Atfer Hbot, go to CoreXY.

I was thinking, what happens if you rotate the part by 45 degrees and then print in the normal configuration?
If you have some interaction between X and Y it might give the same result as the sqrt(2) method.

If you want I can ZIP my marlin firmware and send it to you.
Let me think about your options

0.9° (400 step motor)
DRV8825 put to 16x.

Speed set to 100mm/s MAX. (yes marlin don't have double and quad step :-( ).

And ramps with arduino mega.

Quote
zelogik
...Speed set to 100mm/s MAX. (yes marlin don't have double and quad step :-( ).

?< >?

[github.com]
neverending "discussion" in german ... parts here:
[forums.reprap.org] [forums.reprap.org] [forums.reprap.org]

Quote
Ed Nisley's Blog
Marlin Firmware: Stepper Inerrupt timing
Marlin Firmware: Stepper Interrupt Timing
...a closer look at the step pulses as the motion starts from zero on the way to 150 mm/s:

x-axis-450-mm-s-50-mm-100-us-div-41-9-ms-dly.png

Quote
hardwarekiller

MAX Stepperrate 63,7kHz ( je höher desdo besser ) nix 30khz, oder 48 khz da kommt werder Repetier Sprinter noch Teacup mit !
Marlin blässt von allen Firmware die ich bisher kenne UND getestet habe was die Stepprate UND die Laufruhe betrifft alle in GRUND und BODEN

Stepimpuls nur 3.5 µsec !

Marlin1.jpg

Schritfrequenz 63,7kHz !!

Marlin2.jpg

Damit steht fest Marlin ist WEITAUS schneller !

Regards
Reinhold

P.S.: And for you as a limit searcher (and an octoprint user) (sry again german)

Quote

And ramps with arduino mega.

[sourceforge.net]

You're outside the operating envelope of ramps. 8x is the max safe microstep rate with 0.9 motors and small pulleys. The mega cant keep up at higher speeds.

Besides, anything above 8x is empty resolution with 5% steppers.

---

- akhlut

Just remember - Iterate, Iterate, Iterate!

[myhomelessmind.blogspot.com]

Reinhold,

Meine deutsch ist sehr schlecht, und google translate essen arsch.

Care to tanslate?

---

- akhlut

Just remember - Iterate, Iterate, Iterate!

[myhomelessmind.blogspot.com]

Hi akhlut

Here's your translation (sort of ... more like summary but I've got little time so just typing away)

My 2c's: RAMPS itself (define 'envelope') isn't limit to most anything...(except where that bunch of wiring is too flimsy )
Limiting factor is CPU Power Mega/Due...Quadcore, Firmware, in part (micro-)stepping and of course extruder control < IMHO >

[Hardwarekiller (german uP Hardware developer and member in Frankfurt Hackerspace) is talking (as part of a lengthy discussion) about what firmware is able to produce the most steps in a given period without falling back to 'tricks' like double or quadstepping]
+++++
so for standard software stepping with Pololu A4988 or DRV8255

...Angelo [udoo Developer] quotes:
Arduino Mega ca. 12kHz [[i]for Repetier but others point to recent version closer to 16kHz Performance[/i]]
Arduino Due ca. = 95 KHz [only Repetier version available, about 8x faster]


[MARLINs] MAX Stepperrate [is] 63.7kHz (the higher the better) not 30kHz or 48kHz neither Repetier, Sprinter nor Teacup can reach that!
Marlin, of all firmwares I know up to now AND have tested, does outperform the competition as to Steprate and running smoothness.
(lit.)run the competition into the ground
...[Steppingdata ]
That proves it Marlin is BY FAR faster.
+++++
My last comments: Dual/Quadstepping (2 or 4 stepper-pulses without a new calculation) isn't that good a solution (obviously)
but we are at a hardwarelimit here (it'sdone with interrupt timing = in hardware, look at the not evenly spaced peaks in Ed's data)

IMHO all of the above doesn't really/nearly "explain" anything zelogik sees
(with his motors "1 infill line missing" (out of "3") should be quite a few steps ... much more than 4 (quadstep) too )

hope I didn't bore y'all - got 2 run

Regards
Reinhold

Quote
Reinhold
...
IMHO all of the above doesn't really/nearly "explain" anything zelogik sees
(with his motors "1 infill line missing" (out of "3") should be quite a few steps ... much more than 4 (quadstep) too )

hope I didn't bore y'all - got 2 run

Regards
Reinhold

I don't think Zelogiks printer is missing infill lines, but it is shifted. So the middle line is closer to the first than to the next. (or the other way around)
So it may be possible that steps are delayed.

