Resize a STL File For Me

[bostonman]

I'll try printing it the other way too.

Thought having less support material was better, so I flipped it.

[MarkF]

Overall, it sounds like you should really do the printer calibration and give it a good look-over for anything loose.
Especially since you've made a bunch of modifications and updated Marlin (correct?).
You should be able to do everything through the printer control panel.

I just changed a wore out brass extruder gear to a stainless steel one on my CR-10 today. 
That alone changed my eSteps for the stepper motor from 97 to 100.5.

[bostonman]

Overall, it sounds like you should really do the printer calibration and give it a good look-over for anything loose.
Especially since you've made a bunch of modifications and updated Marlin (correct?).
You should be able to do everything through the printer control panel.

Yes, Marlin and a bunch of modifications have been done. The belt tension upgrades could be implemented, but I never got around to it and always questioned their strength. I got the tension fairly tight and it seemed to make a tremendous improvement (but they may have stretched).

Eventually I found myself spending more time upgrading than actually using it; so I began accepting the prints as they were.

Most stuff I've printed wasn't dependent on such resolution, now I'm in uncharted territory. The printer has never been calibrated as I was unaware this was needed. The only calibration was trying my best to get the Z axis level along with bed leveling.

The test jig is a great idea and may provide just how far off my printer is, so let me provide some updates (hopefully) tomorrow.

[Ian.M]

Scaling off the original STL, as rendered by OpenSCAD, I got dimensions of 6.13mm dia, and 4.8mm

Just to clarify, are you referring to the initial one I posted that's too large (30mm ??) If so, then it would explain why it fit (and loosely) since these others were drawn at 6mm.

Yes, the STL from thingiverse attached to your initial post.

Does it make sense to make the hole as a V shape per the factory knob? As I mentioned, I'm guessing the V part allows for less of the surfaces to touch giving more room for error whereas a (half) hole has to fit a round shaft.

No. That is for a light press fit with injection moulded parts, and will have been carefully optimised for the original plastic used and the wall thickness of the shaft boss adjusted to allow the plastic to flex to make it work.   FDM 3D prints have anisotropic strength, usually far lower than injection moulded parts even with similar plastics, so the feature would need to be redesigned and re-optimised to work effectively without sheering off the shaft boss when it flexes.

Its much easier to simply tweak a D hole to fit reasonably well then figure out how to retain it.  Glue it as you proposed, or maybe a 2.5mm cross hole to allow a M3 grub screw to be tightened on the shaft flat.

My printer isn't anything great as some of you may know. How I acquired it is a different story, but I printed many of the associated brackets allowing me to tighten the belts without bending the acrylic (or less bending). The shaft bearings were replaced for smoother movement, belts were tightened (may have stretched now), and I built an overhead filament holder reducing friction as the roll turns.

Similarly, except my I3 clone has a plywood frame, which was a bit better braced than the Anet design.  I can generally get small parts to come out close enough to what I want.   If in an unheated room in winter, some sort of an enclosure is recommended to block drafts and control warping, which can be as simple as a very large cardboard box with a clear plastic curtain to cover the open front.  Use the Marlin menu to turn on the bed heater (only) to warm up  the enclosure well before staring the print.   If you trust its wiring and MOSFET, use its 90° (ABS) bed preheat setting for faster enclosure warmup.   I tend to do so at least 20 minutes before starting a job, then set the correct bed temperature when I open the slicer and give it 5 minutes or so to stabilise at the new temperature while I'm importing the STL and setting it up to slice and print.

On the subject of calibration jigs, Assuming you can run OpenSCAD, change the line:

Code: [Select]

                dshaft(h+2*$dd,shaft_d+oversize,shaft_f+oversize*shaft_f/shaft_d);to:

Code: [Select]

                cylinder(h=h+2*$dd, d=shaft_d+oversize);for round holes you can check against a gauge pin (or any other smooth round rod of accurately known diameter).  Set shaft_d (first line of script) to the actual diameter of the test pin or rod.

Edit: Another way of getting round holes is to simply set shaft_f (second line) to the same value as shaft_d.

