Unsolved Mystery. Weird Z0 behavior around build perimeter.

[Jimustanguitar]

Hey JJ, how far off was your machine before the firmware change? The carriage offset change in the new firmware didn't help my machine, I'm still off by about 19 thousandths.

New file from the GitHub, cleared EEPROM, loaded new firmware, recalibrated... What am I missing? Has this been the silver bullet for anyone else's machine?

Here are my results:

[youtube]http://www.youtube.com/watch?v=ej_n71nQ5-k[/youtube]

[Mac The Knife]

Before I wipe my firmware settings out again, How do I retrieve them from the printer, then reupload them after I make the change?

[Mac The Knife]

Looking at the firmware I downloaded last night from github, Line 752 has the carriage offset at 38.4, not 35. I assume someone updated this setting on github? I also ASSume, that I can't simply take a peek at my current firmware settings, and make a few tweaks.

[Mac The Knife]

What I finally ended up doing. I opened Repetier, connected to the printer and exported the EEPROM settings. Shut down Repetier, and fired up Arduino IDE. Wiped the EEPROM on the printer, and installed the latest firmware from Github. Shut that down, reopened Repetier,, connected to the printer, and imported the EEPROM settings I had exported earlier. Now I'll check and reset the calibrations.

[RegB]

Looking at the firmware I downloaded last night from github, Line 752 has the carriage offset at 38.4, not 35. I assume someone updated this setting on github? I also ASSume, that I can't simply take a peek at my current firmware settings, and make a few tweaks.

The article on the SeeMeCNC home page seems to think it should be 37.5

[Mac The Knife]

The firmware they uploaded after their announcement was 38.4. Who knows?

[RegB]

The firmware they uploaded after their announcement was 38.4. Who knows?

I have this and SO MANY OTHER things to check when I do my build.
I will do my build when I get my kit, HOPEFULLY two days and a few hours from now.
I will try to contribute SOMETHING when I get to these settings.

The wait is the hard part, the wretched WAIT (-:

[bvandiepenbos]

Hey JJ, how far off was your machine before the firmware change? The carriage offset change in the new firmware didn't help my machine, I'm still off by about 19 thousandths.

New file from the GitHub, cleared EEPROM, loaded new firmware, recalibrated... What am I missing? Has this been the silver bullet for anyone else's machine?

Here are my results:

[youtube]http://www.youtube.com/watch?v=ej_n71nQ5-k[/youtube]

Jim, so did the firmware change fix your printer like SeeMe thought it would?
Anybody else had a measurable improvement with this fw change?

[Polygonhell]

I assume what you are seeing is basically this

I took the matlab model from here http://wp.boim.com/?p=67 and tried to reproduce the error you were seeing.
It's a combination of RodLength set incorrectly, then compensating for the error with DeltaRadius. In the image above I introduced a 1mm error in the RodLength, then tweaked DeltaRadius by 0.3mm to more or less compensate.
The issue is that both introduce doming errors, but because the errors aren't the same degree of function, if you compensate for one with the other you get the high order harmonics you can see in the graph above.

As to fixng it, I would start by trying to get an accurate measurement of the RodLength

[Jimustanguitar]

I assume what you are seeing is basically this

Pretty much, yep. For some reason, the proposed firmware fix didn't have an effect on my machine. Still waiting to hear from others.

[Polygonhell]

The firmware fix, would have just changed the printer radius, that may well be necessary, but the crux of the issue is it's two errors interacting. The first thing you need to do is accurately measure the the length of your arms, center of pivot to center of pivot. Once you have that accurate and in the firmware, it should just be a question of getting the delta radius correct.
Unfortunately I can't think of a good way to get both values correct iteratively.

[snoman002]

The firmware fix didn't have an effect because the parameters in the software for your machine don't match the actual measurements of the machine. Polygonhell showed something I mentioned previously.

On the same note, improper tower rotation will cause the low points between towers to be higher or lower than each other.

