LDO v2.4 / 350 ; mod log, more than a build log.

[YaaJ]

Been scratching my head on this problem : how to have an horizontal drag chain for the umbilical ??? No examples on the web. Every topology failed. Until some surgery was made on a 7x7 drag chain :

 

Hopefully it will work...

 

NOPE ! FAIL ! Always gets stuck somewhere

[EDIT] found this : we don't see the what the chain does when the toolhead is at x=0 y=Ymax ; weird... Where does the "excess" length go ??? It is hidden.

 

Edited November 29, 2023 by YaaJ

[mvdveer]

5 hours ago, YaaJ said:

how to have an horizontal drag chain for the umbilical

AS far as I remember, the original Tronxy X5SA had a horizontal drag chain for an umbilicus. I never liked it though - the chain was extremely stiff. My original  Tronxy is now a VZBot Found a picture on the net of one to illustrate.

  

Found this on thingiverse: Maybe clarify some?

Only way I see a vertical (Umbilical style) chain it will work is to have a swivel point at the rear connection?

Like this design on Cults?

 

[YaaJ]

All these designs have a characteristic in common : all of them have a footprint larger than the enclosure. Found two more videos : this is always hidden, like in this video :

Currently designing two parts, one for each end of the cable chain, with pivot. Testing by hand is not a good idea.

The source of inspiration :

 

Interesting : uses a USB cable, signal and power.

Another interesting mod was made by "Old Man Melts Plastic" : he routed the vertical drag chain another way, so it is not in the way ; it is hanging under the rear extrusion, instead of going up then down. Didn't think of doing this, and it could be part of the solution : it was a real problem while trying different positions. Why is he doing this ? Is he working on a toolhead dragchain, in secret ?

 

[YaaJ]

After a quick and dirty setup / test fit :

Front left :

Front right :

Rear left :

Rear right :

Rear right has a problem : the cable chain touching the belt when the head goes to the Y endstop. Due to the GX12 connector. This connector is for the toolhead (heater + temp probe). Have to see if it could be relocated, or mounted upside down. [EDIT] of course, it can be relocated. Will probably be fine horizontally, just between the extruder and the pivot. doing this, the cable chain will not touch the belt (at least, it will not apply any force aginst it)

Also, wanted to test USB (noCAN) comms between the Octopus and the EBB36 : a USB connector is too large for the drag chain, and it is closed type (AFAIK, open 7x7 drag chains don't exist. Also have open 10x11, but too large / heavy. Will require CAN.

Wire torsion : 270° amplitude ; manageable, for example by spiralizing the 4 wires between the cable chain out and the terminal.

Will not say the problem is solved, but at least something can be tested.

Also implies re-routing everything. The Z drag chain hanging under the gantry will make routing easier and shorter in the electrical compartment (drawer), but the cables will be on the left side under the gantry. New designs for the A and B drives, as well as Y endstop relocation...

Back to the drawing board...

[EDIT] salvaged the cables from the LDO kit. Currently twisting the CAN cables. Should be 50 twists / meter. Not easy. Twist and shout. Or cry.

Edited November 29, 2023 by YaaJ

[YaaJ]

Massive progress today. Decent pivot on the toolhead (these parts are quick and dirty designs, *not* the final parts !!!).

Front right (view from rear)

Back, right (holding the chain because there's no panel...)

Rear left ; crappy test part, printed as fast as possible on the v 0.2, with no calibration at all (thick layers, few walls, top bottom  etc... massive underextrusion). Does not matter. It is likely this part will be made of aluminium : it's tall and too flexible (to be decided after a part is properly printed, and the A drive frame redesigned for better fixtures)

At this point, there's no pivot. Next design.

This is a huge improvement. The pivot now rotates 90-100° on the head. This is 3 times better than the first test (270°). And no more need for the GX12 connector relocation.

To do this, I had to do the same mod than "old guy melts plastic" in order to make room for the horizontal umbilical cable chain and avoid collisions. The Z chain is hanging under the flying gantry, and is much shorter now. It has a cost : the bed had to be raised by 6 mm, due to the chain minimum radius. Will be 10 mm, i.e. 20mm instead of 9 (LDO spacers are 9mm thick)

[EDIT] was wondering where one of the spacers was gone ! It's between the Z motor and the belt !

