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

[YaaJ]

Received a second Biqu Microprobe that will be tested as a Z-endstop/Nozzle probe device :

2 USB hubs, one came with no quartz ! Amazing ! Each is a 1 to 4, need one, should do it... I plan to test this on the gantry, and plug the U2C + a camera (or two...). No idea how it will behave. At worst, there will be a second USB cable from the Pi...

Also received some USB modules for connections

 

[YaaJ]

Now it's time to complete the wiring job. First, the bottom panel. It has to be screwed to the electronics drawer. But nothing to attach it. (the holes in the custom skirts are for "easier" access to the screws. 100% useless !

4 holes + cage nuts (automotive supplies, bodywork aisle). M4.

Some strong magnets will help, catching chips. One on the sheetmetal in order to magnetize it, and others below. Then, vac cleaner. No problems with the electronics later.

One screw had its hex socket damaged, and finally seized in the nut while attempting to remove it (soft A2 stainless, not a great idea for this job !)

Had to use the bolt cutter. Replaced the cage nut with another one, the bottom panel now is in place, and the printer on its feet again.

Am I the only one using a bolt cutter for a Voron build ? Should it be added to the mega List ?

[YaaJ]

Didn't update this thread for a while.
Because suffering PTNE syndrome. Project That Never Ends. Emptyness feelings. Near to depressive desease. Not even kidding. Is there some psychiatrist on the forum that could explain ?

More time spent watching stupid videos on YT than working on the 2.4... But some progess anyway.

The most difficult part : cable management, measuring 5 times before cutting and crimping.

Made some fanout modules, from protoboard :

Also made some toolheads from the old V6, and CHT + hardened nozzles purchased just a couple months before the 2.4 project started.

These toolheads will  be shared with the bed slinger...

A ambient temperature sensor, made with a old 104GT...

(reached the upload limit)

[YaaJ]

The sensor senses the cooling air intake :

The most complicated part of this build : cable management on the rear side.

The drawer cable chains tend to sag a bit, and interfere with the structural parts. There's a part I don't know how to describe, that forces the drag chains to go where they have to go...

limit reached...

[YaaJ]

The electonics drawer is completed ! Added a cover on the SMPSU connections. RSP-200-24-Cover.stl. Perfect fit !

Currently printing the panel latches, two colours, on the good old SuperTornado :

Also discovered a falw in the drawer design : it is free, as the kitchen drawer it is made from... Have to add some safety to this thing. Will try to repurpose this old HDD lock :

Not only is it for safety (think of a kid opening this and touching high voltage...), but also to make moving the printer around easier : the drawer tends to open easily by itself. Kitchen furniture wasn't designed to be moved around...

Waiting for some parts, the parts that are always missing by the end of a project. Currently some Wago's (green ones, for bed earthing)

[YaaJ]

Made 2x 30 of those panel clips. Maybe a bit too many... Of course, nearly out of the required screws...

Hope that they will fit properly ; 3.5mm where there's 1mm adhesive foam, 6mm for 3mm foam, with the LDO panels.

My fingers do hurt...

[claudermilk]

LOL, you went a little overboard. There certainly won't be a shortage of clips.

[PFarm]

Cool drawer set up impressive!

[YaaJ]

2x 6 spares : 12 per panel, 30 printed when 24 needed.
Half of the spares are not usable because of some black residues on the bed that are stuck to the white parts. Despite bed cleaning with acetone.
On the 0.2, discovered that the best technique fo buildplate cleaning probably is printing a one layer very large square (brim), but would have took ages on the 310x310 slinger bed.
Now waiting for screws, while printing extrusion covers for the deck ; after seeing what heat + electronics bay fans can do !

[YaaJ]

1 hour ago, PFarm said:

Cool drawer set up impressive!

Unfortunately, sharing the design is worthless, as it was done with scrapped parts.

Was it worth the effort ? Not sure. Because modifications on wiring still require removing the bed and deck. It is usefull for accessing the Octopus and Raspberry SD card only. Could have be done an other way with SD extenders... The hood mod is better.