Tweaking rod length and radius should allow one to get to the point where the area between towers averages zero, from there tower radius could be tweaked to get close to a true zero.

Calibration using a large radius and small radius 6 point star, plus center, should give and indication of which direction to make changes (longer or shorter rod length).

[RegB]

I assume what you are seeing is basically this

Geometry error.jpg

I took the matlab model from here http://wp.boim.com/?p=67 and tried to reproduce the error you were seeing.
It's a combination of RodLength set incorrectly, then compensating for the error with DeltaRadius. In the image above I introduced a 1mm error in the RodLength, then tweaked DeltaRadius by 0.3mm to more or less compensate.
The issue is that both introduce doming errors, but because the errors aren't the same degree of function, if you compensate for one with the other you get the high order harmonics you can see in the graph above.

As to fixng it, I would start by trying to get an accurate measurement of the RodLength

How "accurate" would that need to be ?
I am thinking;
Take the rods off, U-joints and axles too.
Assemble pairs of arms, U-joints and axles with approximate spacing.
Lay them on the bed and heat to maybe 32C (measure above a running bed to know EXACTLY what temp).
Measure outside of axle to outside of axle, subtract axle diameter.
Maybe average it for the three pairs ?, maybe that would be UBER fussy ? (-:

[Jimustanguitar]

Delta radius is really the only variable out of the two. The arms are manufactured quite precisely. As a test, I picked up a set of the v2 clip style arms, and I can confidently report that this issue is happening regardless of which arms are used. V1, V2, and Trick Laser carbon arms all behave similarly.


Another note... I replaced my belts and pulleys and motors last night, so I'll know later today if any of those parts could have been out of tolerance.

[Polygonhell]

I assume what you are seeing is basically this

Geometry error.jpg

I took the matlab model from here http://wp.boim.com/?p=67 and tried to reproduce the error you were seeing.
It's a combination of RodLength set incorrectly, then compensating for the error with DeltaRadius. In the image above I introduced a 1mm error in the RodLength, then tweaked DeltaRadius by 0.3mm to more or less compensate.
The issue is that both introduce doming errors, but because the errors aren't the same degree of function, if you compensate for one with the other you get the high order harmonics you can see in the graph above.

As to fixng it, I would start by trying to get an accurate measurement of the RodLength

How "accurate" would that need to be ?
I am thinking;
Take the rods off, U-joints and axles too.
Assemble pairs of arms, U-joints and axles with approximate spacing.
Lay them on the bed and heat to maybe 32C (measure above a running bed to know EXACTLY what temp).
Measure outside of axle to outside of axle, subtract axle diameter.
Maybe average it for the three pairs ?, maybe that would be UBER fussy ? (-:

It needs to be accurate to 1/10th's of a mm
The easiest way would be to put pins through the arm ends, use a caliper to measure the distance and subtract off the diameter of one pin.
I'd do it, but I don't have a caliper big enough.

[RegB]

Since I removed the axle clips (in favor of rubber bands) the axle ends on my machine are clearly accessible.
I will try to borrow a long vernier from a machinist friend and measure rod length right after about a 1 hour print above a ~90C bed with a ~225C extruder running.

I may be wrong 'bout this, but axle to axle centers seems better than rod end to rod end.
I know, they SHOULD be the same - and as long as the U-joints are spot on they will be.

[DavidF]

Since I removed the axle clips (in favor of rubber bands) the axle ends on my machine are clearly accessible.
I will try to borrow a long vernier from a machinist friend and measure rod length right after about a 1 hour print above a ~90C bed with a ~225C extruder running.

I'm thinking, not only measure the length, but measure the distance between each arm at the cheap skate and the effector.

[Jimustanguitar]

Motors, Belts, and Pulleys didn't do the trick for me. I'll try swapping my Rambo board this weekend.

[Jimustanguitar]

Swapped Rambo boards, and that didn't do the trick. Also tried Marlin instead of Repetier (loaded from someone else's computer too)... I'm running out of conspiracy theories.