Doing this, the copper length is divided by a factor two or so, and the cables jump directly from one chain to the next one, down to the electronics in the drawer ! As a result, I can reuse the Microfit crimped wires from the kit, despite the drawer ! And save a lot of spares. Winner winner !

The DIY twisted pair is settling, under tension, at 100 twists/meter. Will be reduced later to 40 to 50, according to the recommandations. Funny to see the twist (very) slowly become more and more regularly spreaded along the cable, hours after hours. Was a mess at the very begining.

 

Edited December 1, 2023 by YaaJ

[Penatr8tor]

@YaaJ Excellent solution. The Pivot is key!

[YaaJ]

Oh yes ! The pivots do all the work. Found the ideal positioning on the toolhead and the gantry, so the drag chain is just as long as needed, and does not rub against the Z belts.

The new problem is designing the pivot. Of course, it could be a screw, but the threads probably will eat the chain ends very quickly. Salvaged 3mm diam rods from paper printers. Yet another pin mod ! 3mm also is the BMG idler gear axle diameter... Could even rotate on needle bearings, with a idler as outer race, pressed into plastic (the way the nozzle Z probe is designed).

Searching things among a pile of old parts, found spares for older E3D Titan. Could make a Gucci duesie pivoting system (pin = inner race, gear = outer race, and two bearings in between ; it is the exact required dimensions !) ; but any added millimeter counts, changing the ideal pivoting positionings :

Cannot buy the Bambulab kit I just discoverred :

https://eu.store.bambulab.com/products/cable-chain-assembly?variant=43949265223899

It is cheap ; would require two of them (one would be too short) ; it is the pricing of a 7x7 on Amazon, it is open type, comes with a pivot. Would get two...

But Bambulab parts have nothing to do in a Voron.

Also found this (beautiful engineering)  :

https://www.printables.com/fr/model/385282-bambu-lab-p1p-printable-cable-drag-chain

https://www.printables.com/fr/model/364801-bambu-lab-p1p-cable-drag-chain-assembly-59-updated/files

[EDIT] will be tested soon :

(and the parts are BMG parts, not Titan)

 

Edited December 2, 2023 by YaaJ

[YaaJ]

More progress.

For now, I used screws (not pins) because it's easier. Really promising. Promising to a point USB C comms are an option again.

There will be no need for machined parts. Copied the idea from "[a]_xy_joint_cable_bridge". Just perfect. Rigid enough. (pics don't show the final design, the parts are there for test fitting of course)

And USB C between the Octopus and the EBB36 could be an option... I used M3 screws as inner races for the bearings. It is just for testing ! Because it's easier !

The 7x7 drag chain came closed. Opened it, just to see what happens. Looks like a viable solution. Tried a few cuts. Diagonal ones probably are the best. Will try to cut in the middle, and reshaping with the hotair desoldering station (smallest nozzle, and up to 500°C or so, it's supposed to be made of Nylon)

Less than one year back, I thought the Voron Zero was a toy, made for speed boating... How wrong was I ! It's been the workhorse since it's been completed.

Looking for something between the Zero and the 2.4/350 ; Micron ?

Edited December 2, 2023 by YaaJ

[atrushing]

Looks like some real fun problem solving!

24 minutes ago, YaaJ said:

Looking for something between the Zero and the 2.4/350 ; Micron ?

You could always try the Vorpal 180! I am still working on the manual but it's coming along..

[YaaJ]

3 hours ago, atrushing said:

You could always try the Vorpal 180! I am still working on the manual but it's coming along..

The thread that made me think about a medium size printer.

[YaaJ]

Because of a drawer major flaw (weak brackets, thanks to eSun ABS+...), had to redesign and reprint everything from scratch ! Lost one full week.

Now back to the umbilical.

The final toolhead cardridge connector :

And the final umbilical pivot (could be slightly modified but keeping this geometry) :

The gantry side. This was just for positioning. This arm will be given a needdle bearing (from another cheap BMG kit), and embed a custom breakout PCB : USB planned, rather than CANBUS (because 1-why not ? 2-because simplified wiring if adding a nozzle cam + USB Beacon : USB hub). The pinned B drive frame was redesigned (some symmetry with the A one), but it is not the definitive design. The arm will be replaced with a secondary electronics enclosure that could receive everything needed, from a breakout board to a miciocontroller - there's room for anything)

[EDIT] the Z belt is touching the frame because the gantry was not horizontal !