Was it fun ? No doubt !

[Penatr8tor]

@YaaJ Do you think you could share your SB toolhead mount/s for the Tornado?

I gave mine to my friend and he upgraded it to use an old Afterburner that I had laying around and now he's looking at upgrading to Stealthburner.

[YaaJ]

1 minute ago, Penatr8tor said:

Do you think you could share your SB toolhead mount/s for the Tornado?

Of course. Didn't share because since I made this, lots of makers made the same mod.

It is made for the standard 3-wheeler carriage that can be found on most of chinese bed slingers. But it could reduce the X travel, and even the Y travel. The X-Z joints and Y axis were designed and machined to extend a bit these travels on both ends. For a Clicky and a auto Z offset I removed because of magnets falling apart. And no time, the 0.2 then the 2.4 were more interesting projects. Currently leveling the old school way ! (it works !)

Works on Tornado, CR10, Ender 3, etc.

Just a mix between the Voron X carriage and the CW2 extruder body. No rocket science. But minimal overhang. Prints amazing with Klipper and IS. 3x faster, and better.

1 or two days, just the time I remember how to edit HTML pages (and the HTML tags) on Github for some basic instructions... Will start tonite. (go to bed at 4:30~5:00 AM, it's 8:00 PM)

Will post the STEPs and STLs. License : DWYW 0.0.01 (do what you want). Will PM to you when it's done.

Love the SB. It's been incredibly reliable. Printing half the parts for the 0.2 and 2.4 exclusively :

[PFarm]

3 hours ago, YaaJ said:

The hood mod is better.

Good to know since I'm about to do that mod to my V2.

[YaaJ]

Also considering it.

But without the pneumatic springs because of the huge forces they apply to the frame.

At the end of the todo list, because it's been one month saying every monday "it will print sunday", and it is still not printing. Until sunday... Or the next one...

[Penatr8tor]

3 hours ago, YaaJ said:

Will PM to you when it's done.

Thanks! 

[YaaJ]

Don't tell her what I did with her rolling pin !

Curruntly working on the auto Z offset...

The Z endstop is used as a nozzle probe, and a reference for the ABL device. The Z endstop is declared as a "knockprobe". See https://github.com/TypQxQ/KnobProbe_Klipper/blob/main/knobprobe.py

This way, it is possible to measure any variation relative to a first calibration, and apply the value in the start gcode.

It is not really a "project" : the macros are (nearly) ready for use ; wrote them 1 year ago for the bed slinger, at first for the BLTouch. Then for a DIY Klicky. Of course, I didn't keep the original BLTouch macros, but there's not much difference ! (simpler, have to remove attach/detach macros)

The macros I've been using with a dockable probe on the bed slinger. Stopped using them when a magnet decided to live its own life... No time to lose : one profile, one toolhead, one nozzle for the Voron's parts...

These macros were also using the "Dockable_Probe" plugin ; a pull request on Klipper Git.

Dockable_Probe.zip

Plan to replace the mechanical switch with a Biqu probe mounted upside down, a steel pin and a spring. I don't trust the mechanical switches...
A 5 wire cable has been routed from the Octopus to the rear of the bed. For now, just using the sensor pin.

Edited February 2 by YaaJ

[mvdveer]

This is a Voron-Techno-Majestic! Admire your innovation.

[PFarm]

@YaaJ Love the use of the rolling pin! 

Edited February 3 by PFarm

[YaaJ]

The rolling pin comes from a common advice about the mag mat : some weight on top overnight.

Let's say the rolling pin, when pushed against the surface, has a contact width  of 0.5 cm (5mm), wich is pessimistic. Probably less, depending on the nature of the mat. It's pretty hard.

Then, if we apply a vertical force of 40 kgf (easy, half the operator's weight). This will apply a pressure of 40 / (0.5 * 30) = 2.66 kgf/cm² = 2.66 bar. Probably 2 or 3 times more...