I guess the best test is to go back in microstepping. For a test I would say 4x or even 2x. Also a good test is to rotate the part by 45 degrees in the slicer. If there is backlash in the system it may come out better. (because is was better in the sqrt(2) mode)

Thanks all for your interests. I have try 1/4step and i have the same problem.

I don't remember to have this problem at first, and i don't remember if the problem have appear when i have changed for igus bush. And then have created a little Backlash.

Need to try to rotate file to 45°, i will restart slic3r as cura don't rotate infill :-)

EASY: 50% Infill ---> not equidistant ---> line shifting
---> else...

Regards
Reinhold

I doubt ustep problems: I'm using drv8255+32usteps now
... tested 0.9°+drv8255+32usteps with marlin...
"0" problems - but all on my (1st&different) printer

@zelogik: Cura does rotate, just select the object with the left mouse button and three icon will appear in the bottom left. The first one is rotate.

Edited 1 time(s). Last edit at 06/07/2014 03:40PM by jand.

Quote
zelogik
Need to try to rotate file to 45°, i will restart slic3r as cura don't rotate infill :-)

NOPE... Cura 14.03 generates 90° cross infill (anyway)
...and if you absolutely want to rotate use netfabb basic (takes 10sec)

Regards
Reinhold

14.03 + netfabb [www.netfabb.com]
I should work not surf

Edited 3 time(s). Last edit at 06/05/2014 03:57PM by Reinhold.

@zelogik Sorry I misunderstood you. Infill does not rotate. You were right

Edited 1 time(s). Last edit at 06/07/2014 03:39PM by jand.

@zelogik: I zipped the complete Marlin folder I used for my printer and placed it in my Public share So if you want you can give it a try.

So i have try your marlin package jand, and ......








It's the same problem, I have try to rotate the part at 45° with slic3r, and the circle come exactly on the same place when printing. So it's backlash. It's not gt2, it's not bearing, it's Iglidur problem. So i have the choice now ... quieter and backlash OR lm8uu and louder printer ...
Marlin don't have backlash compensation ....give a try with Repetier.

Edit:

After things I have completly forget to try.

Move -0.1 and 0.1 mm in X and Y ... seem that I have 0.1mm of backlash in Y direction... need to know why in Y and not in X.

Quote
zelogik
...So it's backlash. It's not gt2, it's not bearing, it's Iglidur problem...

Which one do you use ?
RJMP-01-08
RJZM-02-08.
(and sorry but I am too lazy to read all 7 pages to find out)

Among others a friend (who's running a homebrew ultimaker setup with 10mm rods and Iglidur) was discussing in detail about "Iglidur problems" (backlash)
with another "french" (sic) guy and they both agreed that it works (better) if you cut open the bearing and put 1-2 grub screws (vis sans tête) to adjust.
(sry I wasn't really interested and listening in detail and friend is on vacation until wed. ;-/ ... but they both are still using IGUS_not_ metal)

In another (same day) "joint discussion" a tech support of igus said that:
"iglidur® Gleitlager are designed that they fit into an H7-tolerance "holder" with the nominal bearing outer dia. which fixes them via pressure.
(called "Einpressübermaß" i.o.w. it's outer dia is 0,1-0,25mm larger than nominally stated) - that precision and pressure are needed.
In Iglidur labeling "P" is for precision ... or fits without the above pressure (or without cutting and grub screws which is cheaper I learned).

Sorry not my field of play (obviously yours) so before further listening to hearsay maybe look at:
[www.igus.eu]

Regards
Reinhold
_{(the LM8UU overstocker)}

It's GSM-0809-10, inserted in plastic spacer with a 9mm hole (with a 9mm H7 reamer), there are inserted really tight. (at 0.55€/pcs for 25pcs it's cheap ). Finally when I compare between lm8uu and Igus, both have gap and don't like moment.