More and more dead corpses in the grave. Collected all of them since the project was started. This project could be renamed genocide.

[EDIT] forgot to say...

Among the beginer mistakes and fails this week was made of... Replaced the PEI bed steel sheet for a new one on the bed slinger. And completely forgot I had removed the auto Z offset device a few weeks ago... Guess what ? Understood under the shower, 2 days later, why the nozzle was crashing into the bed. Nice one, isn't it ?

Edited December 10, 2023 by YaaJ

[YaaJ]

Got the final geometry ! Nowhere the cable chain will be rubbing against belts or panels. The part that holds the pivot on the gantry will be modified, as it will have to enclose a USB breakout board ; if I can't get USB to work properly, a twisted pair was made and routed from the EBB36 to a U2C. The U2C is in the electronics drawer. Relocating it on the gantry  maybe is a better option (much shorter CANH/CANL line, receiving much less EMI from the stepper cables.

Currently stuck, Waiting for the low profile USB cables and USB breakout boards...

 

A Biqu Microprobe also is on the way. The EBB36 is mounted reversed, and no room for the stupid Dupont connector. Replaced it for a JST, mounted on the "copper side" ; space is tight, the Y endstop will have be moved a couple millimeters (it is on the left side)

Mounting the EBB reversed has some advantages ; cleaner cable management, slicker toolhead, and better cooling for the TMC2209, the PCB groundplane being responsible for most of the heat dissipation (according to datasheet). It will be tested with no additional heatsink. And if a heatsink is required, it will be free air, just over the thermal vias (the golden oval pad).

But for now, thid printer is far from printing...

Edited December 13, 2023 by YaaJ

[YaaJ]

Did some testing. Flashed the EBB36 with a Klipper firmware. Took some time before it was recognized. Until I unticked "bootloader"... Definitely can be used as a secondary extender.

But is it interesting ? Definitely not ! It just adds complexity. At first, it was about avoiding electromagnetic interferences in the electronics drawer cable chains as well as software and hardware layers. Half a meter side by side with Z1, Z2, Z3, Z4, A and B motors cables, then A and B, taking advantage of the USB shielded cable. In the end it appears to be only complexity. The USB C connector cannot go through the 7x7 dragchain, meaning some hacking, and AWG 30 or 32 terminals crimping (reaching the limits).

A middle of the road compromise : a Picobilical-like solution : a relocated U2C. The drawer was designed with this device, side by side with the Octopus ; also considered the Octopu as a bridge. Then learnt about the CANBus miserable bandwidth. Decided it will be on the gantry, connected to the Pi with a USB cable, and with minimal twisted pair length to the toolhead.

But le U2C is a bit too bulky. It has to be downsized a little bit.

Removed everything that's not needed : the huge and crappy screw terminal, made of skinny sheetmetal, the unknown 4pin white connector, and the alien dual CAN over USB dual connector.

Replaced the screw terminal for a smaller and much better quality one. Using rosin based "no clean" flux... No clean ? Seriously ? It had to be cleaned ! With IPA of course.

Long story short : Raspi -> gantry = USB, gantry -> toolhead = CANBUS. A compromise...

Back to the drawing board.

[EDIT] had a bad idea. Listened to old Sigue Sigue Sputnik and Toy Dolls songs. Can't get them off my head...

 

Edited December 14, 2023 by YaaJ

[YaaJ]

Some progress :

The white printed part is just a placeholder ; the will be an enclosure ; just have to find a angled USB-C -> USB A 150cm cable. Doesn't have to be high end, the STM32 datasheet states USB is 12Mbps.
Still didn't cut and crimp the CANBUS wires : measure twice, cut once.

Nowhere the wires and the cable chain will be touching the belts.
 

The toolhead should be operational before the end of this week, and extrude some plastic for testing..

Edited December 19, 2023 by YaaJ

[YaaJ]

CANBUS operational, no errors.
Removed the confusing include file with aliases, and edited printer.cfg by hand. Everything flashed from the Pi, as it is much more straightworward and easy.
Everything tested good. The extruder extrudes, the probe probes, etc.

Except the LEDS with the Barf Mod... (the rainbow LED was installed a while back)

Of course only one mistake could be done... Did it... And had to take everything apart.

