3D Printer yet?

[xrunner]

xrunner once I have taken all the chinesium parts out from my ender 5, I will sell everything to you.
Very messy, realistic old looking, imprecise, and fuzzy 3D print results guaranteed.

 

LOL

I was referring to general weathering from sunlight, storms, vegetation, and people. Not air to ground munitions. 

[mnementh]

What aboot fireballs, claw slashes and tailstrikes...? 

mnem
Oh, and I think that little doggie is peeing on one.

[mnementh]

Currently cogitating on something to make this a thing, rather than just a mockup with tape & paper-clip...

mnem

[Zucca]

Currently cogitating on something to make this a thing, rather than just a mockup with tape & paper-clip...
mnem

How the headphones are charged? Do you want a charger station as well?

[Zucca]

BTW can you confirm me that the nozzle fan (not the cooling part fan) is connected straight to 12V/24V for safety reasons?

[Monkeh]

Cheapness reasons, really.

[Zucca]

Thanks but since I can PWM only two fans on the new board, I will keep the nozzle fan always on.
The two PWM fans will be the cooling part and the heat sink board fan.

I think it is a good compromise.

[mnementh]

I found that annoying AF; my solution was a nice quiet always-on 80mm Arctic fan blowing across the mainboard to replace the always-on 40mm screamer. Cost aboot $6.



Later, I did this mod to enable PWM control of the hotend fan (my board had only ONE PWM fan available); I incorporated the 12V reg for that fan and LED lighting into that little sub-board along with the PWM pass-transistor connected to D7.

https://www.eevblog.com/forum/chat/3d-printer-yet/msg3051586/#msg3051586

Totes worth it; made my annoying little E3 clone into a civilized 3DP.

mnem
 

[mnementh]

   Currently cogitating on something to make this a thing, rather than just a mockup with tape & paper-clip...

How the headphones are charged? Do you want a charger station as well?

Naahhh... I'm not into that waste of real-estate headset-hanger bunk, and it uses off-the-shelf AAA NiMHs which it recharges using a jack for the same 5V power-pack as the transmitter.

I MAY add a pigtail for charging the headPwns to whatever part remains visible; I'm still vacillating over whether to make the adapter plate or just hack in an old iPud dock I found and stash the transmitter inside my credenza.   I think the deciding factor will be whether I ever get around to adding a switchbox for input from the Roku.

mnem
 

[Zucca]

Later, I did this mod to enable PWM control of the hotend fan (my board had only ONE PWM fan available)

nice job, you inspired me.

My plan:

1) Get a ABL Probe (I have a glass bed so I think it will be BLtouch) to do also the Zmin sensor.
2) With the free Z Min sensor do a PLA runout sensor
3) with the free Zmin pin (in my board the BL touch pins are separated) control the board fan
4) hot nozzle fan controlled by µC and not straight to 24V where now the board fan is.
 
Now I am doing my homework regarding the ABL probe.

Thanks!

[mnementh]

You're welcome.

I have had my dalliance with the BLTouch, and to be quite honest, it was a clusterfuck from day one. In principle a good idea, but QC on the sensor (yes, I bought direct from them) was utterly craptacular; it simply did not give consistently repeatable results. After replacing it twice, I cut my losses and went back to manual leveling and a glass bed; I have never once regretted that choice.

My curiosity was piqued by the CR6-SE's approach: mount the entire hotend on a load cell and use that as the leveling sensor. It does seem an elegantly simple solution; one that takes away all the adjustment and offset calculations inherent in using a separate sensor with constantly-changing nozzles, print surfaces, etc. 


Monkeh, if you're still around and can answer this one... How is the mesh leveling implemented in current Marlin FW and slicer profiles?

It seems to me that unless the firmware and/or slicer profile uses some algorithm to average out the difference between the mesh leveling offsets and a virtualized reference flat plane in the first few layers, ultimately mesh leveling just takes the 1-time assache of properly squaring the frame and leveling the bed away, and makes it a 100% duty-cycle assache for metal parts that wear on your machine: the Z-leadscrews/nuts/stepper. 

It also seems to me that constantly moving the Z-axis per the stored mesh-leveling offsets for the entirety of a print would add another place for microstep errors vs increasing weight of the printed part  to creep in and fuck with your head as well. 

Maybe once machines that come from the MFR with mesh leveling get more common, we'll have changes in the slicers and firmware that make this a more interactive process with manual control...?

mnem
 

[Zucca]

My curiosity was piqued by the CR6-SE's approach: mount the entire hotend on a load cell and use that as the leveling sensor.

and why I can't use the sensorless Z home of mine new shiny TMC2209, to detect when the nozzle is touching the bed? I can set the sensitivity...
Ohh FUU with the 2mm Z screw it will be a shit show... got it.