[bostonman]

I didn't see your message until tonight; and I already printed everything.

We've definitely established my printer needs some adjustments, but for the fun of it, I printed a few test items. First off, the test jig results are attached. The bright green were loosely fit (fall off the shaft loose if it was a knob), dark green fit tight but nice and snug (held onto the shaft equal to the factory knob), and red it wouldn't fit at all - notice third from the bottom on the left is a bright green.

I printed a square with a 6mm hole in the middle, using calipers, it measures approximately 5.62mm with some slight variations in all directions. Also printed was a 6mm cylinder (approx 7mm height) and measures 6mm and up to 6.18mm as I spin it (implying one belt is looser than the other - or it wobbled due to it's thin structure).

The latest revision knob was printed, without rotating it in the slicing software (made trying to remove the supporting material fun), one was printed at the default 0.3mm layer height (first layer is 0.35mm by default) and the other 0.12mm layer height.

The hole is measuring around 5.7 and 5.8 on both, but varies (it may be because how the calipers are seated in the hole and/or touching filament residue due to supporting material).

In any case, both knobs fit, it's very (and I emphasize that word) tight fit that requires lots of force, but it does fit. I wouldn't use this much force on the new encoders when I get them, so the holes could be sized slightly larger.

The current knob, with the exception of having to apply lots of force to slide onto the shaft, is perfect. The lines on the side, diameter, height, roundness, etc... are essentially equal to the factory one.

My suggestion is to wait until I get the new encoders to confirm they are the correct size/function and then just enlarge the shaft hole again.

If I mess with the printer belts and calibration, then we'll end up starting over trying to get a hole that will fit the shaft. My suggestion is to leave the printer as is for now, print the knobs with the compensating hole size, and then at some point I'll tinker with the printer.

[Ian.M]

The test jig may be indicating between 0.1 mm and 0.15 mm shrinkage on a 6 mm hole ID.  The red no fits near the notched corner are as expected but the cluster of five snug fits after the first loose fit  are not, so something odd is going on there.

The console log of the echo statements was:

Code: [Select]

ECHO: "[0, 0] +0 mm"
ECHO: "[0, 1] +0.1 mm"
ECHO: "[0, 2] +0.2 mm"
ECHO: "[0, 3] +0.3 mm"
ECHO: "[0, 4] +0.4 mm"
ECHO: "[0, 5] +0.5 mm"
ECHO: "[1, 0] +0.05 mm"
ECHO: "[1, 1] +0.15 mm"
ECHO: "[1, 2] +0.25 mm"
ECHO: "[1, 3] +0.35 mm"
ECHO: "[1, 4] +0.45 mm"
ECHO: "[1, 5] +0.55 mm"
which indicates it calculated the oversize for each hole correctly.  Hole id is [x, y] with [0,0] next to the notched corner.

Its suggestive of printer or environmental problems, possibly warping due to loss of bed adhesion at one end, resulting in 'dogleg' holes at the edge of the area that lifted.

[MarkF]

If you are satisfied with everything else about the knobs except the hole size, you may be better off just having the slicer scale the knob by 101% - 105% for a snug fit.
Whichever you choose, I'm perfectly happy to make any adjustments you want.


One point of interest:
   If your Anet stepper motors and lead screws are like Creality CR-10, each step results in a linear movement of 0.04mm.  For example, choosing a layer height of a multiple of 0.04mm will not require any partial steps.  I always choose layer heights of 0.08mm, 0.12mm, 0.16mm or 0.20mm.  I never go beyond 0.20mm.  That's just a personal choice.  My default is 0.2mm unless I want some fine detail.

[bostonman]

I've never changed any settings except infill and maybe some other basic settings.

Maybe the test jig got slightly warped when I removed it. The bed has painters tape for a base and I struggle to get some items from it. I've seen a rubber (?) sheet that is magnetic allowing it to be removed and making part removal easier, but never invested in it.

At some point I need to learn a bit more about 3D printing along with calibrating this printer since I've never done it. Unfortunately I don't do much printing, and, anything I've printed hasn't been affected by being off by a few millimeters.