In through the out door. Fixed.

(the wavy look is moiré)

The gantry will be belted soon.

The U2C being on the gantry, there should be no EMI problems with the motor cables in the long cable chains (DIY twisted pair, unshielded, made from the LDO wires)

[Austin Dennis]

Wow! I'm super glad to have happened upon this post. I also just had the idea to convert my umbilical into a horizontal drag chain. My regular umbilical hits the top of the enclosure and gets into weird spots that makes me hesitant to print higher than like 170mm. I don't want to go back to XY chains because that balloons the length of the CAN wire by almost 2x.

I will be drawing much inspiration from this post, but I have a couple questions:

• Do you think I could reuse my original XY drag chains here to make the horizontal chain or is there a specific reason you went with 7x7? 
• Did you just use your existing Z chain in the under gantry position or is that a different type from the Z chain spec?

Super cool stuff! I'd be happy to share my work here for a G2E and a more classic gantry mount.

[YaaJ]

I went to the 7x7 because it is much lighter. I wouldn't use a 10x10 or 10x11. The best is probably the drag chain sold by BambuLabs (open type) ; Creality also have such a drag chain on one of their printers. The BambuLabs horizontal drag chain is dirt cheap, but I'm a Bambu hater (closed source, etc.).

This umbilical will not solve the problem with the reverse Bowden ; dont how it will be managed ; the BambuLabs routing is interesting.

The Z drag chain is mounted under the gantry, because it would interfere with the "umbilical" drag chain. It is a 10x10, I will maybe replace it for a larger one (10x15 being the maximum) ; if the curvature radius allows for it. Because of this radius, the Z drag chain removes some travel : the bed has to be raised a bit, the same order as a kinematic bed, no big deal (there will be kinematic mounts at some point, have to design and machine them).

Still untested with the head moving around at high speeds ; just by hand ; currently wiring the motors. Explored many slightly different designs ; the next iteration will have a USB hub inside the relocated U2C enclosure, in order to add a nozzle cam and / or a USB cam attached to the gantry instead of the frame (I'd like to get the same point of view as a conventional moving bed CoreXY).

A bit stuck for now... I thought I had all the required wirings, but no ! The missing parts should be there on thursday AWG18 24V power cable (currently using AWG15 !!!), low profile Ethernet and USB cables, USB hubs (modules used in PCs as internal mini hubs). You know, all these little things that are missing when it's time to complete a project !

This being said, I should be able to do some realistic umbilical testing today or tomorrow (motors moving the head at high speed, on long distances). The printer is belted, and belts are tuned.

 

[EDIT] no luck ! bad contact on the CANBUS connector, hotend side ; but when electricity finds its way, the toolhead is moving around (i.e. when Klipper doesn't shutdown because of lost connection with the EBB36).

Edited December 25, 2023 by YaaJ

[Austin Dennis]

Nice, thanks for the reply and hope you’re having a happy holidays! I’m also a bit of a Bamboo Lab hater so I’ll see of getting my hands of a couple Creality chains is doable.

Not having wire is the most relatable thing I’ve ever heard. I’m sure I’ll have to redo my CAN harness to be PTFE as well so that will be a ton of fun.

Looking forward to that maiden high speed test, good luck!

Edited December 25, 2023 by Austin Dennis

[YaaJ]

@Austin Dennis a few pics ; this is the current moutings (prototypes, bulky, quickly printed) :

Top ; notice the massive, useless 24V wires ! The USB cable also is too thick ; waiting for parts...

Bottom : the cable chain has to be above the bed extrusions of course (rear view) :

10x15mm drag chain, min radius = 50mm OD, 25mm ID

Edited December 25, 2023 by YaaJ

[Austin Dennis]

Oh, ya those pictures are very helpful!

Those parts shouldn’t be an issue for me to draft up in CAD. That chain is definitely overstuffed and USB wires are thick as hell.

I wonder if you could save some space in there by making a data only USB cable (for 2-4 less conductors). I’m not sure if the 24V is regulated and connected to all the UTC electronics or just passed through to CAN devices.

if course you could ditch the Z chain all together and go for a loose umbilical setup there. I’ve seen some people run it with good success. It just looks a bit odd.

