Voron Trident build

Manta M8P plus CM4 cooling

What I really want to do is mount a fan to directly cool the stepper drivers and a second fan to directly cool the Raspberry Pi.

So I found one mod that mounts a 40mm fan on the CM4 module and a second mod that mounts a 5015 blower fan in a air duct for the steppers. However, when I put the models together, I don’t think I can use both of them at the same time.

This might be a design project. Not right away.

Systemd yak shaving intensifies

I have this vague uncomfortable feeling that the instructions in the standard CANBus guide are slightly wrong, but I gave up on worrying.

From my reading, it sounds like I would want to skip the ifupdown section because I’m definitely using a fairly modern version of linux on the Pi and it’s definitely using systemd-networkd instead of ifupdown… but for some reason I actually do have to follow both sections of the documentation and that feels a bit weird.

On the other hand, I can’t really find much documentation on using the CANBus inside of systemd anywhere and I gave up messing around and just did both sides and the whole thing works and this is really just silly and while I try not to pile onto the systemd drama situation I’m still somewhat blaming systemd.

I complained to a friend about this and she sent me a “Yak Intensifies” animated GIF.

CAN bus setup

• I decided to not bother with setting up katapult.
• Formbot helpfully provided a CANbus cable that had the required connectors and was heat-shrinked… and I had to cut off the heat-shrink cable and some of the insulation to stretch to make all of the connections.
• Oddly, nowhere in the manual is the meaning of the LEDs on the BTT SB2209 board specified. One of them is the power LED, one of them is the status LED. The Esoterical guide shows you setting that to on at startup if you take the Klipper Config route… which didn’t make sense at first.
• I accidentally put firmware to have the CANbus hanging off of PB12/PB13 instead of PD12/PD13. Oops! After I got that fixed, things seem to be working.
• The wiring diagram doesn’t show that you need to put a jumper for the terminator resistor on the motherboard side nor does it show that you need to put a jumper for the terminator resistor on the toolboard side and two more jumpers to set the voltages.
• There’s two cable-wrap-y thingies. One bag which is nylon-coated seems to be the “Cable Sheath 150mm” from the BOM, which means that it’s probably meant to be cut up and used to cover the other cables. The other bag is AD11.6 corrugated pipe, and that looks to be the umbilical part.
• The PG7 gland is just perfectly slightly too small to fit a JST connector without being de-pinned. Yay.
• Wrong cPIF part: I got three versions of the part for the umbilical bridge for the StealthBurner toolhead from my cPIF provider. I got the version for the 2 screw cable chain, the version for the 3 screw cable chain, and then the version for the EBB SB2209 CAN RP2040 board. However, what I need is the Formbot-provided umbilical bridge for the PG7 gland. If I’d thought this through, I might have gone straight to the PUG system.
• The prind container is installing the can libraries at every startup, which is I guess how Klipper wants to work? I try not to think too much about this because I don’t want to arrive to either the regular Klipper or DangerKlipper communities with a pile of patches. There’s a fairly specific yet also battle-tested way that we tend to default to in the infrastructure engineering world and Klipper doesn’t comply and it bothers me.

X and Y endstops

• The Y Endstop relocation mod was a bit confusing. Basically, there’s some stuff that’s intended to screw into the gantry that you don’t need. So you unscrew the XY Joint Cable Bridge part, either the 2 or 3 hole version, that you installed during the X axis section and then also the Y axis Y Endstop Bumper part you did during the Y axis section. The XY Joint Cable Bridge is where the bumper for the relocation mod goes and then you unscrew the front two screws from the motor mount and put the switch there.
• The X endstop just triggers against the XY joint without a special bumper?

Lead up to first print.

• Formbot ships a filament sensor for you but it won’t actually work, apparently. The pullup for GPIO22 doesn’t work, so either you need to wire up your own pull-up resistor or you need to switch it around… yeah, just leaving that out.
• The Tap manual klipper config section sucks because it doesn’t actually tell you about the virtual z endstop and the klipper manual doesn’t really cover this very well either.
• The way the stock wiring is designed, you are wiring a PT1000 through the 4-pin connector, so you want to be configuring the MAX31865. So the toolboard config looks something like this:

# This file contains common pin mappings for the BIGTREETECH EBBCan
# Canbus board. To use this config, the firmware should be compiled for the
# RP2040 with "USB" or "CAN bus (on gpio4/gpio5)".
# The "EBB Can" micro-controller will be used to control the components on the nozzle.

