CoreXYZ

[Ryan Carlyle]

Here's the design for my CoreXYZ concept. It's an extension of the CoreXY belt path into 3 dimensions. See attachments for full theory write-up and Sketchup model. License CC-BY-NC-SA.  

The really cool thing about this design is the possibility of a CoreXYZZY variant. (Suggested by Joseph Chiu.) The entire printer can be flipped on its side to exchange the Y and Z axes. Naturally this will require rotating the extruder carriage, having two build plates, and doing some cleverness around tilt-activated relays to flip the endstops and such.

Design in progress.

[Gary Crowell]

Fascinating write-up.

I suspect you've already done this, but I couldn't help working out the logic for the XYZ - ABCD  translation.

ADIR = (XDIR & XSTEP) + (YDIR & YSTEP) + (ZDIR & ZSTEP)

BDIR = (XDIR & XSTEP) + (!YDIR & YSTEP) + (ZDIR & ZSTEP)

CDIR = (!XDIR & XSTEP) + ( !YDIR & YSTEP) + (ZDIR & ZSTEP)

DDIR = (!XDIR & XSTEP) + ( YDIR & YSTEP) + (ZDIR & ZSTEP)

STEP = XSTEP + YSTEP + ZSTEP

As a sanity check I fiddled that together in a simulator and it works just as expected.

NAND-NAND equivalent to AND-OR.

That's really just a handful of gates.  Half of those go away if you're just doing XY - AB.  20 years ago I'd have just done that in a single PAL/GAL chip, but I don't think those exist anymore.  While I've done some work with large FPGA's, I have no idea what the current solution is for small logic implementation.  I'll have to look around.  It may just be simplest to implement it in discrete logic; I count 4 cheap chips.

There is a gotcha there.  Since the XYZstep is part of the direction logic, it violates the set-up time of the driver.  It'll be necessary to delay the step signal to the drivers to meet the direction set-up.  I don't see any problem in doing that.

Gary

[Ryan Carlyle]

I had gotten as far as a truth table and "most of" the gate schematic. (Just another iPad sketch.) But it's been about 10 years since I've done any gate-level logic design and I've forgotten how to most of it. So I was looking at something like, ehh, 10 ANDs, 5 ORs, and 12 inverters. A lot sloppier than your version. 

Seems like this would be pretty easy to wire up on a breadboard with a few hex-NAND DIPS. I suppose the delay requirement will depend on the stepper driver? The standard Makerbot/Clone driver is an A4982, which needs: 

So we'd need a min 200ns, max 800ns delay between a 1us input step pulse and the output step pulse? (How wide is the actual Mightyboard step pulse?) We're rapidly getting beyond my competence level here  :-)

[Dan Newman]

On 27/07/2014, 8:55 AM, Ryan Carlyle wrote:
> I had gotten as far as a truth table and "most of" the gate schematic.
> (Just another iPad sketch.) But it's been about 10 years since I've done
> any gate-level logic design and I've forgotten how to most of it. So I was
> looking at something like, ehh, 10 ANDs, 5 ORs, and 12 inverters.

To which one then might apply a Karnaugh map and see if that results in a
simplification. But looks like Gary did a mental equivalent.

Dan

[Joseph Chiu]

Make sure that there are sufficient cover-terms for de-glitching, if this is an asynchronous design!

--
You received this message because you are subscribed to the Google Groups "3DP Ideas" group.
To unsubscribe from this group and stop receiving emails from it, send an email to 3dp-ideas+unsubscribe@googlegroups.com.
To post to this group, send email to 3dp-...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/3dp-ideas/53D56AD3.80601%40mtbaldy.us.

For more options, visit https://groups.google.com/d/optout.

[Gary Crowell]

I briefly looked at small programmable logic.  What I saw was based on the old 22V10 PAL cores that I was familiar with, but much expanded.  While inexpensive, $2-$5, the smallest package I noted was a 44-TQFP, and then it has to be programmed.  Unless there's other logic you wanted to add in, it looks like a gate level breadboard would be quick and easy.  I've added in a '123 for the delays which makes it 5 chips.  I think it came out with one inverter left over that I didn't show.

Karnaugh maps were in the back of my mind, but no, it wasn't that formal.  I do think that this is going to be as reduced as it gets though.

If the outputs from the microcontroller are well-behaved, i.e., all direction outputs are settled and remain so during a step, and only one step at a time, I don't see any potential for glitches.

A common​​ 74xx123 should work fine for the delays.  Proper choice of tolerance and stability on the RC's of course.  The actual values will change depending on the logic family used; the values shown give a 400ns delay for setup and a 1.3ms step.  Have to check what is needed for a DRV8825.

​

 ABCD LOGIC.jpg

​

To unsubscribe from this group and stop receiving emails from it, send an email to 3dp-ideas+...@googlegroups.com.

To post to this group, send email to 3dp-...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/3dp-ideas/CAL5_QBMPYes8%2B2RT5bWk869PPSq4_XX%3DeHe6BzVM4pfKO4mSFw%40mail.gmail.com.

For more options, visit https://groups.google.com/d/optout.

--

----------------------------------------------
Gary A. Crowell Sr., P.E., CID+

Linkedin  Elance  KE7FIZ  Things RocketryCNC

[Gary Crowell]

Sorry, the scope waveforms don't show up great on that .jpg, but they are there; thin red lines.

And ugh, do I have to drag out my old 16500 logic analyzer?  I think I'd rather spend a couple hundred and get one of the new little ones instead.

Gary

[Joseph Chiu]

I use a LogicPort - 24 channel logic analyzer with protocol decoding... I wish it had a deeper buffer, but it's otherwise perfect. I love the fact that I can carry my logic analyzer in my carry-on luggage. :)

To view this discussion on the web visit https://groups.google.com/d/msgid/3dp-ideas/CAB%2B-abtqHv2T74f_%3DsMEDvRA0AWVbVikL5H31T6wUm8i9Fe7Fw%40mail.gmail.com.

[Joseph Chiu]

Err, 34 channels!

[Gary Crowell]

On Sunday, June 29, 2014 5:00:53 PM UTC-6, Ryan Carlyle wrote:

Design in progress.

So, how much in progress?   If you want to try the XYZ>ABCD hardware conversion, I could cut a board easily...

Used DIPs 'cause SMT would require vias;  not complete is power and discrete R's & C's.  I have an LPKF pcb mill that would cut this easily.

Gary

[Ryan Carlyle]

Man, you're awesome, Gary.

How would you feel about putting 4x stepper driver sockets on there, with pass-throughs from a ribbon cable to the existing mightyboard sockets for vref, etc? I can connect everything manually if you don't feel like doing that much work, but it'd be nice. Hmm. Adding another driver's worth of current probably justifies running dedicated wiring to the PSU for the sub-board. Lemme sketch something up.

I had back-burnered the CoreXYZ build due to wife aggro on 3DP parts purchases, plus the overall difficult implementation scope. So I've been tweaking/upgrading my CoreXY and gradually accumulating parts for a third bot. But if you're helping on the electrical design, I can get back on CoreXYZ and perhaps strip my CoreXY for the frame and electronics. Wouldn't take a lot of extra non-printed parts -- just the spectra line drive, some more rods, etc. Plenty of parts design to do but that's not difficult. Hard part has always been getting the stepper control sorted.

[Ryan Carlyle]

Quick sketch attached, assuming it goes through from my iPad. This is pulled straight off the Mightyboard Rev E schematic. 

Need to pull these from the Mightyboard stepper driver sockets:

Either X or Y Enable (Not Z to avoid Z-hold settings)

Either X or Y VRef (4 symmetrical motion axes = same current to all, I say)

X-Dir

X-Step

Y-Dir

Y-Step

Z-Dir

Z-Step

+5v

Ground?

Need a separate screw terminal for +24v and Ground wired straight to the PSU with some decent gauge wire.

[Gary Crowell]

Easzy Peazy.  Will have a sch tonight or tomorrow.  Layout and cutting will be no prob.

