[X][P][T] CR600S setup and first test prints

A project log for Coaxial8or [gd0144]

Full-colour FFF? Multi-materials with unparalleled interlayer bond strength? Abrasives without abrasion?

kelvinakelvinA 02/03/2024 at 08:131 Comment

It took much longer than I expected, but I've finally turned the pile of parts into a CR600S to conduct printing tests with the 4-in-1-out Machined Coaxial Hotend. Unfortunately, it lost against the leaks and the results I was able to get out of the two test prints were inconclusive other than knowing that I can print 100% in extruder E0 and E1.

Putting the printer together

I started putting the CR600S together on the 23rd Jan, found out what parts needed to be ordered on the 24th and got them in on the 25th and 26th. 

Extruder mounting

I had some Nema17 mounts from when I was testing for the #SecSavr Sublime [gd0036] and I figured out how to be able to mount them to the CR600Ss extruder mount plate:

I figured out that it would be possible to get a 10cm length of 2040 extrusion to actually use all 5 mounts I have to mount the 4 extruders.

E2 gave me the most issue because, as I was installing it, I noticed that the grub was loose. It also wasn't tightening correctly. I put in a grub from the bag of them, but then I noticed that some motors could be spun from the white gear things and others can't. For all but E2, the issue was that the gear was too high up. However, it took me trying the E2 extruder on E3 to realise that the (longer) grub screw is the reason E2 wasn't working. I had to transplant a screw from one of the BMG gears I got for the #SecSavr Select [gd0091].

Some other parts ordered

I decided to get this 8 meter bundle for £12 instead of a 2m for £7 because all this is experimental and there was a possibility I'd need more than 2m. I went into this thinking that I could get 5m for £5 or something. The included cutter makes things so much easier than what I did in the past, which was a scalpel against a hard surface.
I got this instead of the cheaper £4.59 option because I specifically wanted at 5 of the same height so that I could put one under the center of the bed, which is slightly lower in elevation than the corners.

It wasn't until I was wiring things on the 29th that I realised that I hadn't actually seen my box of ferrules in years, so I got 600pcs of those.

Parts redesigned in CAD

The only thing I modified so far was the screen foot:

The dimensions of the BTT TFT24 were very similar to the FysetC RGB 12864 installed currently. The plan was to use this as a test print.

I added a cutout so that the mains wires didn't have to be disconnected when swapping this part out if needed in the future.

More mechanical ideas

Possibility of a micro-bowden (direct drive) fed input?

This is what gave me the idea:

And this is what I plan to use:

BotObjects heatsinks?

Whilst on my Fabbaloo travels, I came across this:

As you'd see later in this log, I think something like those brass inserts is something that is really needed if either the Machined, Printed or Unibody Coaxial Hotend is going to work long-term.

Firmware changes

Swapped the bed and E3 pins so that I didn't have to connect a jumper set of cables just so that I could power the SSR (which certainly doesn't need bed-heater levels of current to run).
I'm not sure why, but there is a specific build environment specifically to use the USB A port on the Octopus Pro. This is used since the leg would block the one on the TFT24.

JST XH crimping using the SN-2549

As I lost one of the jaws from when I was trying to make my own ball-chain splicer for #Tetrinsic [gd0041], I got a new one:
Unlike the one it replaces, this one has a narrower and smaller jaw at the end that I tried (only after multiple unsuccessful attempts with the 0.25 option) that didn't malform the crimp.
My JST crimps came all separate. I've seen some YT videos where they first have to be cut off a line of them. The best strategy I've found seems to be the following:

Trying PTFE tape

I first did an MPC autotune which was very straightforward, then I kept the hotend at 140C so that I could unscrew one of the heatbreaks. 
After that, I tried to put PTFE tape on but ultimately failed. 
I then had an idea of using the hex heatsinks along with thermal compound as a possible future idea:
Anyway, I cut 2 lengths of about 36 and 46cm (35 + distance between edge of plastic and cutter blade) and everything was ready for the test print.