This knob has certainly opened my eyes to issues with my printer. At the same time, I'm unaware just how accurate the printer should be making parts. My goal about two-years-ago was to install the belt tension adjustments. They don't strike me as very sturdy, but, either way, I thought having the ability to loosen them when the printer isn't being used would reduce belt stretching.

Another issue I felt was that it has so many upgrades available that I was basically putting three to four times into the printer whereas I could just buy a higher end one; so I began accepting prints as they were.

I never expected this knob design to be so involved. It's a power supply that someone hit, took out one of the two knobs (and encoder shaft) along with three banana jacks. It's not a high end power supply, but it was in the trash bin at work. Being able to bring things back to life for not only my benefit, but keeping them out of a landfill, is a nice felling.

[Ian.M]

My first useful print was a 'bumper' for an Arduino Mega.  It took quite a bit of fine tuning of my CTC DIY I3 to get one that fitted.  Back then I was also printing on painters tape, but switched to printing on glass with hairspray as an adhesion promoter, after I'd got fed up with replacing damaged tape every few prints.   Glass isn't for everyone, as to get a good first layer, the nozzle height must be more accurate over the whole part footprint, and any warping of the part typically results in total  adhesion failure and a failed print rather than just lifting a corner and continuing with some distortion the way printing on the more forgiving painters tape typically goes.  However, if you can cut the glass to size yourself, its a very cheap bed upgrade.   

I also had tensioner problems, as CTC had made some very strange design choices, and the D shaft used for the tensioner pully axles was trying to grind its way through the tensioner bodies, but I noticed the slackening belts in time and buying and installing the ball bearings they'd muntzed out  in the recesses still provided for them fixed that.  I generally leave the belts tensioned, as the plywood frame can take the strain, and simply switch on and move the axes every month so the belts don't take a set, but if I was storing it long term I would slacken them.
With an acrylic frame its less advisable to leave them tensioned as the plastic can warp or fail under sustained load, but as you need consistent tension for good calibration, its a PITA to have to frequently slacken them and retension to the same tightness.

Unless your bed is very flat and accurately levelled, having a BLtouch installed, with its probe as near the nozzle as practical, makes a big difference to larger footprint parts.  However as an Anet A8 typically has an ATMEGA1284P controller, getting Marlin built to fit with bed levelling enabled can be a PITA, as it only has half the memory of an ATmega2560.

[bostonman]

The glass and hairspray technique is something used previously. As a result, I have a piece of glass. The current setup is the metal original bed with the glass on top, and painters tape holding down the glass.

As mentioned, the item I saw about two-years-ago was a rubber (?) flexible sheet. if I remember correctly, a strip of magnetized metal gets taped to the bed, and this sheet sticks to it using magnets. After a print is complete, the entire rubber sheet lifts off and you can peel it away from the print, and then reapply the rubber sheet.

It looked quite useful, but at the time painters tape seemed to do the trick.

My Anet has a front and rear bracket mounted to it preventing (or reducing frame bending). If I remember correctly, the design shows to mount these utilizing the threaded rod, however, I have a piece of wood (clear stained) as a base and screwed the printed brackets into the wood. Also, the threaded rod has four brackets holding them down to the wood base as well. For some reason I remember using t-nuts, but, either way, the printer is mounted quite tight to the wood base (maybe 3/8" wood - not plywood - some sort of quality stuff I think). It even has the brackets for the upper part to reduce wobbling.

Basically my printer can be spun on every axis in the middle of a room and (hopefully) not see a shift in the frame - at least this is the image I have due to the amount of effort I put into making it sturdy.

The big issues are bed leveling (I absolutely agree that the BL Touch is a great improvement - I just never got around to adding it) and belt tension. Otherwise, I've taken into account all the mounting, reduced friction everywhere I could, etc...

I could be wrong, but believe due to the lack of features I care about (open thermistors shut down is the first feature I care about), I'm able to fit the BL Touch and standard features into the amount of memory available - you're correct though - it's awfully close to running out of room.

The personal issue I had was (as mentioned) felt I was dumping more time/money into the printer and using it to test upgrades rather than just buying some $2k unit that wouldn't need any (or much) upgrades.