# See docs/Config_Reference.md for a description of parameters.

[mcu EBBCan]
#serial: /dev/serial/by-id/usb-Klipper_Klipper_firmware_12345-if00
canbus_uuid: xxxxxxxxxxxx
# Change the canbus_uuid to somethng

[temperature_sensor EBB_NTC]
sensor_type: Generic 3950
sensor_pin: EBBCan:gpio28

[temperature_sensor EBB_mcu_temp]
sensor_type: temperature_mcu
sensor_mcu: EBBCan
min_temp: 0
max_temp: 85

[adxl345]
cs_pin: EBBCan:gpio1
spi_software_sclk_pin: EBBCan:gpio2
spi_software_mosi_pin: EBBCan:gpio0
spi_software_miso_pin: EBBCan:gpio3
axes_map: z,-y,x

[resonance_tester]
probe_points: 100, 100, 20
accel_chip: adxl345

[extruder]
step_pin: EBBCan:gpio18
dir_pin: EBBCan:gpio19
enable_pin: !EBBCan:gpio17
microsteps: 16
rotation_distance: 22.6789511
gear_ratio: 50:10
nozzle_diameter: 0.400
filament_diameter: 1.750
heater_pin: EBBCan:gpio7
sensor_type: MAX31865
sensor_pin: EBBCan:gpio9
spi_software_sclk_pin: EBBCan:gpio10
spi_software_mosi_pin: EBBCan:gpio8
spi_software_miso_pin: EBBCan:gpio11
rtd_nominal_r: 100
rtd_reference_r: 430
rtd_num_of_wires: 2
control: pid
pid_Kp: 22.518
pid_Ki: 1.532
pid_Kd: 82.753
min_temp: 0
max_temp: 350

[tmc2209 extruder]
uart_pin: EBBCan:gpio20
run_current: 0.2
stealthchop_threshold: 999999

[fan]
pin: EBBCan:gpio13
max_power: 1.0
kick_start_time: 0.5 # Depending on your fan, you may need to increase this value if your fan will not start
off_below: 0.13
cycle_time: 0.010

[heater_fan hotend_fan]
pin: EBBCan:gpio14
heater: extruder
heater_temp: 50.0

[neopixel hotend_rgb]
pin: EBBCan:gpio16
chain_count: 3
initial_RED: 0.3
initial_GREEN: 0.0
initial_BLUE: 0.9
initial_WHITE: 0.0
color_order: GRBW

[probe]
pin: ^EBBCan:gpio22
x_offset: 0.0
y_offset: 0.0
z_offset: -1.600 # You need to tune this
speed: 15
lift_speed: 7.0
samples: 4
samples_result: median
sample_retract_dist: 2
samples_tolerance: 0.01
samples_tolerance_retries: 10
activate_gcode:
    {% set PROBE_TEMP = 150 %}
    {% set MAX_TEMP = PROBE_TEMP + 5 %}
    {% set ACTUAL_TEMP = printer.extruder.temperature %}
    {% set TARGET_TEMP = printer.extruder.target %}

    {% if TARGET_TEMP > PROBE_TEMP %}
        { action_respond_info('Extruder temperature target of %.1fC is too high, lowering to %.1fC' % (TARGET_TEMP, PROBE_TEMP)) }
        M109 S{ PROBE_TEMP }
    {% else %}
        # Temperature target is already low enough, but nozzle may still be too hot.
        {% if ACTUAL_TEMP > MAX_TEMP %}
            { action_respond_info('Extruder temperature %.1fC is still too high, waiting until below %.1fC' % (ACTUAL_TEMP, MAX_TEMP)) }
            TEMPERATURE_WAIT SENSOR=extruder MAXIMUM={ MAX_TEMP }
        {% endif %}
    {% endif %}

