Bertha's Build Blog

[claudermilk]

Oh yeah, the wiring error jumps out at me. But I'm not suffering from tired brain after a long build session. That kind of thing is where I'd put it away for the night and come back the next day--and probably quickly go "Oh! I'm a dummy! "

Nice progress and congrats on getting it powered up and to the tuning stage.

As for the connectors, one day when there's maintenance to do and removing that part simply means unplugging it rather than tearing the wire harness apart it will be a blessing. FYI, my XY endstops are the PCB setup with the connector and I'm happy they are that way. Especially after having had to rewire my entire gantry following a failed wire. That's now on a tool head PCB with connectors--next time will be much easier.

[xyleth]

So today's 'work' (in amongst actual work) has been on the belts, mainly the A/B belts.  Tensioning is a pain!

I tried the 110hz tuning method and just can't get a consistent read on my phone, so instead I used this which I bought in anticipation of this very problem:

The instructions say a 9mm Gates belt should read about 6.3 (so the Z belts) and 6mm about 5.3 (A/B Belts), so that's what I aimed for.  I then used Klippan's Shake 'n Tune module to tune the belts.  Many iterations later this is the best I can do:

 

And that is after completely removing the A/B belts and ensuring that they are, as far as I can possibly make them, exactly the same length to the tooth and then de-racking everything again.  From there it is on to Input Shaping and here are my results:

I'm not sure what's causing that second peak on the X axis, but have been advised on Discord that it is probably not serious enough to spend time chasing now.  Let the machine bed in for couple hundred hours and heat cycles and everything will loosen and move around anyway - then you can tune.

Probably won't be pushing plastic today, but it's close.

 

[xyleth]

This is the last of the Shake 'n Tune outputs.  Seems I should avoid around 70-80mm/s speeds.

 

[xyleth]

Okay I fibbed.

First plastic!

Just done calibrating the extruder rotation distance.  Next is on to first layer squish and all that good stuff, but that can wait for another day.

[xyleth]

First print!

The only tuning that has been done at this point is getting the Z offset right, other than that it is pretty much all defaults.  Default Voron PLA profile in Orca slicer using the generic PLA that came as the tester spool with the printer.  53 minutes later, including 8 minutes of bed soak, here is the end result:

Not bad for a first attempt I think.

[zav3nd]

Only gets better from there. You  might have 3 more cubes and a bunch of tuning prints by now though!

I'm guessing the purge line is from KAMP?

[xyleth]

18 minutes ago, zav3nd said:

Only gets better from there. You  might have 3 more cubes and a bunch of tuning prints by now though!

I'm guessing the purge line is from KAMP?

Not KAMP, I haven’t installed that. It’s out of the box with Klippain which implements many of the same behaviours. 

Edited April 7 by xyleth
Fixing autocorrect

[xyleth]

Looking a lot more printer like now!

 

I tried fitting the included bowden tube from the exhaust vent plate (self printed on this printer!) to the toolhead.  It went....poorly:

It went from printing near perfectly to an under-extruded nightmare simply by adding that tube.  I figure that it's probably cheap tat so I've ordered length of Capricorn PTFE tubing to replace it.  In the dim recesses of memory I recall having a similar problem with my Ender 5 and fixing it with quality tubing so hopefully this will resolve it.

The tube also pulled the retaining clip out of the coupler that came with the Galileo 2 extruder, so I've had to ditch that part.  After exploring a few options I ended up opting for epoxying a brass coupler with a compression fitting into the mount on top of the extruder.  It's a proper 2 part epoxy so hopefully it'll hold up to the rigours of printing - worst case I'm going to have print / source another front plate for the Galileo.  At least I know intimately how it all fits together.

The printer is running most of the time I'm in the office with it now, currently printing filament latches to hold the panels on.  I should be in a position to think about printing some ABS by mid-week.  Job number 1 after getting it dialled in - reprint the Jalouise for the Nevermore v6.  Here's what became of the one I was shipped:

It went in fine after being warmed on a build plate as suggested, but it disintegrated the first time I tried to pull it back out

[claudermilk]

Don't use the Capricorn! Trust me, I tried way back when and it caused more issues. I just this inexpensive PTFE (cloudy white) tube that's been working fine. ID is 3mm, OD is 4mm. You want that tube to fit a bit loose so the filament can move through it with as little drag as possible. The little piece between the extruder and hot end can be the 2mm ID stuff, and I used some of the Capricorn for that for a while.