Let's calculate the equivalent, with a weight laying on the surface. 35 cm x 35 cm = 1225 cm². Applying the same pressure with a weight would require 3266 kg ! (3266/1225=2.66).

Two orders of magitude more than 30 kg of books. The books can't compete. Never did the math before, the result is impressive ! Equivalent to a small hydraulic press.

[YaaJ]

The auto Z offset macros are ready and working.

Here's some typical results ; tolerances for retries are set to 0.005 (5 microns) for both the two probes : Microprobe and nozzle probe microswitch.

(knobprobe = nozzle probe microswitch)

Measurement, after a first calibration :

 14:54:46  // KnobProbe triggered at z=-1.155578
14:54:47  // KnobProbe triggered at z=-1.150891
14:54:49  // KnobProbe triggered at z=-1.149641
14:54:49  // KnobProbe samples exceed tolerance. Retrying...
14:54:50  // KnobProbe triggered at z=-1.148391
14:54:51  // KnobProbe triggered at z=-1.147141
14:54:52  // KnobProbe triggered at z=-1.146203
14:54:53  // Result is z=-1.147245
14:54:56  // probe at 268.000,333.000 is z=0.823797
14:54:58  // probe at 268.000,333.000 is z=0.822234
14:54:59  // probe at 268.000,333.000 is z=0.821297
14:55:00  // probe at 268.000,333.000 is z=0.820047
14:55:02  // probe at 268.000,333.000 is z=0.820984
14:55:02  // Result is z=0.821672
14:55:02  // ref_delta_z: 1.945624999992464 , cur_delta_z = 1.96891666666033
14:55:02  // gcode_offset: -0.023291666667865973

Now the Revo nozzle is unscrewed (half a turn ~ one turn or so, the thread being M4x0.7)

 14:58:07  $ GCODE_OFFSET_CALIBRATE
14:58:31  // KnobProbe triggered at z=-0.710891
14:58:32  // KnobProbe triggered at z=-0.705578
14:58:33  // KnobProbe samples exceed tolerance. Retrying...
14:58:34  // KnobProbe triggered at z=-0.701516
14:58:35  // KnobProbe triggered at z=-0.702453
14:58:36  // KnobProbe triggered at z=-0.699953
14:58:36  // Result is z=-0.701307
14:58:40  // probe at 268.000,333.000 is z=0.795047
14:58:42  // probe at 268.000,333.000 is z=0.799109
14:58:43  // probe at 268.000,333.000 is z=0.800047
14:58:44  // probe at 268.000,333.000 is z=0.800672
14:58:44  // Probe samples exceed tolerance. Retrying...
14:58:46  // probe at 268.000,333.000 is z=0.801297
14:58:47  // probe at 268.000,333.000 is z=0.801609
14:58:48  // probe at 268.000,333.000 is z=0.801609
14:58:50  // probe at 268.000,333.000 is z=0.800984
14:58:51  // probe at 268.000,333.000 is z=0.801297
14:58:51  // Result is z=0.801359
14:58:51  // ref_delta_z: 1.945624999992464 , cur_delta_z = 1.502666666659024
14:58:51  // gcode_offset: 0.4429583333334399

We see that the nozzle was lowered by 0.44mm, and the parameteer for SET_GCODE_OFFSET will be +0.44mm

Now the nozzle is re-screwed all way up and seated.

 15:02:40  $ GCODE_OFFSET_CALIBRATE
15:03:04  // KnobProbe triggered at z=-1.145266
15:03:05  // KnobProbe triggered at z=-1.143078
15:03:07  // KnobProbe triggered at z=-1.142766
15:03:07  // Result is z=-1.143703
15:03:11  // probe at 268.000,333.000 is z=0.823797
15:03:12  // probe at 268.000,333.000 is z=0.822859
15:03:13  // probe at 268.000,333.000 is z=0.822547
15:03:15  // probe at 268.000,333.000 is z=0.823172
15:03:16  // probe at 268.000,333.000 is z=0.823797
15:03:16  // Result is z=0.823234
15:03:16  // ref_delta_z: 1.945624999992464 , cur_delta_z = 1.9669374999936533
15:03:16  // gcode_offset: -0.02131250000118934

The Microprobe makes the macros much more simpler than using a Klicky : no need to attach/detach the probe !