• I realized that the most recent time I’d wrapped the belt, I’d accidentally not gotten them around the front idler wheels, so I had to fix that and tighten everything up. The belts on the Zero were, for whatever reason, infinitely easier to deal with.
• Finally, I can go back into the Tap and Stealthburner and EBB SB2209 manuals and finally put the thing together!
• The Formbot PDF instructions that explain how to use their hinge system doesn’t mention that you should probably trim the piece of AD11.6 conduit they gave you. Presumably it’s meant to work on the 350mm x 350mm bed. They also fail to mention that you need to de-pin the CAN connector. I’m not overly happy with AD11.6 conduit. I’m probably going to redo the umbilical.
• I decided to use the Lever Connector Mount using Side Fan Supports to wire up the two side-fans because I don’t think I have any cases where I would want just one of the pair of side fans running and it simplifies the wiring overall.
• I also printed some pieces from the Modular Cable Management Clip set. It would be really cool if I could print a long version of the clip to completely hide the cables but if I scale the Z-height of the clips, I end up with a very very rigid clip.
• I managed to put the hotend cooling fan in backwards. Oops. Generally the label faces inwards, although I’m betting there are some fans floating around that have the label on the wrong side because that’s just how things are.
• Again, the Tap manual is not really useful here. If you go through the startup guide you can mostly follow the first part. When you get into the Bed Locating section, you can ignore the Z-endstop section (because there isn’t one). You can define the zero point as usual. You can ignore the Z Endstop Pin Location section. You do want to run the Probe Accuracy check. You want to run that a bunch because for whatever reason the accuracy gets better after it’s probed a few hundred times. You want to do the PID tune as normal. And you want to do the whole Bed Level process as normal. The Z offset adjust is silly because you want to do PROBE_CALIBRATE instead of a Z_ENDSTOP_CALIBRATE because there is no Z endstop.
• I backed off the tap required accuracy.
• Also there’s some Y loss from the Tap sensor. In my case, I’m losing maybe 5mm off the back of the bed.

First prints

I used the default SuperSlicer 2.4.x profile for a Voron 1 because I figured it wouldn’t stress the machine but would otherwise be mostly working. I loaded up the cube… and there was a tangle in the drybox partway through the cube. Oops!

• I copied most of the calibration settings over from the Zero because they both have the same hotend.
• Also I had to mess with the z side position_min setting to give it enough upward travel to avoid the “No trigger on z after full movement” error message.
• I ended up wasting some time to get things migrated to a version of the better print_start macro.

Finally, I started making prints. I decided to print a compliant cable clip because that’s an actual useful item but also relatively good to understand if the printer’s OK. After two of those, I printed the Gridfinity bin that I was frustrated about because I wanted it but it was too big to fit on the Zero’s bed.

So far the tally is that my Voron Trident has printed zero calibration cubes and zero benchies.

Labor Day sale

My long-term plan has always been to run either a Dragon Burner or XOL toolhead to replace the StealthBurner and also to replace the Tap with a Beacon probe.

Coincidentally there’s a 20% off sale on the Beacon site.

Next up to work on

• Motion compensation measurement and adjustment, making it print faster and cleaner.
• Bed meshing. I was kinda thinking of using the KAMP plugin but Klipper has native meshing at this point and that’s one less thing to install.
• Mount the Pi’s heatsink
• Mount the nozzle brush
• Figure out the nozzle purge situation. It’s less important to use pellet-style purge with a giant bed as compared to the Zero but I’m realizing that I really like not having purge stripes all over the place. Because of the Tap sensor I can’t put the pellet bin in the usual place.
• Mount the panels. I printed out some of the Voron 2.2/Annex clips that I’m probably going to use for most of the panels I’d want to remove. The Coroplast sheets for the back seem to be more like 3.5mm instead of 3mm.
• One of the screws on the right side of the gantry is accidentally a M3 T-nut and so I need to dismantle that a bit to get an M5 T-nut in there. Ugh.
• Refactor the macros
• Nevermore air filter
• Bed fans
• Dragon Burner or XOL and Beacon
• Filament sensor
• Serial number

Conclusion

The Trident is sitting where my Ender 3v2 used to sit. It’s not finished but it’s usable. And, like my Zero, the first prints actually look pretty darn good.