I have an ulterior motive.  I've got a Misumi2525 frame 300x300x600mm that I was planning to wire as CoreXY, but I haven't bought the Zscrews yet...

Err, 34 channels!

To view this discussion on the web visit https://groups.google.com/d/msgid/3dp-ideas/CAL5_QBMPYes8%2B2RT5bWk869PPSq4_XX%3DeHe6BzVM4pfKO4mSFw%40mail.gmail.com.

For more options, visit https://groups.google.com/d/optout.

--

----------------------------------------------
Gary A. Crowell Sr., P.E., CID+

Linkedin  Elance  KE7FIZ  Things RocketryCNC

--

----------------------------------------------
Gary A. Crowell Sr., P.E., CID+

Linkedin  Elance  KE7FIZ  Things RocketryCNC

--
You received this message because you are subscribed to the Google Groups "3DP Ideas" group.

To unsubscribe from this group and stop receiving emails from it, send an email to 3dp-ideas+unsubscribe@googlegroups.com.
To post to this group, send email to 3dp-...@googlegroups.com.

[Joseph Chiu]

But couldn't you have gotten by with, say, a direct belt drive on the Z stage, though?   The Phoenix 3D (Easy3D) guys have been doing that and seemed to have a pretty decent success with their machines operating that way?

To unsubscribe from this group and stop receiving emails from it, send an email to 3dp-ideas+...@googlegroups.com.

To post to this group, send email to 3dp-...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/3dp-ideas/fafc1359-de21-4198-99f3-727cd158c1d2%40googlegroups.com.

[Ryan Carlyle]

Ah-hah. I've been looking at a 300x300x1000mm frame for CoreXYZ but haven't bought it yet. Trying to find a good option for supported rails to suit -- I was thinking about using 2040 or something similar to mount supported rails to. I doubt 4x unsupported 10mm rods will be stiff enough when the carriage is slinging around at mid-span. Might just give up and use the extrusions themselves for Z rails though. It's a little more complex than you might think, because gantry skew is how you tram, so some angular play is required where the XY gantry meets the Z rods.

One thing I haven't sorted out yet is how to do the spectra drive. For the first rev, I'll probably just print smooth pulleys and use fisherman's knots to make loops that can roll over grooved pulleys without issues. Maybe steal some Tantillus/Ingentis parts.

[Ryan Carlyle]

Issue with direct drive belts on Z is the minimum step size being kind of large for low layer heights. (With normal 1.8 degree steppers and ~18t pulleys.) The Z resolution may be too rough unless you use a gearbox or pulley arrangement to "gear down" the Z axis.

Yes, this is an issue with CoreXYZ too. The X,Y,Z resolutions are necessarily equal. And the Atmega is very limited on clock cycles to handle high steps/sec rates. So fast XY motion means rough Z resolution, and high Z resolution means slow XY motion. I haven't worked through all the implications of doing a "gear reduction" for CoreXY. It should be easy enough to put a loop in the Z-legs to give them a 3:1 mechanical advantage relative to XY, but I haven't worked through all the possible implications of that. (It should tilt the planes defined by each belt from 45 degrees to 15 degrees or something like that.)

One alternative that crossed my mind is using the stepper multiplexer circuit to double-step on XY moves and single-step on Z moves. That doesn't actually increase the Z resolution but it does reduce the processing load on the Atmega to fire the stepper interrupt for X and Y, which means on net the total XYZ resolution can be made finer. Same basic outcome. I think I'll build the symmetrical version first to see if it performs ok as-is, then try the 3:1 Z, and then save the electric fix as a last resort.

[Vernon Barry]

Just jumper the microstepping in the drivers since you are breaking it out.

 

Just saying 1/8 stepping is plenty and leave 1/16th for Z.

 

Makes the whole setup a lot easier.

[Ryan Carlyle]

I wanted to do that too, but it won't work here. All four CoreXYZ steppers have to fire during every Cartesian move. What determines carriage direction is what direction combo they fire in, not which steppers fire.

I'd love a system that changed the microstepping level depending on which direction it was stepping, but (didn't you tell me this?) the drivers lose microstep position registration when switching stepping levels. So the shortcut is double-stepping X and Y to make effective 1/8th stepping. Or just use the gantry pulleys.

[Ryan Carlyle]

Alright, I'm buying parts now. Just ordered a Misumi frame kit and all the linear motion hardware for a 5:1 pulley reduction Z motion implementation. With 18mm diameter smooth stepper pulleys, this will give ~57 steps/mm on X and Y and ~285 steps/mm on Z. In other words, a fast bot for large prints.

I'll start modeling up the motion mechanism parts over the weekend. There's actually only a few CoreXYZ-specific parts to model -- the carriage, the X ends, and the Y ends mainly, then a few angled pulley mounts to attach stationary pulleys to the 2020 frame. Should be pretty easy.

Most everything aside from the motion mechanism is just going to copy my CoreXY bot. HBP, electronics, extruder, etc all the same.

[Ryan Carlyle]

Work in progress...

[Ryan Carlyle]

X ends and Y ends done:

[Ryan Carlyle]

Motor arrangement, tensioning, and tramming sorted:

(the idlers on the 2020 rails near the motors will be "loosen, slide, tighten" to adjust line tension)

[Gary Crowell]

Looking really interesting.

Here's the schematic for the XYZ-ABCD converter board.  

• I've got a breakout board for the MIghtyboard Botstep sockets - it uses a 10-pin header/cable to bring the signals to the converter board (I've got a bunch of 10-pin cables here somewhere).
• I added step select jumpers for AB, and CD.
• There are jumpers to reverse any ABCD direction.
• Sockets to mount the ABCD Botsteps.
• Mightyboard-type Molex 0.156x4 headers for motor connection.
• On-board 5V generated from 24V.

I have the board routed except for a few impossible routes and some cleanup.  Not too late to make changes though, if you have any other ideas.

Gary

--
You received this message because you are subscribed to the Google Groups "3DP Ideas" group.
To unsubscribe from this group and stop receiving emails from it, send an email to 3dp-ideas+...@googlegroups.com.
To post to this group, send email to 3dp-...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/3dp-ideas/e52a141a-8d77-41ab-a684-2eaf013edbf2%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[Ryan Carlyle]

Looking really good. Here's a few thoughts:

• The step select jumpers are great. Caveat is that I think all the drivers have to be at the same microstepping level. If you put two drivers at 1/8th and two at 1/16th, motion won't be Cartesian any more. So you probably ought to use the same pair of jumpers for all four motors rather than breaking out two separate pairs. (Just dummy-proofs it a little.)
• Is there any risk from having two different 5v regulators in the system? Versus pulling the main 5v leg off a board driver socket. IE someone connects the two 5v legs together somehow. Not sure how the risk tradeoff here plays out since the board 5v is under-spec'd on the older hardware. I don't have the answer, just asking the question.
• The direction reversal jumpers are a really good addition. (Much easier than rewiring the steppers.) And now that you have those jumpers, you have the option to simplify the stepper header traces. Specifically where OUT1B and OUT1A cross to go to stepper header pins 3 and 4. If you de-cross that wiring, it will flip the stepper direction. Then the direction jumpers will flip it back. 
• I don't think it matters in the slightest to the function, but the OUT1 and OUT2 phases are reversed against header pin numbers 1,2 and 3,4 compared to the Rev E Mightyboard. You have OUT2=1,2 whereas MBI has OUT2=3,4. Again, doesn't hurt anything, but I figured I'd mention it in case you're picky about consistency. Yeah, this is kind of the opposite of the last bullet, I know.

On Sunday, August 3, 2014 1:36:22 AM UTC-5, Gary Crowell wrote:

Looking really interesting.

Here's the schematic for the XYZ-ABCD converter board.  

• I've got a breakout board for the MIghtyboard Botstep sockets - it uses a 10-pin header/cable to bring the signals to the converter board (I've got a bunch of 10-pin cables here somewhere).
• I added step select jumpers for AB, and CD.
• There are jumpers to reverse any ABCD direction.
• Sockets to mount the ABCD Botsteps.
• Mightyboard-type Molex 0.156x4 headers for motor connection.
• On-board 5V generated from 24V.