Edited February 4 by YaaJ

[YaaJ]

Was tired with all these crappy gcode macros inceptions. All the code now is in a custom plugin, based on KnobProbe by TypQxQ

For the standard Voron nozzle probe.

Renamed the plugin NozzleProbe ; added attributes and code are surrounded with '##########' separator lines, making it easy to locate and edit. Very first version, but unlikely it will be massively modified. File name : nozzleprobe.py

printer.cfg :

[nozzleprobe]
; see https://www.klipper3d.org/Config_Reference.html?h=probe#probe
; NOZZLEPROBE_CALIBRATE :
;     set reference delta z ; save with SAVE_CONFIG
;     must HOME first
;     auto returns to initial position
;
; NOZZLEPROBE_AUTO_GCODE_OFFSET :
;     to be inserted before running gcode
;     first HOME / QGL / MESH
;
; generic probe settings
pin: ^PB7 ; BLTouch sensor PIN
deactivate_on_each_sample: False
z_offset: 0 ; must be provided, but not used...
speed: 5.0
samples: 3
sample_retract_dist: 2.0
lift_speed: 5.0
samples_tolerance: 0.01
samples_tolerance_retries: 3
; nozzleprobe specific settings
pos_nozzle_probe: 269, 356  ; position for probing nozzle ; probe coordinates
pos_ABL_probe: 268, 333     ; refeerence position for ABL probe ; probe coordinates
travel_z_lift: 5.0          ; z lift before traveling, uses 'lift_speed'
travel_speed: 100           ; horizontal travel speed

Now only gcodes (no macros)

First one is

NOZZLEPROBE_CALIBRATE

Gets a reference delta Z between nozzle trigerring point and ABL probe triggering point.
The value is saved in printer.cfg

#*# <---------------------- SAVE_CONFIG ---------------------->
#*# DO NOT EDIT THIS BLOCK OR BELOW. The contents are auto-generated.
#*#
#*# [nozzleprobe]
#*# ref_delta_z = 1.9641666666669928

Then, delta z will measured again before each print, after HOME, QGL, BED MES, etc. by inserting the command

NOZZLEPROBE_AUTO_GCODE_OFFSET

This command calls

SET_GCODE_OFFSET Z=nnn.nnnnnnn

before starting print, and compensates for any variation in nozzle length / toolhead height.

It was yet another project inception...

 

nozzleprobe.py

Edited February 6 by YaaJ

[YaaJ]

Today, first print ! With PLA. A 20mm x 20 mm cylinder, vase mode. First one at 40mm/s (near to perfect), second one at 400mm/s (1000% multiplier) ; crappy, it exceeds the Revo HF max flowrate : 0.45 line width, 0.2 height -> 36mm3/s = a bit too much ! Had forgotten how slow 40mm/s is since I switched to Klipper 1 year ago ! Had to verify in the slicer, thought something was wrong !

Without any calibration. Beautiful. The 2.4 is a beast. Just a bit too loud at 0.5 meter from the desk...

Also did some testing with the bed heater. The famous rule of thumb 0.4W/cm²...

(opened a thread about this rule on Voron Design Forums)

Did 3x BED_MESH_CALIBRATE, at 20°C, at 60°C/60%, and 60°C/100%, after 30mn. Of course, with a PID_AUTOTUNE before each test, as the power is not the same. Starting from a cold bed (ambient temperature)

Tested open air : no potential problems with a warm bowed gantry, and worst temperature gradient between the heater pad and the top surface. The ABL probe is the Biqu Microprobe, 3 samples, 0.01 samples tolerance, and 16 microsteps resolution.