Things that have been replaced:

All wooden parts (V1 to V2 conversion)
Build plate (window glass, boro glass, and machined aluminum)
Arms (V1, V2, and Trick Laser)
Carriages (V1, V2)
Effector Platform (V1, V2)
Motors (swapped from Wantai to Kysan)
Pulleys (stock plastic to all aluminum)
Belts (different brands)
Rambo board (both 1.1a)
Wiring (not troubleshooting, just accessorizing)
Endstops (different brands)
Firmware (multiple versions of Repetier and 1 of Marlin)


Things that have not been replaced:

skate bearings, covers, and cams
belt idlers and spacers
aluminum extrusions


Unless my extrusions are bent, I can't imagine any of the parts that I haven't replaced yet having anything to do with this... I have extra aluminum to try this with, so that's next on the list. I've also got a smoothieboard to try in the next couple weeks, and a sanguinolulu if needed.

Am I crazy or is there something obvious that I'm overlooking? My only remaining theory is that there might be some gremlin hiding inside my Rambo board version since I tried 2 of the same version. Is anyone else who's having this issue using something other than a 1.1a? Maybe there's a math error and switching to a different microstepping setting or swapping the pulleys for a different tooth count is next?

Any suggestions besides seeking professional mental help?

[geneb]

If I were you, I'd bring it to Dr. Guanu and see if he can work his magic. The dude's a frigging genius when it comes to calibrating cranky machines.

g.

[guanu]

I fix machine for happy printing time! it work very much good

[Jimustanguitar]

I fix machine for happy printing time! it work very much good

Thanks, Andy. JJ offered help as well. It's just tough for me to get it out to you guys on a normal work day. Any chance that somebody would be there for me to drop it off over the weekend?

[geneb]

What you need is an abnormal work day.

g.

[626Pilot]

I've also got a smoothieboard to try in the next couple weeks, and a sanguinolulu if needed.

Let me know if you decide to set up the Smoothieboard. I'm adding tower rotation code to the Smoothie firmware. It will require a Z probe. There is some experimental code for Smoothie that tries to adjust delta radius, arm length and tower rotation simultaneously but it usually trips over itself. My code tries to do one thing at a time.

[RegB]

Thinking about tower rotation...
How hard would it be to say;
Lay a straight edge across towers X & Y, scribe a line, find the mid point.
Strike a line from the side of the Z tower to intersect the X-Y line.
Wouldn't THAT tell if the Z tower is facing the right way ?
Repeat for Y-Z against X tower and for X-Z against Y tower.

Now I'm thinking what REALLY matters is the cheapskate axles - THOSE need to be a PERFECT 120 degrees to each other.
I think if the cheapskate axles were temporarily replaced with very long pieces of very straight drill rod one could measure the length of the sides of the outer triangle where they intersect - somewhere around 26 - 27 inches at a guess.
This feels like a reasonable technique for amplifying the error in order to detect & correct it.

Getting them the same length would get the angles between them the same.
If they are wildly off... I would double check for a sloppy build and/or sloppy cheapskate adjustment.
Some adjustment may be possible by loosening up the 1/4 inch tower bolts to merely snug, then twisting/tweaking towers and tightening down again.
Shimming the U-joint carriage bases against their cheapskate inside plates would work, but I would want to tighten up on everything else first.


How "upright" ?
I would HOPE that the distance from the top of any tower to the base of an adjacent tower is some constant.
Quick and dirty check; Cut a metal straight edge (steel yard stick) down to about 29 inches, exact length doesn't matter.
Lean it up against the Z tower with its bottom against the base of the X tower, mark the Z tower carefully.
Move it around to the other side of the Z tower, with its bottom against the base of the Y tower, mark the Z tower on that side.
Use a carpenter's square to transcribe the marks to a face where they can be compared.
Repeat for X, repeat for Y, they should all be the same on both sides of all three towers.