I have the board routed except for a few impossible routes and some cleanup.  Not too late to make changes though, if you have any other ideas.

Gary

On Sat, Aug 2, 2014 at 10:02 PM, Ryan Carlyle <temp...@gmail.com> wrote:

Motor arrangement, tensioning, and tramming sorted:

(the idlers on the 2020 rails near the motors will be "loosen, slide, tighten" to adjust line tension)

--
You received this message because you are subscribed to the Google Groups "3DP Ideas" group.

To unsubscribe from this group and stop receiving emails from it, send an email to 3dp-ideas+unsubscribe@googlegroups.com.

To post to this group, send email to 3dp-...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/3dp-ideas/e52a141a-8d77-41ab-a684-2eaf013edbf2%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[Ryan Carlyle]

On the mechanical side -- I'm designing it for: 

• Misumi 2020 frame 340x340x580 to match my existing CoreXY lasercut parts, but size is very flexible. Main thing is the corner bracket style. The Y-end alignment and pulley size match Misumi HFS5 brackets since I'm using them to hold the Z pulleys. It could be modified for another bracket style or for a printed corner bracket without too much trouble.
  
• 12mm linear rods on Z, mounted with VXB shaft end flanges. Wouldn't be too hard to redesign for supported rod or extrusion slide. 
• 8mm linear rods on X and Y
• These V-groove pulleys from Robotdigg: (need 52 = 9 sets) http://robotdigg.com/product/72/Fishing-Line-Pulley-Bearing
• Pretty much all M4 hardware

I'll provide Sketchup files and BOM/PNs for anyone who wants them. Depending on the motion parts you end up wanting to use, I can make alternate versions of printed parts. It's just some modelling work to change pocket sizes for bearings and stuff.

[Ryan Carlyle]

Lots of parts are coming directly from my CoreXY Cube build. Find the BOM for that here: http://www.thingiverse.com/thing:393155

• Extruder carriage in its entirety
• All the electronics
• Lasercut panels to hold electronics

[Gary Crowell]

Current layout for the ABCD converter.  100% routed, even passed DRC for clearance.  I'll still be tweaking for awhile to shorten traces and increase spaces where possible.  Trying to get a better grid on the 5V/Gnd but I think it's about as good as it's gonna get.  Layout may look odd in some respects 'cause it's done for a milled board.  There's no thru vias except where a component pin or added wire goes thru the hole (and that pin must be soldered both sides).  You can't put a topside trace to a component pin where the body of the component covers the pad (connectors, headers, etc.) 'cause you can't solder the topside on those.  The body of the screwterm stepper output overlaps the molex connectors, so that's an either-or case.

--

You received this message because you are subscribed to the Google Groups "3DP Ideas" group.
To unsubscribe from this group and stop receiving emails from it, send an email to 3dp-ideas+...@googlegroups.com.
To post to this group, send email to 3dp-...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/3dp-ideas/bdc6e7a0-2a5a-48b0-86e0-8f67ad2cc09a%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[Vernon Barry]

Gary, I'm glad youa re doing this as a routed board, FYI James Armstrong who is also here in Charleston also has a PCB mill and can cut these for me as well.

 

Sorry, that I've been quiet on this thread, just trying to get a few other major projects done in the very limited spare time that I have free right now.

To view this discussion on the web visit https://groups.google.com/d/msgid/3dp-ideas/CAB%2B-abu4mh%3DB31%3DLah3ucnETmJ_yA9hvK%2BSZkxspoTs_cn61dw%40mail.gmail.com.

[Ryan Carlyle]

You going to build one too, Vernon? 

I would be knee-deep building mine already if I didn't like my CoreXY too much to strip it for parts. As it is, I have a lot of stuff to source.

On Monday, August 4, 2014 8:09:13 AM UTC-5, Vernon Barry wrote:

Gary, I'm glad youa re doing this as a routed board, FYI James Armstrong who is also here in Charleston also has a PCB mill and can cut these for me as well.

 

Sorry, that I've been quiet on this thread, just trying to get a few other major projects done in the very limited spare time that I have free right now.

On Sun, Aug 3, 2014 at 6:12 PM, Gary Crowell <garyacr...@gmail.com> wrote:

Current layout for the ABCD converter.  100% routed, even passed DRC for clearance.  I'll still be tweaking for awhile to shorten traces and increase spaces where possible.  Trying to get a better grid on the 5V/Gnd but I think it's about as good as it's gonna get.  Layout may look odd in some respects 'cause it's done for a milled board.  There's no thru vias except where a component pin or added wire goes thru the hole (and that pin must be soldered both sides).  You can't put a topside trace to a component pin where the body of the component covers the pad (connectors, headers, etc.) 'cause you can't solder the topside on those.  The body of the screwterm stepper output overlaps the molex connectors, so that's an either-or case.

On Sun, Aug 3, 2014 at 2:24 PM, Ryan Carlyle <temp...@gmail.com> wrote:

Lots of parts are coming directly from my CoreXY Cube build. Find the BOM for that here: http://www.thingiverse.com/thing:393155

• Extruder carriage in its entirety
• All the electronics
• Lasercut panels to hold electronics

--
You received this message because you are subscribed to the Google Groups "3DP Ideas" group.

To unsubscribe from this group and stop receiving emails from it, send an email to 3dp-ideas+unsubscribe@googlegroups.com.

To post to this group, send email to 3dp-...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/3dp-ideas/bdc6e7a0-2a5a-48b0-86e0-8f67ad2cc09a%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--

----------------------------------------------
Gary A. Crowell Sr., P.E., CID+

Linkedin  Elance  KE7FIZ  Things RocketryCNC

--
You received this message because you are subscribed to the Google Groups "3DP Ideas" group.

To unsubscribe from this group and stop receiving emails from it, send an email to 3dp-ideas+unsubscribe@googlegroups.com.

To post to this group, send email to 3dp-...@googlegroups.com.

[Vernon Barry]

Maybe one day after Maker Faire.

 

I have only a few weekends left to get the big machine up, running, tuned, and cleaned up for September.

To unsubscribe from this group and stop receiving emails from it, send an email to 3dp-ideas+...@googlegroups.com.

To post to this group, send email to 3dp-...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/3dp-ideas/07ef43dd-567c-48e1-aabf-63f43445aa0e%40googlegroups.com.

[Chris P]

Ryan, your idler pulleys and motor drive pulleys strongly imply you're going cable drive on this one...  Two reasons I can think of are a) cost vs open ended belts and b) easier to route cable drive through the system, do plane changes, etc, vs a belt.  Did I miss any?

[Ryan Carlyle]

It's mostly the non-planar turns. You'd have to put a 90 degree twist in the belts between the Y ends and X ends.

Plus yeah, spectra drive is WAY cheaper with this many idlers.

[Ryan Carlyle]

Gary,

What sort of stepper headers are you intending for this board? I see two rows of pads for different pitches?

I was thinking of using these, which I believe will match stock Mightyboard RevE:

http://www.digikey.com/product-search/en?KeyWords=WM50000-24-ND

http://www.digikey.com/product-search/en?KeyWords=WM2336-ND

On Sunday, August 3, 2014 5:12:45 PM UTC-5, Gary Crowell wrote:

Current layout for the ABCD converter.  100% routed, even passed DRC for clearance.  I'll still be tweaking for awhile to shorten traces and increase spaces where possible.  Trying to get a better grid on the 5V/Gnd but I think it's about as good as it's gonna get.  Layout may look odd in some respects 'cause it's done for a milled board.  There's no thru vias except where a component pin or added wire goes thru the hole (and that pin must be soldered both sides).  You can't put a topside trace to a component pin where the body of the component covers the pad (connectors, headers, etc.) 'cause you can't solder the topside on those.  The body of the screwterm stepper output overlaps the molex connectors, so that's an either-or case.