After all bad 'conclusion', it's seem that only Y (when motor move on the opposite way) have backlash, not X. I need to make a brain-compute phase to find why Y and not X with a CoreXY setup ...

[fr.wikipedia.org]

Quote
zelogik
I need to make a brain-compute phase to find why Y and not X with a CoreXY setup ...

As long as you (can) trust your "findings" that's OK (with me)
...dare I say that (in my experience) "mechanics" tend to find "mechanical problems"

Regards
Reinhold
...who has a chirurgien friend whom he is not visiting for medical problems
and who "feels" that not all that you showed is explained with xy backlash ...(ducking&running)

This might be a stupid question as I haven't read the whole thread yet, but why is the belt path the way it is near the motors? Couldn't you remove the 90 degree angle redirection and save 4 pulleys? Wouldn't removing 4 pulleys reduce friction and backlash a bit?

Also, I'm not sure if I've seen it linked in this thread but the fabtotum uses a two level CoreXY belt path. Anybody has experiences with this or are there already a open design using this?

@dejay: the reason with the 4 more pulley are wrote on the previous page :-) to resume, it's for easier belt alignement and easier belt tightening.

@reinhold: i don't have forget you for endstop :-)

For Reinhold



X+ and X-


Y+


Y-


Edited 1 time(s). Last edit at 06/08/2014 05:23PM by zelogik.

@zelogik: Thanks !

Quote
Dejay
fabtotum uses a two level CoreXY belt path[/url].

!NOPE! NO corexy - that honor belongs to Ilan Moyer!

I wrote a long reply ... which in the end ... just resembled a diatribe ... ... - erased it ...
In short: "it's commercial speak" "How to make cost cut engineering look good"
In my wordbook indeed "to mitigate" goes by "lessen the gravity of (an offense or mistake)".
But then I reprap and have nothing to sell!

I'll wait for a look at their "supercar racing chassis"
YOU might want to read the whole thread here.

Regards
Reinhold

Quote
zelogik
@dejay: the reason with the 4 more pulley are wrote on the previous page :-) to resume, it's for easier belt alignement and easier belt tightening.

Ah lol of course just the previous page! Thank you for answering anyway

Quote
Reinhold

Quote
Dejay
fabtotum uses a two level CoreXY belt path[/url].

!NOPE! NO corexy - that honor belongs to Ilan Moyer!
...
YOU might want to read the whole thread here.

I am reading as fast as I can but I keep getting sidetracked with links to cool videos and shops!

Ok, done lol but I didn't mean to imply they invented the coreXY, just that they adjusted it?
I still don't understand all that they wrote on the blog post, and probably the belt crossing is less important
than they think. Is there a flaw in their belt path?
Overall, I think their fabtotum is a really neat deal for a printer, scanner AND mini cnc mill.

Thanks you all for the info and for the great design!

Sorry for going a bit off topic here, I hope you won't mind

I'd like to build (or actually I would like to have ) a printer with a really large print volume. Something like a 600 to 1000mm cube print volume. And it should be able to carry a dual direct extruder, or rather a quad extruder that can directly feed something like the kraken but with only 2 steppers.

Is there a coreXY design that uses aluminium extrusion with a wheel carriage for linear motion instead of heavy stainless steel rods? I figure aluminium extrusions would be inherently less prone to bend compared to rods and would have less weight. Basically like the delta / kossel mini wheel carriages that also run on extrusions.
This would reduce the weight of the moving carriage so that less force is needed for acceleration and might let you get away with the usual nema17 stepper motors, or at least use normal stepper drivers to drive nema23 motors (current limit).

As I understand coreXY it applies the forces in a way that would allow a kind of wheeled carriage, while hbot would probably not.

Aluminium extrusions won't be as precise as steel rods so you might need to calibrate bed leveling, but this might be a good idea anyway because of the large print bed will also be prone to not be as even or have slope.

Alternatively you could of course you could just use round hollow aluminium tubes, polish them and use them for the x axis and keep with steel rods for the y axis, since they don't contribute to the moving weight. This would even keep the current zelogik design.

Of course there would still be problems with heating a large print bed.

Would this be a sensible approach and what problems would you see with this?