X axis endstop

I had to reorient the x axis endstop because the cover got in the way:


Well, looking for a slicer. Cura still only supports 8 extruders and the UI gets laggy. I looked into Orca Slicer but it seemed to only support things like AMSs and MMUs and not multiple extruders.
In PrusaSlicer, I was easily able to add 32 extruders with no issue, but I dropped it to 16 as that was the amount of virtual tools set in Marlin. Also unlike Cura (even with the filament plugin), materials had things like custom GCODE and dependencies, so in theory I could add M165 commands for each mix and say that it should only be visible if specific materials are loaded into Extruder 1 - 4.
Also unlike Cura, PrusaSlicer supports volumetric flow limits which means that I probably won't need access to almost every setting through the material like I do in Cura.
Actually, I just looked through what I'm actually changing now. All the speed settings would be solved by the "Max volumetric speed" setting, I've never actually used ironing ever since I could get Cura (and now PrusaSlicer) to print concentric topmost/bottommost surfaces, typical things like extrusion multiplier and retraction is already in the PrusaSlicer filament setting pane, linear advance could be set in the start GCODE and I've actually never changed the bridge settings between materials (other than speeds).
I threw together a profile that seemed to look OK (see above) and then modelled up a test mix swatch. I'm going straight into this because I don't know how long this heatblock is going to last so I should "move fast and break things" as Elon would say.

First print attempt

The first thing that happened is that nothing fed in.  I opened the ptfe's and only one even got in. The other 3 got caught.
It turns out that the bowden coupler itself has a catch that the ptfe will catch on unless it's effectively perfectly straight.
I already expected that whatever bowden coupler I used in this hotend would cause issues sooner or later, but I wasn't expecting before I've even gotten a print out.

Expected ratios for vtools (starting at T0 = 1.):

  1. 1/0/0/0
  2. 0/1/0/0
  3. 0/0/1/0
  4. 0/0/0/1
  5. 0.5/0.5/0/0
  6. 0.75/0.25/0/0
  7. 0.9/0.1/0/0

This is the first result:

This is the hotend after the print:

I looked at the virtual tools after and for some reason, T0 was set as a 9:1 ratio. Also, the wipe tower volume was too small to get proper transitions.

Second print test

I thought that I might as well test to see if the heatbreaks just needed to be tighter or something, so with the help of one of the silicone levelling spacers, I swapped it out:

Then I sliced with a larger purge amount and even increased the amount of skirt length for extra measure:

I found out something about M165 that it sets the mix for the active extruder (it's not a 1-line version of M163) and so added it to the filament gcode of each virtual tool (instead of the start gcode). 

Expected ratios for vtools (starting at T0 = 1.):

  1. 1/0/0/0
  2. 0/1/0/0
  3. 0/0/1/0
  4. 0/0/0/1
  5. 0.7/0.3/0/0
  6. 0.8/0.2/0/0
  7. 0.9/0.1/0/0

This is the result:

The hotend was in an even worse leakage state, with the experimental V6 heatbreak being awful:

Unlike the other 3, the filament was stuck in this one even when heated to 160C. I did manage to get it off though:


It's interesting that the first test was too black and the next test seemed kinda white. In terms of mixing, the results seem inconclusive. What I expect is that the overextrusion I observed had something to do with it. It could also be that material went in places it wasn't supposed to go, which also made me realise that I'm actually trying to get a "controlled leak" through the nozzle orifice and nowhere else. I saw some research a bit back which suggested that hotends work at 2MPa when printing, so it's more understandable how some can slowly ooze out.

Even the nozzle and coaxial-insert threads are starting to leak now, which is not something that future machined designs have addressed. The deckingman approach, which is a machined-flat mating surface and plates where the M6 threads are tapped and then screwed ontop, sounds like the method to go, especially with a printed heatblock solution. 

Also, I certainly need to be using compression fittings or that cool brass heatsink seen in the BotObjects 5-in-1-out hotend.


kelvinA wrote 03/20/2024 at 21:16 point

I was making changes to my Marlin config when calibrating for the SanBrother DDE's and found that my microstepping was set to 8 for extruders even though the steps/mm were the usual 16-microstep value of 418.5. It's possible that these test prints were trying to extrude 200% of the intended amount.

  Are you sure? yes | no