Edited December 26, 2023 by Austin Dennis

[YaaJ]

I had to replace the 10x15 Z drag chain with a 10x10, because it was interfering with my EBB36 mod (the relocated BLTouch pinheader on the back) ; it shouldn't be a problem with a conventional setup.

Still no high speed test... But tested the resonances. With both the EBB36 accelerometer, and the conventional one, attached to the tool cardridge. The input shaper recommandations remain the same ; these calculations and recommandations are based on the sum of all accelerations, meaning that the accelerometer can be in any position ; but the power spectral density will not be the same ; 45°, got more or less a 0.7 ratio, near to sin(45°)... The rotated accelerometer giving largest value (x+y+z)

The resonance tests, with the cardridge (standard) accelerometer ; seems ok, similar to others found on the web... Frequencies, PSD, and IS recomandations

Please tell me if you see a problem !

@Austin Dennis The two Z drag chains I've been testing (10x10 and 10x15) will *not* require risers under the bed. They both have the radius shown in the previous post. Got them on Amazon. If you do this mod, beware the minimum radius, it is not always specified by the vendors. First test was with another drag chain that did not fit well.

Edited December 28, 2023 by YaaJ

[Austin Dennis]

I’ll keep that in mind while browsing on Amazon, thanks for the advice!

I don’t see any problems with your input shaper graphs. Recommend accels look great for a 350 and spikes look tight. Well done!

I use an input shaper workflow module that gives me a ton of data and I think it could help you during the build to spot any potential mechanical issues before they get hard to fix if you’re interested. I use the belt tension graph to dial in my belts that last little bit so they are exactly the same tension.

https://github.com/Frix-x/klippain-shaketune

[YaaJ]

I knew Klipain only about the adaptive bed leveling. Didn't notice there's information about accel curves interpretation. Will have a look ! (started with Klipper less than one year ago). Just watched one video on YT.

About accelerometers, I asked a question on Klipper forums, and uploaded the curves by the EBB36 accelerometer as a comparison. Had a very interesting answer by Sineos, and a link to an abandonned pull request :

https://klipper.discourse.group/t/ebb36-accelerometer-orientation/12639/1

https://github.com/Klipper3d/klipper/pull/4931

 

[Austin Dennis]

Ya, he recently moved the input shaper stuff to a standalone repo (which was totally the right call). It’s still pretty new. I think it’s brilliant since I’m not very savvy with python myself and just being able to run a few install commands to get it up and running really appeals to me. I’m still learning my way around Klipper as well, it’s been about 10 months for me since I first brought my Voron 2 online.

Strictly speaking, all of what the PR is asking for is already possible with python scripts and macros (and I think the klippain module has an auto axes mapping macro already)so I’m not sure that’s ever going to be a priority for mainline Klipper. Would be nice to have, but I think they’re focused on scaling up to a 64bit architecture sustainably for the foreseeable future. They have some cache management challenges to come through.

Edited December 28, 2023 by Austin Dennis

[YaaJ]

Sorry, I didn't see your answer !
Will test all this when the printer can print. (will not be until some time...)

Had to do minor, but extremely time consuming mdifications. The drag chain for the drawer low voltage wiring was too small. Had to order a 15x15. Of course, the 15x15 is a bit thicker than the 10x15. By 2mm. Had to redesign the brackets and retest, etc. Very little room between the drawer and the screws that hold the deck under the bed.

Pretty happy with the result :

Yes, I know. The power lines on the Octopus are daisy chained ; but it is 1.5mm², and can be considered as a bus bar.

The printable heat shrinks make the wirings look very professional. Really a game changer. Well worth the money.

Put a blank on the 4th motor JST terminal, after I spent 2 hours wondering why the head was moving everywhere except where I wanted it to go...

Had 4 5015 fans lying around. Made 4 brackets, they will be used as bed fans. Orientations differ, hopefully they will generate some swirl in the chamber, and help for temperature repartition. The Nevermore will suck the hot air between the bed extrusions.

Didn't cut the wires too short ! But there will be some cable management to do...

Also, made some "bed indexers". Was tired with the spacers never where they have to be, and falling on the electronics.

Not sure they can wistand the temperature, but they are easy to machine with just a 16mm endmill and a 8mm drillbit. And there's a kinematic bed in the todo list...

(to be continued, problem while inserting images...)