On Sun, Aug 3, 2014 at 2:24 PM, Ryan Carlyle <temp...@gmail.com> wrote:

Lots of parts are coming directly from my CoreXY Cube build. Find the BOM for that here: http://www.thingiverse.com/thing:393155

• Extruder carriage in its entirety
• All the electronics
• Lasercut panels to hold electronics

--
You received this message because you are subscribed to the Google Groups "3DP Ideas" group.

To unsubscribe from this group and stop receiving emails from it, send an email to 3dp-ideas+unsubscribe@googlegroups.com.

To post to this group, send email to 3dp-...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/3dp-ideas/bdc6e7a0-2a5a-48b0-86e0-8f67ad2cc09a%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[Ryan Carlyle]

Actually, now that I'm thinking about it, I'd like to go ahead and order everything I need to get this working. All the rest of the necessary CoreXYZ parts are on their way now. Are we at the "release a BOM and fab some boards" stage?

[Ryan Carlyle]

Build update: most of the parts are in. I have:
Frame kit
Electronics (FF mightyboard etc, PSU)
Linear rods and bearings
Spectra line
Stepper line pulleys
Most of the spectra idlers
Steppers
Most of the printed parts (all the gantry components)
Extruder & hot end
Tons of M4 hardware

On the way / in progress:
Lasercut structure plates
More spectra idlers
HBP
Gary's kick-ass stepper multiplexing board
Some more printed parts (stepper brackets)

Hoping to get the frame and 3-axis gantry put together this weekend.

[Ryan Carlyle]

Latest render. Changed the motor arrangement and line path around to make mounting the HBP and electronics easier. It's not super pretty, but it's the least-bad routing I've come up with (of like five different options).

[Dan Newman]

On 14/08/2014, 5:39 PM, Ryan Carlyle wrote:
>
>
> Latest render. Changed the motor arrangement and line path around to make
> mounting the HBP and electronics easier. It's not super pretty, but it's
> the least-bad routing I've come up with (of like five different options).

Well, three points do define a plane. But you need to add leveling feet to those
three stepper motors ;) (Also, I want to see rainbows shooting out of the extruder
and maybe a my little pony on the build plate. You know, to appeal to the bronies.)

Dan

[Ryan Carlyle]

Ask and ye shall receive.

Friendship is magic!

[Dan Newman]

On 14/08/2014, 6:58 PM, Ryan Carlyle wrote:
> Ask and ye shall receive.
>

> <https://lh4.googleusercontent.com/-TKa6tsHEKR0/U-1pA7onyAI/AAAAAAAAAic/lMFmJ1TPlik/s1600/Capture.PNG>
> Friendship is magic!

Now that's a serious printer. Sign me up!
(P.S. we're they printed without support?)

Dan

[Ryan Carlyle]

Rainbows have excellent bridging performance. No supports needed.

[Ryan Carlyle]

On a more serious note:

[Christopher Matthes]

Looking awesome, please keep up posted ;) 

May I ask where you got the metal stepper fishing line pullys from ? 

[Ryan Carlyle]

The line pulleys / spools / drums came from a Delta parts site:

http://www.tridprinting.com/Mechanical-Parts/#3D-Printer-Spool

To be honest, they're way too large diameter. In the current design, the XY steps/mm is ~57 and the Z steps/mm is ~171. So it should run really fast, with lower precision. That's perfectly fine for a test build but not really where I want to be for a production bot. 

Compare to 88.88 & 400 for the Replicator 2 line (with 18t GT2 pulleys and 8mm pitch lead screw). 

I may print some smaller spools eventually, but the aluminum ones are just so shiny and pretty!

[Chris P]

Printing spools is tricky, because any runout on the diameter will impact the motion/position of the print head.

[Ryan Carlyle]

Chris, I agree. Printing something that requires spot-on diameter is very tricky. That's why I'm starting with aluminum, even though the diameter is 50% larger than I want. How are you printing the spools for your machine? 

One thought I had is printing slightly over-sized, rotary-sanding it down to spec.

This wouldn't be a problem if I had a lathe. Although the spools I bought have a wire-fastening hole through the body so they really can't be turned down far enough. It'd mean starting over from scratch. And since all four steppers have to run in sync plus precisely control the Z stage, they need to have extremely accurate and matching diameters.

[Ryan Carlyle]

On the build progress front, I had a brain fart and only ordered 4x 12mm shaft flanges. (Got another 4x on the way now.) So I was only able to build half the gantry yesterday. But that half is working beautifully. See pics below.

(The Misumi bearings I ordered came dripping in oil, so I have some paper towels tied around the rods to catch the drainage.)

I should probably mention that I have no idea where I'm going to put the Z and Y endstops. Still working out details like that.

[Joseph Chiu]

Wow, that is gorgeous.  

--

You received this message because you are subscribed to the Google Groups "3DP Ideas" group.

To unsubscribe from this group and stop receiving emails from it, send an email to 3dp-ideas+...@googlegroups.com.

To post to this group, send email to 3dp-...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/3dp-ideas/5d2282d1-8057-44a6-8745-c90b7f1b5e91%40googlegroups.com.

[Ryan Carlyle]

I printed out some temporary 12mm rod flanges and "finished" the gantry. It's smooth and sexy. Will post pics when I get a chance. (I can't wait for all my 624vv bearings to come in so I can string it all up on the spectra line.)

BUT, it's occurred to me that I have no idea how I'm going to run the wiring and filament feed tube. I built this thing for a 400mm tall build volume, and with the gantry moving up and down, that means I either need a ~800mm long overhead U-loop (ulgh) or some kind of clever routing solution that attaches to the moving gantry itself. Need some more thought on that one.

[Ryan Carlyle]

More build pics.

This was the worst possible way to build the gantry, but the final two 12mm rod flanges were still printing. I should have build the two separate Y stages, then lifted one up to  twist the gantry and insert the X rods. 

Special thanks to Dan (and his father) for the aluminum carriage plate.

Gantry finished:

Extruder assembly. E3D v6 hot block, QU-BD nozzle and cooling bar, MBI stepper, Ebay extruder, random 3mm thick aluminum "nut" that I cut and tapped from a sheet of stock because I couldn't find the "magic brass nut" I was planning on using.

I wanted to make sure everything went together ok before insulating it. I'll do that later.

[James Armstrong]

Can't wait to see how it all works. I just picked up my Dan's special alum carriage from Jetguy (Vernon) Saturday. It is so much better than the acrylic one we originally did. Thanks Dan. 

[Dan Newman]

On 18/08/2014, 5:35 PM, James Armstrong wrote:
> Can't wait to see how it all works. I just picked up my Dan's special alum
> carriage from Jetguy (Vernon) Saturday. It is so much better than the
> acrylic one we originally did. Thanks Dan.

You're quite welcome. There's actually a newer edition which my father did.
Only 2 mm thick Al plate so as to accomodate the thin jam nut trick. And it
is double slotted so you can do the old Mk7 mounting method. One of thoe
plates is used in this picture,

https://www.flickr.com/photos/d-newman/14628705904/in/set-72157645313526624/lightbox/

Dan

[Ryan Carlyle]

More build pics.

"MK3" build plate with heat traces put on aluminum instead of FR4:

Ponoko parts:

I ended up jigsawing off some material from the floor plate to make room for bearings and rod flanges, and redoing all the mounting holes:

First power up, everything works EXCEPT the LCD. Haven't had time to troubleshoot yet.

[Ryan Carlyle]

LCD issue turned out to be flipped wires in my jumper harness. This is why you don't A) work late when exhausted and B) use a bunch of single-wire jumpers instead of a proper wiring harness.

So the electronics all work, sans stepper multiplexing board. The last mechanical parts came in. I'm now at the "how on earth do I make this stuff fit together" stage. My XYZ gantry horribly interferes with the lasercut panels for the LCD screen, buttons, and power switch. (This is why you should put the electronics in the 3d model.) Plus motor mounts and spectra lines are EVERYWHERE.  So I'm printing some standoffs and hacking things together right now. 