[xyleth]

2 hours ago, claudermilk said:

Don't use the Capricorn! Trust me, I tried way back when and it caused more issues. I just this inexpensive PTFE (cloudy white) tube that's been working fine. ID is 3mm, OD is 4mm. You want that tube to fit a bit loose so the filament can move through it with as little drag as possible. The little piece between the extruder and hot end can be the 2mm ID stuff, and I used some of the Capricorn for that for a while.

Ooo, thanks for the heads up.  I can return the Capricorn to amazon I guess!  Makes sense that you'd want a loose fit, maybe that's the problem with what Siboor supplied.

I ordered some transparent 3mm ID / 4mm OD PTFE tubing instead.  Figure it'll be good to be able to see the filament moving down if I ever implement a MMU.

[MrSprinklz]

2 hours ago, xyleth said:

Ooo, thanks for the heads up.  I can return the Capricorn to amazon I guess!  Makes sense that you'd want a loose fit, maybe that's the problem with what Siboor supplied.

I ordered some transparent 3mm ID / 4mm OD PTFE tubing instead.  Figure it'll be good to be able to see the filament moving down if I ever implement a MMU.

Capricorn is good/fine for any bowden between your extruder and hotend which is why its good for enders and such as well as the short length in the toolhead. It's just bad for reverse bowden

Edited April 9 by MrSprinklz

[dasbooter]

@xyleth for parts for the pin mod and beefy idlers are those parts typically printed yourself or purchased? If say I wanted to use the print it forward program should I forgo certain parts and maybe ask the printer as a separate transaction to produce those parts? Is that at all reasonable? Sorry for the novice question but seems like a chance to not waste time money and plastic.

[claudermilk]

As far as I know PIF prints standard parts, and just what  you need to get the printer built. From there it's up to you. So either build it stock with PIF parts, print the mods and add the mod yourself (requires unbuilding a bunch of the printer), or talk to your PIF provider and see if they are willing to swap out parts for you; from there that's between the two of you. It can't hurt to ask, the worst case is they say no.

[xyleth]

On 4/11/2024 at 5:17 AM, dasbooter said:

@xyleth for parts for the pin mod and beefy idlers are those parts typically printed yourself or purchased? If say I wanted to use the print it forward program should I forgo certain parts and maybe ask the printer as a separate transaction to produce those parts? Is that at all reasonable? Sorry for the novice question but seems like a chance to not waste time money and plastic.

I purchased the parts for the Beefy Idlers and the Pin mods from a small 3d printing company in my country who specialises in printing Voron parts.  I found them via Etsy but ended up dealing direct because I wanted a lot of non-standard stuff.  This is how I ended up with my SB parts printed in PC-CF instead of standard ABS.

I didn't use the PIF program so I can't comment on how that works.  But I would say you can always ask - the worst that will happen is they'll say no and you'll have to look elsewhere.

[xyleth]

So, it's been a while between updates.  This is both because work has gotten busy and I have entered the inevitable 'every time I touch it something breaks' phase of the project.

Everything was clearly going too well as I developed two, possibly related, problems.  First the G2E extruder developed a horrible habit of unscrewing itself from its motor mounts during longer prints:

My working theory is that this was caused by the strain relief mount that mounted on the back of the motor mount bolts.  The suspicion is that the torque motion applied to that mount was slowly turning the screws.  This is not helped by the fact that I was not fully tightening the motor mount bolts because if I did so the whole extruder seized up, so I'd found that I had to back them off by about a turn in order for the extruder to run.

In addition to this I started getting intermittent comms failures to the toolhead.  This is _probably_ down to a issue in the CAN umbilical but as it was an intermittent problem it is very hard to troubleshoot.

So two problems present themselves, I need a new umbilical mount and I'm not 100% sure I can trust my umbilical cable.