Quote
Dejay
Sorry for going a bit off topic here, I hope you won't mind

I'd like to build (or actually I would like to have ) a printer with a really large print volume. Something like a 600 to 1000mm cube print volume. And it should be able to carry a dual direct extruder, or rather a quad extruder that can directly feed something like the kraken but with only 2 steppers.

Is there a coreXY design that uses aluminium extrusion with a wheel carriage for linear motion instead of heavy stainless steel rods? I figure aluminium extrusions would be inherently less prone to bend compared to rods and would have less weight. Basically like the delta / kossel mini wheel carriages that also run on extrusions.
This would reduce the weight of the moving carriage so that less force is needed for acceleration and might let you get away with the usual nema17 stepper motors, or at least use normal stepper drivers to drive nema23 motors (current limit).

As I understand coreXY it applies the forces in a way that would allow a kind of wheeled carriage, while hbot would probably not.

Aluminium extrusions won't be as precise as steel rods so you might need to calibrate bed leveling, but this might be a good idea anyway because of the large print bed will also be prone to not be as even or have slope.

Alternatively you could of course you could just use round hollow aluminium tubes, polish them and use them for the x axis and keep with steel rods for the y axis, since they don't contribute to the moving weight. This would even keep the current zelogik design.

Of course there would still be problems with heating a large print bed.

Would this be a sensible approach and what problems would you see with this?

I made a corexy machine out of vslot aluminum extrusion and vslot wheels. It runs very smooth with no appreciable flexing of the extrusions. The problem I am facing is with my extruder/hotend. I can't come up with an extruder design that can match the mechanical capability of the printer.

Quote
tkole
I made a corexy machine out of vslot aluminum extrusion and vslot wheels. It runs very smooth with no appreciable flexing of the extrusions. The problem I am facing is with my extruder/hotend. I can't come up with an extruder design that can match the mechanical capability of the printer.

Sounds interesting! Of course openbuilds vslot is made to be specially suited for linear motion, so they are probably much more precision than the usual extrusion that are just for structure building. So you would use cheaper extrusions for the frame and the more precise vslot for the carriage. I also found this shop that ships them from europe.

Do you have any link to discussion / photo / designs for your project? I've only found this CoreXY Build thread. What are your problem with extruder?

When you say that is what you mean you can't extrude at a high rate of speed reliably. What are you using for an extruder now if you look at some of my previous posts in the mechanical forum on this site you'll see something I came up with it is rocksolid reliable my extruder with an E 3-D hot end. This is the second one I built and I've never had a failure grinding the filament refusing to feed in my opinion the feed wheels are too small of a diameter and don't allow enough teeth to be embedded in the filament at any one time. I use a mig welder feed wheel

Dejay, I don't have a thread or build blog, but here are some pics below. The black extrusion is vslot and the silver extrusion is from Misumi. You will notice my E3D heat block dangling in one photo because I am currently fixing a broken piece.

cnc dick, you are correct. My hotend can't extrude fast enough. I have tried mk7 and mk8 direct drive as well as a geared wade's for my Bowden setup but I think the 3mm filament diameter is what is killing my speed. I saw your post about the mig welder feed wheel and completely agree with you that most extruders are plagued by the lack of tooth grip. After you posted that, I asked about where you got the mig feed wheel but never got a response. I'll try anything at this point.











Printing movie clip

Thanks for sharing the pictures tkole! Looks like a very neat design.
What would you recommend for a 600 mm or even 1000 mm wide x axis gantry? You've gone with the 20x20 but probably 20x40 or 20x60 V-Slot is stiffer without that much more weight.

Going back to the belt tightening and the additional edge the belt path takes, I've been thinking about this and I had an idea, but then I found this one that is pretty close to what I had in mind. It looks like a very compact, printable and near perfect belt tensioner. Or is it important that the belt is tensioned with a kind of elastic spring?

I just got my drive wheel off eBay my wheel was meant for 1.2 mm wire and I use 1.75 filament a little bit under works the best but you're going to have to build everything to put it together. mig welders have two different kind of drive wheels the one you want is the serrated one I don't know if you can find one big enough for your 3 mm but maybe I just don't know if they make make welders to feed wire that large

Edited 1 time(s). Last edit at 06/09/2014 09:31PM by cnc dick.