I'm absolutely going to make CoreXYZ  Rev 1 work, but Rev 2 is going to involve some significant redesign.

[Christopher Matthes]

Stumbled across this : http://forums.reprap.org/read.php?2,377858,page=1 thought you might find it interesting ;) It starts with a coreXZ and in the end they think about building the bot you already designed.

How are you getting on with your corexyz ? Any news ? 

[Gary Crowell]

Very interesting discussion there, but I don't see anything that impacts or rivals Ryan's concept.  I do want to keep an eye on it though.

I've ordered all the parts for the ABCD converter, they should be here early next week.  I'll have a final board milled out by then too.  Then we'll see if I remember how to solder.

Gary

On Fri, Aug 29, 2014 at 11:21 AM, Christopher Matthes <christoph...@googlemail.com> wrote:

Stumbled across this : http://forums.reprap.org/read.php?2,377858,page=1 thought you might find it interesting ;) It starts with a coreXZ and in the end they think about building the bot you already designed.

How are you getting on with your corexyz ? Any news ? 

--

You received this message because you are subscribed to the Google Groups "3DP Ideas" group.
To unsubscribe from this group and stop receiving emails from it, send an email to 3dp-ideas+...@googlegroups.com.
To post to this group, send email to 3dp-...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/3dp-ideas/61c45ce6-8f50-4f2f-8bf5-0d68d797b36e%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[Ryan Carlyle]

I saw that CoreXZ thread last month, a few days after writing my post in this thread on July 29th. I was already toying with the 3:1 Z reduction pullies before I read it, but I'll freely say that seeing Nicholas Seward do it on his CoreXZ made up my mind to do the same for my CoreXYZ build. 

The guy who SAID he was making a CoreXYZ in that thread was just going to hang a separate Y stage off the CoreXZ carriage, which is a shitty idea. (Same issues as Cobblebot.) 

I am making good build progress. Sorting out endstops and wiring this weekend. Basically strapping additional plastic brackets to the existing gantry components. Will post pics in a bit.

[Ryan Carlyle]

Draft Zmin endstop design. Uses standard MBI-style endstop switch. Bolts onto a Y-end with existing hardware. Has adjustable screw to set build plate gap (not rendered):

Putting together the MK3 aluminum heatbed. Aluminum heatbed = GIANT HEATSINK = horrible to solder on leads. DEAR LORD this took me an hour and a 1200 watt heat-gun to preheat the whole plate hot enough to get solder to stick with a big-ass 120 watt soldering gun. I should've used a torch.

Hacking an endstop switch board to act as a HBP thermistor pull-up board:

Thermistor attachment:

Prepping the wood floor plate. I'm a bit terrified of this MK3 aluminum heatbed shorting from a PCB heat trace to energize the entire plate, so I went ahead and covered all the electronics-mounting hardware with Kapton. A layer of aluminum tape went over that to reflect heat back at the heatbed.

Stuff going together nicely:

Extruder and HBP both work like champs. I'm working on endstops now. (They are surprisingly tricky to fit in this design.) After that, I need to design the filament feed / wiring harness path to the extruder. I think it's going to jump off one of the Y-ends to reduce the length of the overhead U-loop. Haven't decided yet.

I really want to rig up the steppers and spectra lines, but my final 12mm rod flanges got lost in the mail and won't arrive until Tuesday. $%#&ing UPS is useless sometimes. No point in running the Spectra line before I have the final rod brackets/alignment done.

[Gary Crowell]

So here is the final cut of the board.  Came out well; should be perfectly usable.

But in the meantime, I recall that I haven't done anything more on the BotStep breakout boards.  They're laid out, but I still have to cut them.  And, I just realized that I didn't order the connectors for them.  I might look around here, I know I have some; but if they don't turn up soon, I'll make another DK order.

On Fri, Aug 29, 2014 at 9:50 PM, Ryan Carlyle <temp...@gmail.com> wrote:

Draft Zmin endstop design. Uses standard MBI-style endstop switch. Bolts onto a Y-end with existing hardware. Has adjustable screw to set build plate gap (not rendered):

Putting together the MK3 aluminum heatbed. Aluminum heatbed = GIANT HEATSINK = horrible to solder on leads, dear lord this took me an hour and a 1200 watt heat-gun to preheat the whole plate hot enough to get solder to stick with a big-ass 120 watt soldering gun. I should'be used a torch.

Hacking an endstop switch board to act as a HBP thermistor pull-up board:

Thermistor attachment:

Prepping the wood floor plate. I'm a bit terrified of this MK3 aluminum heatbed shorting from a PCB heat trace to energize the entire plate, so I went ahead and covered all the electronics-mounting hardware with Kapton. A layer of aluminum tape when over that to reflect heat back at the heatbed.

Stuff going together nicely:

Extruder and HBP both work like champs. I'm working on endstops now. (They are surprisingly tricky to fit in this design.) After that, I need to design the filament feed / wiring harness path to the extruder. I think it's going to jump off one of the Y-ends to reduce the length of the overhead U-loop. Haven't decided yet.

I really want to rig up the steppers and spectra lines, but my final 12mm rod flanges got lost in the mail and won't arrive until Tuesday. $%#&ing UPS is useless sometimes. No point in running the Spectra line before I have the final rod brackets/alignment done.

--

You received this message because you are subscribed to the Google Groups "3DP Ideas" group.
To unsubscribe from this group and stop receiving emails from it, send an email to 3dp-ideas+...@googlegroups.com.
To post to this group, send email to 3dp-...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/3dp-ideas/63d16e95-29c8-4595-8b81-fa9d7887236d%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[Joseph Chiu]

Wow, Gary, that's nicely milled!  What kind of a setup do you have?  

I had a mini CNC mill (currently mothballed) that I tried to do this with -- but I kept breaking bits and I eventually gave up...

To view this discussion on the web visit https://groups.google.com/d/msgid/3dp-ideas/CAB%2B-abuMh-uv%2B-QzaHPUHHzZ-kj5RoSCf8AVz9XobpgS%3DZCXfw%40mail.gmail.com.

[Ryan Carlyle]

Yeah, sexy board. I'm kind of glad I don't have room for it under the bot -- that thing needs to be visible :-)

[Christopher Matthes]

There is nothing really which rivals Ryans design. Just thought you might find it interesting to read what other people are going in a similar direction.

Looking awesome Ryan, can not wait to see running the first time :)

[Ryan Carlyle]

First rev of the Z endstop mechanism:

Apparently I don't have any 35mm long M4 screws. Thus the 50mm long screw and extra nut. (I probably should have checked that before I designed it specifically for M4x35mm.) I also don't have any endstop switches. Those are on the way. 

I'm a little worried about the gantry falling when it loses power, and crashing the nozzle into the build plate. That's one reason why the Zmin endstop is as stout as it is. (It's also why I went ahead with spring-loaded build plate support posts even though I don't plan on leveling with them.) Depending on the detect friction in the four steppers, that endstop might be taking a pretty hefty impact when the bot loses power.

It's crossed my mind to add a pulley and counterweight system like Vernon mentioned for his Big Bot, so the gantry doesn't fall. But that would mean roughly doubling the moving mass of the Z stage. Not excited to do that.

[Gary Crowell]

Oh, and I break bits too.

[scott.e...@gmail.com]

Ryan,

Looks like this might be a nice application for an inexpensive and low-mass constant force (reel) spring.

[Ryan Carlyle]

Yeah, I reckon CF springs would be pretty good here. That's a thought.

I'm going to see how violently it drops once the spectra lines are rigged up to the steppers. Detent friction from four Kysans and a 3:1 pulley reduction may be enough to suspend the weight. 

Which reminds me, I need to do some torque/friction testing on these spectra pulleys. I need to figure out how many wraps are required for good holding force without going crazy on it. Also need to figure out whether the default Mightyboard Vref is going to drive these steppers with enough current or not. 