PS: When you have a magic brillant idea, 99% is just BS 

[mnementh]

Even on 8mm pitch leadscrew I suspect the force applied to get a valid reading would still be orders of magnitude greater; I suspect repeatability would be a total crapshoot and I doubt soft print surfaces like Build-Tak would survive, even if that approach didn't habitually crack glass, which I'm not sure it wouldn't. 

mnem
 

[xrunner]

Just enable & use manual mesh bed leveling in Marlin. It has a menu selection in the Marlin menu that starts it and you just follow it from point to point. I'm doing 16 points and it works great. I leveled it once months ago and haven't had to touch it since. 

[Zucca]

Just enable & use manual mesh bed leveling in Marlin. It has a menu selection in the Marlin menu that starts it and you just follow it from point to point. I'm doing 16 points and it works great. I leveled it once months ago and haven't had to touch it since. 

I am using the 3x3=9 mesh with good result too. But I love to explore.
Anyway according to here :

Bed Leveling Methods

Marlin includes various methods of probing and leveling:

    “3-Point” probes a triangle to determine the height and tilt of the bed plane. During printing the nozzle is adjusted in X, Y, and Z, so you can even print on a badly-tilted bed. However, this method requires a very flat and even surface.
    “Linear Grid” probes a square grid (as much as possible on DELTA) to determine the height and tilt of the bed. After that it works just the same way as 3-point leveling.
    “Bilinear Grid” probes a grid in the same manner as Linear Grid, but during printing the Z axis is adjusted according to bilinear interpolation between the measured points. This allows the printer to compensate for an uneven surface,
    “Mesh Bed Leveling” works in the same manner as “Bilinear Grid” but takes different G29 parameters. (This feature is superseded by combining the manual probe option with bilinear leveling, and will not be included in future versions of Marlin.)
    “Unified Bed Leveling” combines elements of bilinear and planar leveling and includes extra utilities to help improve measurement accuracy, especially for deltas. See (link) for an article specifically about this feature.

soo I am moving to AUTO_BED_LEVELING_BILINEAR  Grid with PROBE_MANUALLY, I want to go 4x4 too.

EDIT: I am also doing that because the new board has 512KB of juice!!!! Even if the the spec are 256KB 
that's also the reson why I got that baby... 

[xrunner]

soo I am moving to AUTO_BED_LEVELING_BILINEAR  Grid with PROBE_MANUALLY, I want to go 4x4 too.

I see, well I haven't seen them remove Mesh leveling yet, but yea if they do I'll go with the other procedure of course. 

[mnementh]

Yeah, I'm not talking aboot the marketing wank  ;  I'm wanting to know the nuts & bolts of how it actually works in the course of a print. How it THINKS. How many layers does it take to compensate? How exactly does it calculate compensation?

Obvi, with my CR-6SE I have little choice ATM. I'm just ready to grok a little deeper so I can decide whether to adopt this tech for my "dinosaur" 3DPs.

mnem
 

[xrunner]

Yeah, I'm not talking aboot the marketing wank  ;  I'm wanting to know the nuts & bolts of how it actually works in the course of a print. How it THINKS. How many layers does it take to compensate? How exactly does it calculate compensation?

You mean you didn't GOOGLE?

https://en.wikipedia.org/wiki/Multivariate_interpolation

https://en.wikipedia.org/wiki/Bilinear_interpolation

...

That's a start

[Monkeh]

There is a fade height option, how that's specifically implemented is documented by the code. I don't even know if there's a default value or whether it effectively defaults to off without being explicitly set.

[xrunner]

There is a fade height option, how that's specifically implemented is documented by the code. I don't even know if there's a default value or whether it effectively defaults to off without being explicitly set.

It's default Zero somewhere deeper down because that's what it's always been in my menu, this is all there is about it from configuration.h -

#if ANY(MESH_BED_LEVELING, AUTO_BED_LEVELING_BILINEAR, AUTO_BED_LEVELING_UBL)
  // Gradually reduce leveling correction until a set height is reached,
  // at which point movement will be level to the machine's XY plane.
  // The height can be set with M420 Z<height>
  #define ENABLE_LEVELING_FADE_HEIGHT

There is a set option in the Motion - Bed Leveling menu, mine is set to Zero (apply leveling to all layers). I've never set it to a height, but they seem to imply the leveling can be ignored after a certain height. My machine is printing well with it left at Zero, so why not leave it be. 