My plan was to get a new PC, expand my interests on the mechanical side (I've taken on more interests doing 2D drawings and 3D designs), and maybe then upgrade to a new printer. My hurtles are that in order to transfer my extensive amount of software that I "obtained" years back when it was easier to "obtain", I'd need to "buy" updated versions. If I use a virtual XP system (or run two PCs) so I can still use my software, then I use old software that I can't use new files in. An example is say Altium. I can't take a new Altium file and have it open in a decade old version (or maybe I can, but just using an example).

Normally I use my work laptop for stuff and my XP to bang around. It's getting more ridiculous though to work around such issues as I just experienced with this knob where I have very limited options on taking a STL file and tweaking it or working with someone who has a CAD file who can share it so I can tweak it. My work laptop is limited on what I can do since I can't install anything and certain sites are blocked, but my XP is limited for obvious reasons. Also, I just dread new computers. Having to work through 30-day trial software, the quirks, installing codecs, etc... Wish XP would have just remained the OS forever.

[themadhippy]

Have you tried to see if either your older machines,or possible the works machine can accesses the likes of tinkercad (https://www.tinkercad.com ) or onshape (https://www.onshape.com)

My hurtles are that in order to transfer my extensive amount of software that I "obtained" years back when it was easier to "obtain", I'd need to "buy" updated versions.

or leave the old machine alone and use it for those tasks,with the new machine dealing with modern stuff

[bostonman]

I'll try later.

FreeCad (link was suggested in a previous post) didn't work with XP.

Years ago I had two computers aside of each other. One was used for old stuff and other stuff while the new one (current) was the one that would only run software I knew was safe. Eventually it got to the point where the old one took forever to boot and wasn't worth keeping.

Obviously options exist to keep my old system such as what I mentioned, a PC (forgot what those are called) that runs within a virtual window. From experience such as having two PCs is that it becomes not worth it anymore.

The answer really is to get a new PC. I should be posting questions about needing help with making a knob rather than having someone do it for me; or people helping by working around my PC limitations.

[Ian.M]

I printed an encoder knob from my reply #19 OpenSCAD script, for a 5.95mm dia, 4.45mm thick D shaft today.  0.05mm oversize wasn't quite enough on my printer, 0.1mm would have been better choice, but a little work with a very small file to smooth out the slight layer ridges in the hole and it went on well enough.

[MarkF]

Seems to me that you need one of these (Not for gaming):     
   https://store.minisforum.com/collections/all-product/products/minisforum-n40-mini-pc


There are older versions of FreeCAD (version 16 lists WinXP):
Quite old.  Don't know how well it works.
   FreeCAD.0.16.6706.f86a4e4-WIN-x86-installer.exe

Plus there are Linux and Mac versions:
I believe the 0.18.4 is the last of the 32-bit builds.
   https://wiki.freecad.org/Download


Oh.  You might want to determine if the knob 6mm hole is the issue or the distance to the flat.
A round ~6mm hole would determine the hole size and the flat adjusted afterward.

[MarkF]

Well, that was an adventure...

I installed FreeCAD 0.16 on my WinXP computer.

After install goto Edit->Preferences and:

• Select Part Design as your default workbench.
  
• Select navigation style to Touchpad.
  This will allow you to:
     Pan (hold Shift Key, no mouse buttons) moving the mouse
     Rotate (hold Alt Key, no mouse buttons) moving the mouse
     Zoom In/Out with mouse wheel
  

Creating the Knob is a totally different procedure than in the latest FreeCAD version.
Here is the knob using verions FreeCAD 0.16. 
The dimensions may be all wrong.  I did it from memory.


Here are the dimensions for the last non-FreeCAD_0.16 knob:



Edit:  Updated CAD file with latest dimensions.

[Infraviolet]

I note you mention the knob is a bit loose, it's because 3D printed D shaft holes just aren't very good. The plastic is soft enough that in the really small volume of plastic wich experiences all the forces when torques are applied between the D shaft the the attached plastic thing, the force tends to distort the plastic. I'd always recommend trying to insert some nut slots and screw holes so you can use a screw to clamp against the flat side of the shaft.