[scott.e...@gmail.com]

On smaller systems it is not uncommon to use a collection of metal tape measures to simulate and analyze the effectiveness of constant force springs.

Just a thought.

(Peel the Stanley label off and you go from prototype to production). ;)

[Ryan Carlyle]

According to a cheap luggage scale, my gantry weighs 1.55kg / 3.4lbs right now. I think I'll go ahead and order four of these in case I end up wanting them: http://www.mcmaster.com/#9293k18/

All four would be ~3.7 lbs, for a slight upwards net force. Which sounds perfect to me. 

I need to think a little more about backlash mechanisms with this design. On a lead screw, you want positive Z-stage weight to avoid backlash. But I can't think of any source of backlash in this design at all, unless the spectra lines are loose.

[Gary Crowell]

On Sat, Aug 30, 2014 at 2:45 PM, Ryan Carlyle <temp...@gmail.com> wrote:

Yeah, I reckon CF springs would be pretty good here. That's a thought.

Which reminds me, I need to do some torque/friction testing on these spectra pulleys. I need to figure out how many wraps are required for good holding force without going crazy on it. Also need to figure out whether the default Mightyboard Vref is going to drive these steppers with enough current or not. 

Would 8825's do?  I have layouts for those, and I could do a board that integrates the drivers and the conversion.

If I'd looked out my front door earlier I'd have noticed that the DK parts were here.  I'd forgotten how fast DK was.  Everything seems to fit pretty well.  I may take a trip to work tonight to assemble it.

[Gary Crowell]

Ryan, you wanted the 0.156 Molex headers on that, right?

[Ryan Carlyle]

I have a bunch of spare MBI-style drivers (with 4988s) that take an external Vref, and I have a bunch of spare Pololu Black Edition A4988s with onboard trim pots. I was planning on using one of those. Hadn't decided which option yet. I won't mind if you do something different. 

My FF Mightyboard Vref is outputting an actual 1.79+/-0.02 V across the five drivers. (Nominal is 1.91v for the as-built 7.5kohm pull-up and 5kohm digipot.) That means 0.83A with the stock FF drivers. I've heard the stock XY steppers are rated to 0.84A but have not been able to confirm this. Nominal current output at digipot[118] is 0.89A and the max for the digipot +/-20% tolerance rating is 0.99A, which kind of explains why people on the FF board keep complaining about their steppers running hot. (I mean, they should be hot, but they shouldn't be super hot.) This was all a GIANT PAIN to sort out, because the voltage divider circuits do not match the Mightyboard RevE schematics.

[Sidenote: Fuck you, clone companies, for making a bunch of minor modifications to OSHW designs and not releasing updated schematics. I should not have to take a magnifying glass to the board and follow PCB traces to figure out what you built.]

Then there's the chopper off time. MBI-style botsteps short ROSC to ground which means 30us off-time and forced mixed-decay mode. This can make a difference on low-current rising microsteps (eg the first step after a 0 current crossing for the coil). Whereas the Black Editions have a 10kohm pull-down resistor so, according to the datasheet, 12us off time and automatic mixed-decay (use mixed on falling current, use slow-decay on rising current). 

On one hand, faster = quieter, because the chopping frequency goes higher pitch and perceptual volume decreases rapidly over 8khz or so. So the fast off-time on the Blacks will cause less audible noise, but if it's not long enough to adequately decrease the motor current before the next PWM on-cycle, will cause skipped microsteps due to current being too high at certain parts of the microstep curve. So potentially the Black Editions could cause microstep position registration issues.

I'm not competent enough to calculate whether 12ms is enough off-time to hit the 1/16th microstep current. (First order decay equation?) Anyone feel like doing the math? I can't find a minimum on-time for the A4988 so I'm not sure if we even have enough info to calculate it.

Swapping to 8825s but using the same digipot vref and sense resistors (0.27ohm) will provide 1.6x more current (1.33A). Which would be pretty good for 1.5A rated Kysans. But if the stepper multiplex board takes Vref off a separate trim pot, then I suppose it doesn't really matter and I would just set the current by hand. I haven't looked at the on-time/off-time limits on the 8825s -- that would merit checking. But I do like the idea of being able to try 1/32 microstepping.

And yes, I am using 0.156 headers. This is the plug I put on all my Kysans: http://www.digikey.com/product-search/en?x=0&y=0&lang=en&site=us&keywords=wm2336-nd  I have break-away pin headers I can solder up if you leave them unpopulated. But I won't mind if you feel like doing the soldering  :-)

On Saturday, August 30, 2014 4:23:44 PM UTC-5, Gary Crowell wrote:

[Ryan Carlyle]

Extruder cable management parts are printing now. The gist is that the gantry will have a traditional overhead U-loop, but the loop "launches" off one of the Y-ends so the Z travel doesn't cause issues. It might be overly-complicated, but I think it's going to work. Plus, cable chains are cool. 

(No, the chain will not hit the stepper spool below it at the Zmin position. I checked that.)

I probably could have merged the blue and yellow parts into one piece, but this way made it easier to design for printability (wrt 45 degree rule and bridging). Plus maintenance is easier if the wiring run + cable chain can be pulled without a lot of extra stuff attached. 

From a mechanical engineering standpoint, this is a TERRIBLE design. The fastening system is lopsided and the forces are all unbalanced. It's going to have a lot of unwanted flex in use. But for extremely light loading like this, convenience in attaching stuff together trumps good design practices.

[Ryan Carlyle]

Just did some stepper testing, plugging one of my Kysans straight into the Mightyboard and hanging weight off the spool. The stepper can lift >2kg at low speeds. Which is "hella plenty" since the gantry weighs <2kg, and will be supported by 4 steppers, through a 3:1 pulley reduction. So... 12x safety factor at low speeds/acceleration means I should be good to quite pretty fast.

Another datapoint, it takes 5 wraps around the spool to generate enough capstan friction (with 325g back-tension) for the stepper to slip before the line/spool does. Which means the spectra-on-aluminum friction is pretty low. But I really need a good pull-scale or a set of weights or something to take full fiction/slip measurements. I might rig up something with a water jug so I can calculate the weight from water volume.

I'm a bit concerned about the spectra line walking up and down the spool and crossing itself during long Z moves where the spool is turning a ton due to the 3:1 reduction. I do have a workable solution for that but it will add another, oh, twenty? 624vv bearings to the system. Not my first preference.

[Ryan Carlyle]

Stepper / spool testing:

Installing brackets and steppers:

[Ryan Carlyle]

The cable chain parts finished printing after I got the steppers mounted. Basic chain from Thingiverse (http://www.thingiverse.com/thing:34661 scaled up 33%), end brackets custom. I ran out of white PLA and everybody I like is out of stock, so I'm using white ABS. Probably a good thing since the cable chain was ridiculously difficult to assemble and I would likely have broken in PLA. 

[Admittedly not my best photography there. White ABS on off-white MDF, under 2700K CFLs. Taken with an iPhone no less. I should really be using my real camera and EyeFi card for this. Meh.]

Y-end mount turned out about how it was supposed to, but you might notice my R2x is printing with some X/Y backlash. Visible in the first layer on the chain-end bracket. But I'm too lazy to chase that down while I'm just making a bunch of mechanical parts.

Range of motion turned out perfect. Engineering is easy when you have the entire thing modeled in high fidelity.

I feel like this printer is getting uglier as I kludge together all these miscellaneous bolt-on parts. I did not factor endstops or wiring into the original 3-axis gantry design. Aesthetics are necessarily low on the design priority list as I try to jam all this last-minute functionality into the gantry. Oh well, it's a prototype. It's supposed to look rough!

[Ryan Carlyle]

Last update for the night. Got the X endstop sorted. (Note: on this build the gantry X and Y are flipped from the normal Replicator orientation. Long story.) So that's all three endstop mounts taken care of. Just need some misc M4 hardware and FF endstop boards to come in this week, and I can do all the wiring runs. 

The endstop wiring will run through the cable chain along with all the carriage wires.