Got this :

Ambient (20°C)

60°C, heater @60%

60°C, power @100%

At ambient and 60°C, this bed is near to dead flat whatever are the power and the temperature. A nice piece of metal, most likely grinded (stoned, not machined with a shell mill). Beautifull. Will cry the day the mag mat will have to be replaced...

Not sure grinding makes a huge difference over milling : there's the mag mat and the plate on top...

We can see that heating at 60% or 100% makes absolutely no difference !

Already printing on the v0.2 at 80% (in theory should also be 60%), never had to re-level the bed for the last 3 months, and no first layer issues. OTOH, printing ABS only, on the textured side. Will set it to 100%.

Love this machine so much that I procrastinate everything. Did the same with the v0.2. It will live in another room (no alone, with its best friend the v0.2).

[YaaJ]

At 100°C, open (no panels)

Cold :

100°C, power 60%

100°C, power 100%

I see no reason not to use 100% power !

Installed Klippain, and plyed with it, using the ADXL345, mounted to the side of the tool cardridge (not the EBB36)

For the belts, I was unable to get perfect results with the "GT2 Belt Tension Meter" : https://github.com/Diyshift/3D-Printer/tree/main/GT2 Belt Tension Meter

It is great at coarse tuning, but not at fine tuning. But very handy because we don't have to bother with a belt length.

Ended with Spectroid that showed not better than 10%. A few minutes later, got this ; tuned at 120Hz over 150mm instead of 110 Hz :

The flat peak (!) around 125 Hz is the LDO fan, rattling as hell.

Then, the Input Shaper :

And the Vibrations Calibration :

Next, the Ellis' calibration for eSun PLA+ "natural", a Benchy sliced for less than 30mn, and dsomething I was asked to print : https://www.thingiverse.com/thing:4750836

[EDIT] forgot to say... Had to push the current to 1.4A in order to get rid of crazy noises (of course spreadcycle, not stealthchop). And BTW validated the slanted fans (electronics fan) blowing to the drivers. They spin at 40% when the motors are energized, 30% idle. Turned them off, kickly got a overheating error and Klipper shutting down. A/B motors run at 55°C with 1.4A (delta T = 35 °C). Should do it at 60°C chmaber, given the temperatures steppers are rated to.

Edited February 11 by YaaJ

[YaaJ]

Some more numbers. And a huge crash !

First, printed a speed Benchy without any calibration. Forgot to use the chronometer, I rely on Fluidd. 23mn 30s, including all the PRINT_START gcode... Later ran the PRINT_START macro (HOME + QGL + HOME + BED_MESH_CALIBRATE + NOZZLEPROBE_AUTO_GCODE_OFFSET 4mn 30s : it was a 19mn benchy. Nothing crazy, as I tested the Super Tornado bed slinger at 29mn... but happy.

Sliced with accelerations at 10000mm/s². Decided to test with 20000...

Dumb idea !

Got this :

Not sure what happened, but 20000 was a lot too much ! I suspect that the head was not at height after the auto Z offset probing and started too soon. Or a problem with the Z endstop. Don't know... The plugin auto-moves the toolhead to its previous position after probing (i.e. the last meshing point). We see were it entered right into the PEI. Can imagine how badly the bed would have been scratched if "my" pluggin allowed the printer to go directly to the printing area !!!

But I didn't only damage the bed : also destroyed a brand new Revo HF 0.4mm nozzle ! Had to break it for removing the heater cardridge...

Luckily, having many quick change toolheads, it was temporarily replaced with a V6 CHT. Two nozzles are on the road.

Finally RTFEPTG (The Elli's PG). Speed/accel tests.

Got :

500mm/s : pass
600mm/s : fail
19000mm/s² : pass
20000mm/s² : fail

Will be limited to max 400mm/s and 15000mm/s²

While slicing the crazy speed benchy, completely overlooked the fact that the motors are 0.9° steppers, and much slower than usual 1.8° ones. I was basing the tests on speeds/accels I found on the web.  Lesson learnt.

Edited February 15 by YaaJ