This video:

shows how a screw and a nut can be used on a D shaft's flat, though if you go in from both sides 180 degrees apart with 2 screws you often do better as one screw has its end clamped against the flat and one has its end clamped against the curved side of the shaft opposite the flat. This should be a fairly simple geometry to integrate in to your knob. You can have the nuts slotted in from the bottom face, so the top one sees will still look smotth. The only visual difference will be two holes, of diameter sufficient for the screw's heads (M2.5 or M3 will do fine for low torque situations) coming in from the sides.

[Ian.M]

If you are going to add a grubscrew, or if you need to add captive nuts in general to 3D printed parts, but don't want to mess with heat-staked inserts, do yourself a favour and use a square nut.   
e.g. https://www.mcmaster.com/97259A101/
Compared to a hex nut, the fit is far less critical, they are far less likely to spin in their hole. and have a greater surface area to transfer end-load to the part.

[bostonman]

I don't have much experience making 3D parts, but for me (and some of you from previous posts know I tend to go overboard), I think utilizing metal in certain applications is beneficial.

In the case of the shaft, make the 3D part so it can hold a metal insert (maybe brass) with a few spline lines (?) internal, drill a hole in the metal, thread it with a die, use two-part epoxy to keep the metal from slipping (the epoxy will go inside the spline lines for extra strength), and now you have a much stronger piece for a set screw.

My other thought is to make a part, if the software allows it, inverse the design so it's a mold, print it, and then use something like JB Weld to fill the mold. Now you have quite the heavy duty part.

Most likely printing with other material makes for much stronger stuff - but I only have experience with PLA.

[MarkF]

It's possible you may want to take the knob off or need to repair the power supply.
If the knob fails or starts coming slipping off, you can always print another or change its dimensions.
Maybe you want to change the color to match your outfit of the day... 

[bostonman]

It's possible you may want to take the knob off or need to repair the power supply.
If the knob fails or starts coming slipping off, you can always print another or change its dimensions.
Maybe you want to change the color to match your outfit of the day... 

Were you referencing my previous message? If so, I was referring to the previous message about using a set screw for the design used in the posted video. In the case of my power supply, the existing design will be fine.

[bostonman]

Well, that was an adventure...

I installed FreeCAD 0.16 on my WinXP computer.

Did you get this error: DLL load failed %1 is not a valid Win32 application

[MarkF]

Well, that was an adventure...

I installed FreeCAD 0.16 on my WinXP computer.

Did you get this error: DLL load failed %1 is not a valid Win32 application

No. I didn't get any errors installing version 0.16.
Are you sure you downloaded the 32-bit version?
I downloaded the x86-installer.

[MarkF]

It also ask to install the Visual Studio libraries. Accept and let it do them

[Ian.M]

Revised knob with grub screw hole and slot for a square nut.  Still based on resizing the original Knob_D_shape_rotary_encoder.stl from the first post.

Apart from some parameters for the nut, it only needs a few lines of code added in the cuts section of the difference().

Code: [Select]

    translate([0,knob_d/4,knob_h-shaft_l/2-recess_h/2])
        rotate([90,0,0])
        cylinder(h=knob_d/2, d=nut_size+slop, center=true); //grub screw
    translate([0, nut_thick/2+nut_to_shaft+shaft_f-shaft_d/2, 3*knob_h/2-shaft_l/2-recess_h/2-nut_af/2-slop/2])
        cube([nut_af+slop,nut_thick+slop,knob_h],center=true);  //nut slot


[bostonman]

No. I didn't get any errors installing version 0.16.
Are you sure you downloaded the 32-bit version?
I downloaded the x86-installer.

Sorry...Realized after I went to bed that I didn't specify the error occurred after installing, when I loaded FreeCad.

The link had several files, and I downloaded them all (maybe 12 or so). From what I could tell, two .exe files existed that both needed to be installed.

I'll have to do more investigating tonight. From what I could tell, it needs or depends on Python and/or VS stuff. I'm guessing something isn't linked correctly or I didn't install some library.