[Chris P]

Ryan are you wrapping your drive lines around the pulleys like kossel does, or like a sailboat winch?

[Dan Newman]

On 31/08/2014, 10:04 PM, Ryan Carlyle wrote:
> Last update for the night. Got the X endstop sorted. (Note: on this build
> the gantry X and Y are flipped from the normal Replicator orientation. Long
> story.)

Flipped which way? Will you be using X min and Y min homing? If so,
then standard builds will not support P-stop since standard builds
use X min for it.

Dan

[Ryan Carlyle]

My plan was sailboat winch (simple capstan, ~5 wraps) with a centralization mechanism to limit wrap walking, but the kossel style may be better. That's the kind of pulleys I bought anyway.

Problem is fitting ~1600mm of wraps onto the spool for full travel including the 3:1 reduction pulleys on the tall Z axis. I think the spool is long enough to fit that much line if it packs efficiently, but the changing fleet angle may cause stacked wraps and that will throw off the steps/mm.

What's your bot use? Would you be willing to share a close-up pic of your CoreXY pulleys?

[Ryan Carlyle]

Flipped, as in the gantry bridge provides the Y axis. So the gantry is mechanically rotated relative to a normal CoreXY bot. Homing is to the "back right" of the bot relative to the screen, just like a Replicator. But the Y endstop is on the carriage and the X endstop is on the Z stage.

I decided this over the weekend, so the endstops would match the normal orientations. Yes, this is in conflict with my parts naming convention. But the LCD screen is rotated 90 degrees wrt the normal gantry to fit all the parts in, so it kind of works out. Bot cardinal directions are normal relative to the LCD.

[Ryan Carlyle]

Did it this way because I liked the way stuff fit together better. Easier access to the nozzle for leveling, easy build plate removal window, etc.

[Chris P]

I use kossel style.  Even with 5 wraps winch-style I was getting slip between the cable and the pulley.  I was really surprised to see this but it was undeniably there.  The other problem I had was keeping a winch-style wrap centered, because the walking forces are proportional to your cable tension and end up being extremely hard to overcome.

[Gary Crowell]

Ok, last week I cut out the ABCD converter board, and I populated it over the weekend.  I had a bit of a nightmare tracking down shorts, including one on the 5V bus that needed six or seven trace cuts to track down.  In the end I just ran a knife tip along a short bit of trace and around a pad and the short went away.  I never saw a thing, even under a very good microscope.  There were a few other simple shorts, but I think I've got everything now.  Checked continuity on all traces, and fixed a few misplaced components.  I've applied power to it and there was no visible smoke; regulator is giving a good 5V.  I'm hoping to be able to put a function generator  and scope on it tonight.  

I still have to cut the botstep breakout boards but those are simple; can maybe do those tonight too.   There will be a 10 pin ribbon cable from each breakout board to the ABCD board, though they really only carry step, dir, and ref;  I have 10 pin cables laying around.  Ryan, I have 4" and 9" cables.  Which do you prefer, or will the 9" even be long enough?   In a pinch, they could be pretty easily just jumped with wires point-to-point.

Gary

--

You received this message because you are subscribed to the Google Groups "3DP Ideas" group.
To unsubscribe from this group and stop receiving emails from it, send an email to 3dp-ideas+...@googlegroups.com.
To post to this group, send email to 3dp-...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/3dp-ideas/85567e44-110c-4cb7-a174-a36aed07d5ae%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[Joseph Chiu]

Something I learned from a guy at the NSL hackerspace about milled PCB's -- apparently, there can be small amounts of leftover copper that will bridge between traces.  They had a RF board that was failing because of it.  They finally solved it by etching the milled board with a quick swipe of etchant -- the really small fragments dissolve right away, while the traces were left intact.

To view this discussion on the web visit https://groups.google.com/d/msgid/3dp-ideas/CAB%2B-abs0AGq6R0fUD5P8bcN-%2BeOwbq_OxVehq_1BKvFp7_np6Q%40mail.gmail.com.

[scott.e...@gmail.com]

Hah!

After I read Gary's post I was wondering quietly to myself if something like that would work......  

So for the next 3 minutes I will not feel stupid.  ;)

[Ryan Carlyle]

Gary, I'll probably have to put my hands on the boards to sort out how to fit them onto the bot. The whole build is going to need reorganizing eventually but the priority now is function, not looks.

I'm guessing for now I'll just jigsaw/drill some plywood for a base, and hang it off the side of the printer with standoffs for the stepper boards. So ribbon length is not too important as long as the boards can fit side by side. I do have a bunch of point-to-point jumpers if needed.

On Thursday, September 4, 2014 2:58:34 PM UTC-5, Gary Crowell wrote:

Ok, last week I cut out the ABCD converter board, and I populated it over the weekend.  I had a bit of a nightmare tracking down shorts, including one on the 5V bus that needed six or seven trace cuts to track down.  In the end I just ran a knife tip along a short bit of trace and around a pad and the short went away.  I never saw a thing, even under a very good microscope.  There were a few other simple shorts, but I think I've got everything now.  Checked continuity on all traces, and fixed a few misplaced components.  I've applied power to it and there was no visible smoke; regulator is giving a good 5V.  I'm hoping to be able to put a function generator  and scope on it tonight.  

I still have to cut the botstep breakout boards but those are simple; can maybe do those tonight too.   There will be a 10 pin ribbon cable from each breakout board to the ABCD board, though they really only carry step, dir, and ref;  I have 10 pin cables laying around.  Ryan, I have 4" and 9" cables.  Which do you prefer, or will the 9" even be long enough?   In a pinch, they could be pretty easily just jumped with wires point-to-point.

Gary

On Thu, Sep 4, 2014 at 8:51 AM, Chris P <chris.pad...@gmail.com> wrote:

I use kossel style.  Even with 5 wraps winch-style I was getting slip between the cable and the pulley.  I was really surprised to see this but it was undeniably there.  The other problem I had was keeping a winch-style wrap centered, because the walking forces are proportional to your cable tension and end up being extremely hard to overcome.

On Monday, September 1, 2014 7:22:05 PM UTC-4, Ryan Carlyle wrote:

My plan was sailboat winch (simple capstan, ~5 wraps) with a centralization mechanism to limit wrap walking, but the kossel style may be better. That's the kind of pulleys I bought anyway.

Problem is fitting ~1600mm of wraps onto the spool for full travel including the 3:1 reduction pulleys on the tall Z axis. I think the spool is long enough to fit that much line if it packs efficiently, but the changing fleet angle may cause stacked wraps and that will throw off the steps/mm.

What's your bot use? Would you be willing to share a close-up pic of your CoreXY pulleys?

--
You received this message because you are subscribed to the Google Groups "3DP Ideas" group.

To unsubscribe from this group and stop receiving emails from it, send an email to 3dp-ideas+unsubscribe@googlegroups.com.

To post to this group, send email to 3dp-...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/3dp-ideas/85567e44-110c-4cb7-a174-a36aed07d5ae%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[Ryan Carlyle]

Chris, I've got some tricks up my sleeve on the capstan walking problem. I regularly work with giant (>70,000 lbs) spooler units for 10,000 ft long subsea electrohydraulic umbilicals. And high-tension wireline pulling systems, and constant-tension motion-compensating systems, and 2.5 million lbs capacity oil rig derrick blocks. Managing motion of wires and hoses is a surprisingly large part of my job. What's hard for me about this is just miniaturizing these giant pieces of equipment to 3d printer scale using cheap parts.

This is roughly what I had in mind for the spectra drive:

Except with one driven spool and a stack of idlers. With one spool being smooth, you need the idlers to take up all the wrap shifting. (Otherwise it will still walk -- not forever like a plain spool capstan drive, but just back and forth over the distance between the idlers when reversing direction. It's a very minor problem but enough to put some motion error in the system.) 

Preventing walking entirely would normally be accomplished with offset-axis idlers... not sure exactly how to explain that without a picture. Ok, found one in a vendor's sales brochure:

(I would guess this is a ~100,000 lbs tension rated pulling unit, but that's a SWAG.)