[mnementh]

Yeah, I'm not talking aboot the marketing wank  ;  I'm wanting to know the nuts & bolts of how it actually works in the course of a print. How it THINKS. How many layers does it take to compensate? How exactly does it calculate compensation?

You mean you didn't GOOGLE?   https://en.wikipedia.org/wiki/Multivariate_interpolation   https://en.wikipedia.org/wiki/Bilinear_interpolation   ...That's a start

That's how it generates the entire offset table from the 16-point (or however many points) matrix, isn't it? That does NOT tell me how that table gets turned into an actual flat plane over several layers, or even if it does in fact "average out" rather than just apply that table throughout the entire print.

I've looked several times, and all the documentation (especially what's in the code... well, duh ) seems to assume a certain much-higher-than-average knowledge of the programming involved; nowhere could I find any clearly defined (as in basic principles using English rather than math with no clearly defined target) "theory of operation" for the mesh-leveling or whatever it is evolving into now.

I run into this almost any time I try to wrap my brain around some concept in Marlin; I mean, I get that this is mostly volunteer-generated code, and parsing user-level documentation takes time that could be used to refine the code. But having the majority of the documentation buried in the code is a large part of why these things remain perpetually hobbyist-grade... 

So then comes the balance between making a tool that's turn-key ready enough that even management-types can use it (whereupon it will instantly become a commoditized, closed-source system available only for retail) vs leaving the personal, intellectual cost-of-entry high enough to keep the professional grifters out so that "the few who bother" can have a working product at an affordable price through their own efforts.

*sigh*

mnem
"It's always sumthin'." ~Rosanne Rosanna-Danna

[xrunner]

That's how it generates the entire offset table from the 16-point (or however many points) matrix, isn't it? That does NOT tell me how that table gets turned into an actual flat plane over several layers, or even if it does in fact "average out" rather than just apply that table throughout the entire print.

It will interpolate every layer if the Enable fade height is left at Zero. If it is set to a height other than zero it stops interpolating at that height (gradually until it gets there), because it's already manufactured a surface to build upon - the actual model layers.

It interpolates a Z height wherever the X and Y position is at any point in time (from the initial points you measured to begin with). I'm not sure what it is you want to know beyond that. I think you are making it harder than it is. 

Maybe Monkeh can 'splain it better ...

[mnementh]

Are you sure that's what it's saying here? What I see sounds more like a description of different methods of generating the offset table; not explicitly how that table, once generated, is used in each mode to affect the movement of the Z-axis. 

Obviously, how it affects the first layer or three is pretty self-explanatory; but after that, once the first few layers go down and adhesion is achieved, then what? I guess I'm not seeing where it explicitly states that it averages out in small increments to a zero offset, or if it just stops using the offset table, or if it does something else entirely. 

That's what I'm looking for; a brief ELI5 explanation of how each of these modes affects the movement of the z-axis after the offset table is created; not a deep-dive into how the math for that table is derived.

EDIT: Oh, wait... I was looking at the wrong post; let me look into that bit of config.h...

mnem
 

[mnementh]

Oh, fuck me... that section of the FW for my Diggro is buried in layers of ELSE IF conditionals because the Longer LK4 it is based on has an autolevel package available from the factory. 

Every time I poke around in this thing it's like that; it looks like there are 3 or 4 different machines and a dozen variations of screen/UI, printbed, hotend and accessories that are all serviced by this one firmware. 

Okay, yeah it does look like that section does exactly what you suggest. Exactly the way I want to do it. Now all I have to do brave the conditional statement jungle and figure out exactly what is and isn't enabled here.

mnem
*girding my loins*

[xrunner]

Obviously, how it affects the first layer or three is pretty self-explanatory; but after that, once the first few layers go down and adhesion is achieved, then what? I guess I'm not seeing where it explicitly states that it averages out in small increments to a zero offset, or if it just stops using the offset table, or if it does something else entirely. 

Yea after the first few layers you can have it fade-out until it doesn't use any corrections at all. At that point (what ever the fade-out height is you set in mm) it just follows X & Y "machine true" because your rails aren't going to have any imperfections like the bed had. It's past the point it needs to follow the bed. The malleability and "smooshyness" (new tech term) of the PLA enables it to go from an initial phase of following imperfections to a (hopefully) flat plane which is machine true.

I had just left mine with a fade-out of Zero which meant that it used the corrections all the way up. Since my glass bed is really flat I didn't see any issues. But since I have re-thought this I did go to a 3 mm fade-out today. I did check to see if the Z screw did stop small adjustments after 3 mm and sure enough it did. The parts still look just as good to my eye, so I'll leave it at that.