On the mount side I found this mount in the user mods section of the G2E repo.  It mounts into the cable chain mount and only one of the motor bolts, so hopefully should prevent any torque transference issues.  So off to the old printer that file goes to print in PETG.  Once printed it becomes clear that although it fits perfectly the BTT CAN umbilical connector doesn't work with this mount.  It's simply too long and not flexible enough.  So I thought I'd see if I could remove the housing and make it fit.  Bad idea:

Well at least I now know the umbilical is no good

Off to the shops!  Once again onetwo3d came to the rescue with a Linneo umbilical.  At the same time I took the opportunity to grab a replacement Stealthburner LED harness as I'm pretty sure the one I got was either DOA or I killed it during installation, and a USB 2 CAN board just in case because it's been suggested that the Octopus Pro USB-CAN bridge can sometimes have problems like this. 

The Linneo umbilical has the nice feature of a CNLinko connector at the A Drive mount which makes the umbilical much easier to remove for maintenance.  So it's not all bad!

After a lot of testing on the bench I have reached the conclusion that some spacers were needed between the G2E gearbox housing and the printed parts.  In the end 2xM3 washers per side seem to be the sweet spot - I tried 1 as in the photo below but while that worked reliably on the bench as soon as I put it on the toolhead it went right back up to binding again.

Re-assembling everything this _seems_ to have resolved the issue.  The printer is certainly humming along now running a 3hr test print that has failed previously so we will see.

I also took advantage of the downtime to fit a chamber thermistor on the Z-chain guide.

On the plus side I finally have enough latches to fit all the enclosure panels and have made up the clicky-clacky door mod so the enclosure is complete!  If this PLA print finishes okay I'll put the lid on and start experimenting with the ABS

[xyleth]

1hr 40m in and still going strong.  About half way now.

On the topic of the ClickyClacky door, mine was the LDO kit with Linneo's NanoNest door panel.  Instead of a single pane of acrylic this comprises of 3 pieces of 2mm clear PC.  Two of them are square window parts and one is a thin spacer that holds them separate.  This, in theory, produces better insulation and sound dampening.  Although as the other sides of the printer are just single panes of acrylic I doubt it'll have much impact in practice.

There are two notable side effects from this.  First is that the 6mm stack of panels is a tight fit into the extrusion and no longer require the rubber weather fitting strip that comes with the LDO kit.  This is mentioned in the description of the NanoNest panels, but I obviously missed this and spent a frustrating 20/30 minutes swearing at the weather strip when it wouldn't fit.  Don't be like me, read the description fully!

The second impact is that the two layers of PC are not as clear as a single layer of acrylic would be.  You can still see through, but everything is slightly foggy / out of focus.

[xyleth]

Successful print!  Seems things are working again 

[xyleth]

Big milestone today - the printer printed its first major own part in ABS:

added to Keep the steppers cool while its running. 

Edited April 15 by xyleth

[xyleth]

Having got ABS printing going I went ahead and printed some more in-chamber things, including this nozzle scrubber and purge bucket:

This installed nicely like so:

However it also introduced a new wrinkle, as is the way of things.  The scrubber is at just the right height to catch the Cartographer probe when the head is printing right up in that top corner of the plate which was causing layer shifts when moving from the 1st to second layer.  Took me half a dozen failed prints and about half a day to work that one out!

In the end my solution, because I want to keep the scrubber, was to make use of Orca slicers ability to mark areas of the plate as no-go zones and mark off that corner of the bed.  A rectangle described by 220x300, 300x300, 220x263, 300x263 did the trick.  It (obviously) costs me a corner of the build plate but the functionality gained is worth the trade off IMO.  I may in the future look to see if I can drop the scrubber by a few mm, enough to clear the probe.

Finally picture of the printer working and printing some skirt parts

[xyleth]

In common with many other people I started hitting the 'Communications failure during move' error.  This was especially noticeable during QGL and was getting to the point where it would succeed maybe one time out of three or four tries.  There is a big thread along with a number of suggested remedies over on the Klipper discourse here

The common factors seem to be people using multiple MCUs on lower powered SBCs with multiple devices connected to the USB bus.  Apparently the more devices you have the worse it gets.  There were several potential solutions listed, some more hacky than others, but they consensus always seemed to be a more powerful unit, or one with a different USB implementation, solved the problem.  So time for a brain transplant!