It’s a bit of a relief to be solidly in the “using it” side of the fence? There’s other things going on as well that have distracted me from building the printer but at the same time, there’s a bunch of half-finished projects that I can wrap up now.

Getting tired of not having a large printer, actually.

Also, I have a bunch of other stuff going on.

Tap

Okay, so two problems here… First, the manual is from Tap R6 and then there’s an errata file explaining what the differences are. Second, part of why you are urged to build a stock-configuration Voron is to avoid flipping around in the different manuals except that if a Voron comes with a default Tap, you can’t build it as a stock-configuration Voron, so flipping between the different manuals is a requirement.

Oh yeah, and I’m not sure if, were I to be a core Voron designer, I’d make Tap the default. Things are enough in flux these days.

Either way, I feel like if someone wanted to make a lasting improvement to the Voron infrastructure, making it amazingly easy to do documentation updates while retaining the existing positive qualities of Voron docs is probably a huge thing. What I get is the R6 manual with R8 errata and no clear directions for how to build a new printer with Tap and what I want is an R8 manual and enough hints in the Trident/V2.4 and Tap manuals such that I can assemble things.

• When I first got my PIF parts kit, I was worried that I wasn’t going to get the parts for Tap but after looking through the parts, I realized that they were there, in an unlabeled bag, so all is well and the system works. Unfortunately the two magnet holders are a little bit roughly printed.
• On pg 28, the Tap manual blandly assures you that “We could say that there is only one way they fit, but we don’t want to underestimate you” and … sure … but how is it supposed to fit? I think what it meant to say was so that you can see the heatsets through the side.
• The rail comes cut to the correct length already but you still need the spacer piece, which they also provide.

X Axis

• It would be really nice to remind the user on pg 99 which extrusion is used for the X axis instead of a bunch of pages later because I was getting all of my bits together.
• I had a bit of a struggle because I was inserting the wrong bolt into the wrong hole and I had to go back and look carefully at the images.

Belts

Folks usually say that the full-sized Vorons have easy to work with belts compared to the Zero… This did not turn out to be the case. I have spent a lot less time on the Zero’s belts than this.

• Okay so now the part where I get to flip between manuals. On pg 29 of the Tap manual, there’s some M3x6 BHCS screws that the instructions suggest I thread through the center part but it doesn’t tell me where. I’m assuming those are going into the carriage on the X rail.
• My cPIF provider printed the tap center part in black not the accent color. I guess I get no love from my cPIF provider. (That was a pun)
• Now I get to flip into the R8 errata file and realize that the tap center needs to be taken off the X-rail again so I can put the belt covers. The illustration there is quite unclear because I thought there was only one belt-cover, not two. Either way, now that the center is installed with the new belt covers, we’re in a situation where I can actually thread the belts.
• The Formbot Trident kit comes with two separate belts instead of one giant belt, although I ended up feeding one side through and using that to gauge the belt length anyway and then trimming both belts. This turned into a disaster. I thought I’d fed it through properly and left a reasonable length. I used a low-squish belt clamp to try and keep the two belts aligned while I tried very carefully to get exactly the same number of teeth. And then when I tried to lace it up the gantry wouldn’t be trammed right, which generally means that there’s the wrong number of teeth and it definitely felt like the first time through I might not have actually fed it properly. So I tried a few times and then gave up and ordered a fresh belt.
• The second time, I tried a different approach. I printed a bunch of back-to-back belt aligner clips and used them to split the entire length of the belts into two identical pieces and then fed the whole length through, which seems to have done the trick.

Bed

• As before, I’m streamlining the Fabreeko magnet bed that’s stronger and stickier as well as the Fabreeko bed heater that’s sticker and covers more of the plate. The only downside is that this Fabreeko magnet sheet is precisely sized to the build plate with no excess, which made the process a bit nerve-wracking. But, same technique as the Zero.
• I found the blind joint alignment nut to be quite handy for keeping the bed extrusions from getting all floppy. If I could go back, I’d suggest people print a plate of them at the start and use them in all of the blind joints.
• It looks like there’s no Z-endstop because the Tap replaces that. There’s no parts for making one regardless.
• I realized that I’d probably want to put in the frame thermistor corner in before I got much father, so I printed and added that one.
• Formbot provided a Bakelite insulator to replace the M4 thumb nuts that are usually used. I’m not sure how well they compare to the Ultem spacers that some people have been using but I’m imagining that they are far better insulators than a chunk of metal.