The driven drum has a horizontal axis like you would expect for an overboarding winch. The idler sheaves are on a skewed axis such that the top of each idler is across from one wrap on the main drum, and the bottom of each idler is across from the adjacent wrap. So the wire makes a long spiral around the driven drum and idlers with no walking, and no side-loading the sheave/drum flanges. Nice and tidy. Just takes some engineering for the proper angles, and high diameter/thickness ratio idler sheaves. But all my idler pulleys are low diameter/thickness ratio so I don't think this style will work without getting different parts.

Here's a jury-rigged little non-walking arrangement I tried a few days ago. It only gets three half-wraps of contact so it's not going to work, but it had no line rubbing and no line walking.

The Kossel-style drive is new for me because we never have more than one line take-off from a drum -- but it's pretty straightforward and I'll probably start with that. Just need to rearrange some angles to the nearest idler pulleys near the spools to ensure efficient wrap packing.

[Gary Crowell]

Did some testing on the board last night, and all seemed well.  The step one-shots are good, but I know that the way I did other testing did not exercise all of the logic.  More testing tonight.

To unsubscribe from this group and stop receiving emails from it, send an email to 3dp-ideas+...@googlegroups.com.

To post to this group, send email to 3dp-...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/3dp-ideas/c9e6e54a-5d67-49f9-82c4-80936e3f8d5f%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[Ryan Carlyle]

Awesome, Gary. I don't think I would have even attempted to build this without your help on the electronics. 

To view this discussion on the web visit https://groups.google.com/d/msgid/3dp-ideas/c9e6e54a-5d67-49f9-82c4-80936e3f8d5f%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[Christopher Matthes]

May I ask how you secure the fishing line to the pulley ? Is there some kinda scrub screw or is i just by smartly winding the fishing line around it ?

Do you use these holes ? Does the green marked hole go vertical through  the pulley ? Im asking because I will probably have the opportunity to make some  custom pulleys out of steel next week. As there is a huge variety of them :

or:

or: http://www.thingiverse.com/thing:186988 

I wanted to ask you for your preferences as I would love to help. What kind of pulley design do you prefer or would you change something? Do you want a 10mm diameter or 11, 12 ? Would you rather have a horizontal or vertical hole to stick the fishing line through it ? 

[Chris P]

Hehe I tried something similar a while back.  I think if you size it to run everything on bearings, it may work fine but I didn't have the space to do so, and the BOM cost rose astronomically.

http://forums.reprap.org/read.php?2,151871,186357#msg-186357

[Ryan Carlyle]

Yeah Chris (P), there's nothing truly new under the sun. I've seen a lot of different 3DP spool drives too. Although reading through that RepRap forum thread -- I feel really bad for all the people who throw tons of metal at HBot gantries and still end up with quality issues. GANTRY FORCES MUST BE BALANCED! HBot is simply an inferior design concept. 

Chris (M), for now I'm doing a Kossel style pulley. Not positive it will work with this much line travel, but the idea is that you run line from each end of the spool, and one end is winding up while the other end is paying out. The holes through the spool just let you do a little loop from one side to the other.

Once you get the filament looped through the spool, you string up one side of the carriage in the farthest-most position for that line. Then "wind up" the spool to pull the carriage to the diametrically-opposite position, and run the other side of the line on the opposite side of the gantry. It's kind of a pain.

To make this work properly and not cross the wraps (which throws the steps/mm off and potentially tangles the line) you need to be careful about fleeting angles. The nearest idler to each side of the spool must be roughly level with that end of the spool, and sufficiently far away that as wraps accumulate, the fleet angle is never large enough to cause them to stack.

On the issue of what the ideal spool would look like -- all I know is that it should have a theoretical diameter of 40/pi=12.7325mm. (Maybe more like 12.4mm or 12.5mm to account for line thickness. I'm not sure what the effective pitch diameter is of this spectra line yet.) You really want the steps/mm value to be as close as possible to 80 for 1/16th microstepping. That's the sweet spot for Sailfish performance and quality. "Integer circumference" is really, really helpful in avoiding rounding error on the positioning system. It's not so important on XY gantries but it's critical if you want to use spectra line for the Z stage and avoid rounding-error banding.

I'm stringing up my gantry tonight. I'll take pictures when I get it done. Giant PITA to run 32 feet of spectra line through a hojillion idlers...

[Ryan Carlyle]

Hmm. I made an attempt to string it up tonight. The line routing around the electronics is kind of nightmarish. I made some new little idler brackets to move some stuff around, but in doing so I lost my main line tensioning mechanism. The motor bracket slots and adjustable carriage clips aren't anywhere near enough adjustment range for this.

I need to rethink my carriage clips entirely. I'm thinking about something that can take up a lot of slack line, like a guitar tuning peg. Needs more engineering...

[Christopher Matthes]

I guess thats the only downside of fishing line( thats it is  more complicated to string then belts). 

Would you rather go with the theoretical value of 12,73mm or with 12,5mm ?

Another question. How do you check your assembly with sketch up ? Isn't that rather hard to do ? 

[Ryan Carlyle]

Chris (M), what do you mean by "check"? Make sure the STLs are manifold, or verify the as-built against the model, or what? For STLs, I use the Solids Inspector and CleanUp3 plugins.

For verifying the as-built against engineering drawings... I don't! The beauty of modeling the entire printer to scale is that all the parts come out pretty much exactly how they looked on the screen. At worst I do a fit-check, adjust a dimension or two, and reprint. The areas where I'm having trouble are the places that I didn't model rigorously.

Fishing line has some unique challenges, but belts would also be a pain with this design. It's just a lot of loops and routing. Belts would be easier, but not easy. I think what I need to do is rig up some sort of "spectra needle" to fish line through the pulleys easier. Maybe tape the end to a piece of nylon filament or something.

I guess if I were ordering custom pulleys to experiment with, I would do 12.5mm as a scientific guess. But the "correct" value depends on the size of spectra line used -- when it is under tension. My line is nominally "0.5" but the actual wrap pitch is about 0.3mm side-by-side and it's probably around 0.2mm "deep" on the spool. I haven't taken any rigorous measurements yet. But the effective diameter in mm should be [40/pi - thickness]. (Technically 2*thickness/2 but the 2s obviously cancel.)

[Dan Newman]

> I think what I need to do is rig up some sort of "spectra needle" to fish line through the pulleys easier.
> Maybe tape the end to a piece of nylon filament or something.

Our see if the local Target/CVS/whatever has floss threaders which are stiff
enough. E.g., BridgeAid Dental Floss Threaders.

Dan

[Ryan Carlyle]

I'm not sure they'll be long enough for the Y-ends. Worth a shot though.

[Christopher Matthes]

With checking your assembly I was wondering if there is a plugin which allows you to actuate the parts. To see if there is for example enough space.

Looking forward to see a guitar peg on a 3d printer :D Wait a second it needs to be a lower chord ;)

[Ryan Carlyle]

Sketchup does have a half-assed "animate" function, but no, there's no real actuation design functionality. (Sketchup is really architectural software.) When necessary, like when designing the cable chain, I just manually rotate/translate parts to verify the range of motion. Sketchup is far from the optimal design program but it's intuitive and other people can easily/freely modify my source files. 

Here's a few pictures from "attempt #1" to string up the gantry. I wasn't going for full range of motion, just getting all the lines in place to see how they run.

And now it's time to put my inventin' cap on, and channel Tony for a while...

[Ryan Carlyle]

Sending to print.

[Ryan Carlyle]

Ran into some pretty wicked clashing issues, because apparently I do not plan ahead enough. The "rev 1 spectra tuner" managed to run into both the X and Y endstops. So here's rev 2 with big chunks taken out of it:

For the record, I hate this tensioning solution, but I think it'll work.

On Sunday, September 7, 2014 5:19:37 PM UTC-5, Ryan Carlyle wrote:

Sending to print.