I opted to go with a refurbed Beelink BT3 Pro running an Atom X5 Z8350 with 4Gb of RAM and 64Gb eMMC for mass storage.  It's probably a four or five year old box and no speed demon by modern standards but its low power and more than sufficient for Klipper.  It's roughly on a par with a rPi 4 but an Intel chip.  Onto this I loaded the x86_64 version of Armbian which, while marketed at ARM boards, does support Intel natively. 

I chose Armbian because it is targeted at the IoT market and comes out of the box with settings and package choices tuned for that market, so it was ideal for my needs here.  Once that was set up and updated I used KIAUH to install Klipper, Moonraker and Mainsail without any issues and then added the various addons I needed like the Cartographer plugin and so on.  The I installed Klippain, copied across my custom config files and moved all the USB and HDMI connectors from the BTT Pi to the Beelink.

The one bit that wasn't configured out of the box in end was the CAN interface.  The Octopus pro showed up in lsusb as the CAN bridge device and the CAN0 network was present but down and refused to come up.  All you need to do is add an interface config as per Estorical's guide for ifupdown systems (not the Systemd config) and then after a reboot everything just...worked.  I haven't tried cross-compliling firmware yet but I did see that KIAUH installed all the necessary tools onto the system, and in a worst case scenario I do have access to rPis I could use to build the Klipper firmware modules.  That does remind me that I need to re-install Katapult though.

The Beelink only has a single USB3 port which I have left free for future expansion like cameras, but USB2 seems to be more than enough for Klipper.

I managed to get one test print in which worked fine with no errors during QGL before I had to head out.  Work have me away in Madrid for the week so there will be no more printer based shenanigans until the weekend

Pic shows the current location of the Beelink - I will probably have to think of a better mounting solution long term.  It is currently powered by a wall wart but takes 12v DC via a barrel jack so there is scope for tidying that up too.

[xyleth]

So as I feared that Y stop JST connector in the A drive pod has proven very unreliable.  It will work for a while and stop for no apparent reason with the end stop registering as triggered.  If you don’t notice it before you home it can cause some spectacular toolhead crashes!

Pulling the JST out and reseating it, with a fair bit of wiggling, restores the function for a while but it always ends up failing again.  Just getting to the connector requires disconnecting the CAN umbilical and removing the plate that holds the umbilical in place on the A drive. I _think_ it’s the pins in the socket and the receiver in the plug moving in their respective bits of plastic and coming apart.  But troubleshooting that, let alone fixing it, would require unmounting the entire gantry and tearing down the A drive pod to get to that part.  Which, frankly, is not something I’m down for right now!

So this evening I sat down and set up sensorless homing on both X and Y axes, which actually went fairly painlessly. Unsurprisingly Klippain already has most of the config in place, you just have to uncomment the line for sensorless homing with relevant steppers (TMC2240 in my case) and then tune the sensitivity.  You don’t, as I learnt, need to add the macros in the Voron setup guide for sensorless.  The existing Klippain macros already take everything into account once you enable the correct line.

I found there was a very narrow range of SGT values that worked.  Anything below 0 and the toolhead doesn’t move.  Anything above 4 and it tries drive the toolhead through the extrusions!  So taking the guides advice I settled on a value of 2 for both axis and that seems to work.  You still hear the micro switch click as they’re still there which is comforting, perhaps falsely so.  But it does seem to work.  Hopefully this will be more reliable.

Edited 2 hours ago by xyleth

[dasbooter]

Sorry for not ignorance is this a possible result of your pin mod? Funnily enough was about to press purchase on a kit and just can't get brave enough. I think it's more of a time thing though?

[xyleth]

5 minutes ago, dasbooter said:

Sorry for not ignorance is this a possible result of your pin mod? Funnily enough was about to press purchase on a kit and just can't get brave enough. I think it's more of a time thing though?

Yes and no.

the pin mod has the option to print an alternative A drive base plate that relocates the Y end stop into the A drive assembly.  That, or more accurately the recommended JST connector method, is what is causing my issue.

You can absolutely do the Pin mod without doing this, just leave your Y stop as-is. You could also skip the JST connector and solder the end stop wires directly to the micro switch, that would prevent this issue.

I wouldn’t let what is probably my dodgy soldering, crimping and / or terminal assembly put you off doing the otherwise very worthwhile pin mod.