The Fabreeko Honeybadger Bed

The Formbot bed has the thermal fuse wired into the wire bundle. Apparently this isn’t super-great because this means if the thermal fuse pops (which it will, eventually) it’ll be annoying to replace it.

The Fabreeko Honeybadger bed doesn’t have the thermal fuse wired in. Furthermore, it’s got a funky wiring scheme that lets them ship a single bed that can either work in 110v or 220v mode. The 110 wiring diagram looks like this:

I ended up making a parametric Wago 221 mount based on an existing OpenSCAD generator from Thingiverse. If you leave it at the default settings it’s almost exactly the same geometry generated.

Conclusion

We’re finally at the “it looks like a printer” portion of the build.

An accidental clee day

I noticed that, because of Clee Day on the Voron Discord, West3D had a deal on titanium backers and a kinematics mount. So I was planning on getting steel backers and the kinematics mount later, but I guess we’ll have those on the way now.

Siboor Trident R1 vs Formbot Trident Pro

This came out after I’d ordered my Formbot Trident. It’s interesting! I look forward to people’s build logs.

Neither printer is going to have a taller Z-axis, you need LDO or MagicPhoenix CBT Pro to get that.

Using Cartographer instead of TAP is interesting. I wonder if there’s going to be a usable mode for some sort of Z-tap mode on the Cartographer in the same way as the Beacon has to make nozzle change and different print surface thickness a problem of the past.

The CNC components are interesting. The plus side is that they won’t break and they let you reach Forbidden temperatures. On the other hand, a lot of folks have spent a lot of time explaining why they don’t really improve your in any useful fashion, although the AWD mod is designed for CNC instead of printed parts… so maybe that’ll be better?

This one has the Manta v2.0 instead of the v1.1 on my Formbot. I think there was like 1 or 2 things that annoyed me about the 1.1 that made me want the 2.0 but I am presently forgetting.

I am glad that everybody’s putting in filters these days, although overall I am more a fan of the overkill StealthMax than the included Fume Pack or the smaller Nevermore Micro.

And, dono, I was probably going to do the Monolith Gantry at some point for mine, the sheet cooler was a thought, etc. The only thing that feels supremely wild is the beefy motors that may or may not make a useful difference to speed or print quality but definitely might score you drinks at the super-secret Voron cabal meetings.

My big concern about most of these kits is that we don’t have an organized fashion for kit-makers to communicate how they have deviated from stock, nor is the manual well-constructed to make it easy for deviated printers. The problem is that if Siboor doesn’t provide adequate support and self-help resources for their modded printers, it’s going to end up being an unfair drag on the Voron community. And this is really the same concern I have with the printer I actually purchased, it’s just that the Siboor one goes farther off.

Obviously, most people build a Voron only because they are prepared to actually go all of the way and, furthermore, an unmodded Voron is kind of a sad Voron. Also, if we’re looking at the kits people tend to get, there’s presumably a lot of Siboor kits out there.

Ugh wrists

So I’m pretty sure I strained a tendon in my right wrist while I was building the zero. SO I looked at the Project Farm and Torque Test videos and decided that the Skil Twist 2.0 looked the most reasonable for something to reduce the wrist-stress. I already have a set of reasonably good Makita metric drill bits and accessories but I did pick up some Ball-End Hex bits and a flexible shaft from McMaster Carr.

The problem is that blind joints need to fit inside of the blind joint hole and these won’t, so it’s not useful for the frame itself. Otherwise, it’s nice and most other things don’t use a wrench hole.

Linear Rail prep

I couldn’t print a bathtub for the rails so I ended up using a 2.5 gallon ziploc bag, which was kind of a mess because the bag leaked. But the rail lubing was a lot easier the second time around.

