along the way I'll stumble into what is causing issue here but so far it sure doesn't seem like it's anything I'm doing.  

Any chance you could run my g-code and see if you have the sm issue?  If you don't: doesn't tell me much... but if you do - then it's def something in the g-code.  I don't want to lean on you too much, so totally fine if you don't want to.  

Either way - I very much appreciate the help from both of you - I've actually learned a lot just troubleshooting this so... I think I would probably have learned less if things just went w/o issue.  Please forgive if I don't always take your advice verbatim - it's not that I doubt you... it's just that I have to learn and that means sometimes not blindly following but picking the wrong path and making mistakes, and finding out why something isn't a good idea.

well I've found that learning is seldom pleasant in the short run so... if this is the worst thing that happens to me this year - I'm good!

again, thanks for all your help.  I do very much appreciate it.

24 minutes ago, curtisa said:

Well that could be something to explore, as your pickup cavity Gcode doesn't do any moves where X and Y are driven at the same time. The way the pickup cavity is clearing the material is in a zig-zag pattern, whereas the neck cavity has those little 45 degree traverse moves in between each spiral move. Seems odd that you were able to do a more complex shape like the body outline without issues though; that should be full of simultaneous X&Y moves.

 

Gecko G540 is parallel port only. Isn't your machine USB?

that's a very good point about the body and was thinking that earlier myself... totally blows my theory.  

yes, the gecko is parallel, I'd be adding on a pci slot parallel card to run that.  Would feel a bit like a step backwards.

I guess another thing could be the laptop it's running on.  I was warned that laptops are not great for cnc but I've setup the power options to never sleep and kept the power supply way away from the setup... that said, I have a tower as I just built a new pc that would be terribly overbuilt for this purpose but i don't have another use for it... just need to get a monitor.  Would be worth a shot anyway.

seems like what it is is... every time it tells it to travel at an angle but in the y direction... it is likely just going perfectly straight... but it things it's traveling at that angle... so it just ends up off by that angle compounded by however many passes it makes.  I think (and keeping in mind I have no idea) that it has something to do with this machine not following commands where it is moving along x and y at the sm time.  this is backed up by how it reacts to the keyboard jog.  if you hold two arrows... it only moves in one direction.  it can't receive two direction at once.  

when i bought this machine - I had read a lot about them... and everyone says the mechanical is solid, steppers are decent, spindle is great... but controllers are just not good.  I've read thread after thread of folks trying things and eventually buying a gecko g540 and then you never hear from them again.  Further - usually those sm threads have 5or 6 posts of folks telling the person to get a g540 and not waste time trying to troubleshoot the chinese junk!!

I hate to throw money at something w/o really knowing what's going wrong... but am getting a bit discouraged by the sm result over and over.  

I've bought the cables to redo them... if I buy a new controller or not I'll redo those cables so... should have them setup in the next week or so. 

I've found a g540 build but w/o spindle control for $450... and I could reuse this vfd or get a new one for $150... patch that into it with just a few components and that should do it... so thinking pretty hard on that!

thinking about that earlier run... with the pickup AND the neck pocket... even more baffled as it runs the pickup right afa going west of the centerline... and even stopped it after the pickup and we are still on center... but as soon as it starts running the neck pocket - if you stop it... ctr is no longer ctr.

sorry, first time i ran the wrong file (hadn't renamed it to .txt so mach3 didn't see it).  ran the file to do the profile and maybe saw a little bit of a 45 at the two corners where it's rounded but def square pattern at the uppermost corners.  

ran it... can't really see any 45deg movement.  just looks like it's moving in the square spiral pattern.