Component prep

Pg 13 on the manual and the BOM generator both suck and there is already a bug report. For my 300mm bed size build, I was able to spot the A extrusions (there’s 9 of them) and then the B extrusions (there’s 4 of them, with the particular set of holes). The C and F extrusions are the same length, the difference is that the C extrusion is tapped and the F is not. Then you end up with a pile of short extrusions. For the 300mm bed size, the H is the shortest, the G is only 2mm longer, the D is a bit longer, F is longer, then the rest. If you are building the even larger 350mm bed size Trident, the same sorting order should apply.

Ideally the kit-maker should just label the darn things. However, the BOM generator should show the letters of the extrusions and the manual should contain a better phrased of what I just wrote. I got the sorting order wrong, so I’m not sure what the right way is to sort this out.

Frame

Misumi makes an HCJ5 screw joint for 2020 extrusions that contains an BHCS M5 screw but has this little metal bit that makes the alignment process just a bit easier. Presumably a kit-maker could make some cut-rate version of this part and bundle this as a real pro feature. Or, I dono, you can always just buy a bunch of them, at $1.93 each for I think 21 of them?

The frame was a fairly quick assembly, it’s just that it’s a lot bigger than a Zero, so I ended up needing to get it to it’s designated spot on the shelf. And I used my 1-2-3 blocks and clamps to square all of the joints up. Basically, 1-2-3 blocks are supposed to be very square and if you get a pair or a set of four, they should be square in exactly the same way and dimensioned exactly the same. You could also use an engineer’s square but the nice part about 1-2-3 blocks is that they are heavy chonky things that are easy to clamp to and presumably more square than a cheap cast 90 degree clamp.

I ended up using a set of calipers to measure things to get the two middle A extrusions at the right height because using the extrusion as a reference only works with the 250mm version and I don’t quite trust the printed spacers.

Oh, and a note about my shelving unit: I designed it to replace an IKEA wire shelving unit that used to live there and it turns out that a shelving unit a smidgen bigger than 24 in × 24 in was about the right size, so I built it out of 2020 extrusion because that way it’s sturdy yet flexible. I assumed that I’d eventually replace my Ender 3v2 with something bigger, so I ended up looking at the size for a 300mm × 300mm × 300mm Voron 2.4 or a Jubilee printer or maybe just a CR-10, took the largest dimension of any of the three of them and used that to make a reference cube for the shelving design. So it was no surprise that it fits, but it’s good that it really does.

So, yes, I’ve got a 2020-extrusion based printer sitting in a 2020-based shelving unit.

A/B Drive and Idler

• The included shims on the Trident kit are a lot less variable than the Zero’s. I still decided to use the shims I got from McMaster-Carr.
• Also, I wanted to start the printer on an auspicious day in my spouse’s faith but wasn’t ready to actually order the kit, so I got some F695 bearings and some hammerhead M3 nuts. I’m using the F695’s I got on the auspicious day for the set of bearings on the idlers.
• The cPIF parts in bags with labels make it a lot easier to find stuff than the intricately packed foam sheets.

Y Axis: blocked by bad nuts

I was going to stop at the A/B Drive and Idler step, but I noticed the next step needed the roll-in nuts and I might as well see how well they fit… and yep, both the M3 and M5 roll-in nuts are the bad batch.

This is a bit easier for me to critique because I’ve got some genuine Misumi roll-in nuts, which I use to mount things on the shelving unit. You can see one of my Mismui nuts to the left side of the picture, with the supplied nuts on the right side. And the supplied nuts, they work as pre-assembly nuts if you want to do your Trident build the way you’d build a Zero with all of the nuts pre-loaded or you can kinda maybe sometimes crunch it down such that it acts as it’s supposed to, except not very well. Because the Mismui nuts work, it’s clear that someone just didn’t quite make it to the right tolerance.

So I decided that this was enough for this weekend and sent a message to the support address instead of fighting with it. I’m just savoring the irony that I got a bag of hammerhead nuts instead of a bag of roll-in spring nuts.

Conclusion

Considering that I did the Zero first now a Trident, everything feels absurdly giant! And it feels weird to not have a laundry list of things to preload ahead of time, it feels like I’m screwing up.