Tag: Improvements

Making the world a better place, one piece at a time

Windows 11 for the Linux Guy: Remote Desktop Protocol

With Windows 11 and LightBurn running on a little PC perched atop the laser in the basement:

BeeLink Win 11 PC – overview

I wanted to work with that desktop from my Comfy Chair upstairs, because I’m unwilling to stand up a Windows box specifically for another LightBurn installation, along with a nightmare KVM switch tangle for all the displays / keyboards / trackballs I run with Linux.

At the Win 11 PC, turn on Remote Desktop connections:

Remote Desktop enable

The Administrator is automatically allowed access, but I also allowed access for my local User (who does not have a Microsoft account), which requires the Administrator’s password. You’ll want to store that in a password manager, because typing line noise gets tedious.

Upstairs on the Comfy Chair at the Linux box, install Remmina from the repository, then tweak some preferences:

Remmina prefs – General

This being a LAN connection, pick the highest quality scaling, although that shouldn’t matter with a fullscreen display. I added a screen resolution matching my desktop landscape monitor:

Remmina config – screen resolutions

Somewhat to my surprise, selecting an RDP screen resolution larger than the HDMI monitor on the Win 11 box worked perfectly.

Because the remote display will fill the entire screen in fullscreen mode, set the toolbar to “Peeking” mode making it barely visible at the top of the screen:

Remmina prefs – Appearance

I have yet to (figure out how to) enable the hotkey turning fullscreen mode on and off, so if the toolbar isn’t readily available there is no way to get out of fullscreen mode.

Set up the RDP connection to the Win 11 box, using either the static IP address or whatever name the router assigns:

Remmina config – Basic

I set the Win 11 box for a static IP address, then told the router to assign that IP to the box if it ever woke up asking for an address through a DHCP request. The process differs depending on which router you have and may not be needed. I (try to) nail down all the IP addresses, so anything using DHCP will be obviously in need of attention.

Select the highest quality compression:

Remmina config – Advanced

With all that set up, double-clicking the appropriate line should fire up an RDP connection, perhaps with a peephole view of the Win 11 desktop:

Remmina – small RDP window

Hit the Toggle Fullscreen icon (hollow square, fifth down) to embiggen it:

Remmina – fullscreen RDP window

Your Win 11 desktop will be different; that’s the Apollo 17 lunar module ascent stage.

The thin line along the center top is the Remmina toolbar, peeking over the edge. Move the mouse cursor up there to roll it down into view:

Remmina – fullscreen RDP window – detail

Because this is a fullscreen view, hitting the Toggle Fullscreen icon (highlighted blue) is the only way out. It required a disturbing number of iterations before realizing none of the hotkeys worked, then figuring out how to enable toolbar peeking.

Moving the mouse pointer to the bottom of the screen rolls up the Win 11 Task Bar (which I always set to Hide mode to get it out of the way):

Remmina – fullscreen – task bar

I pinned the LightBurn icon to the task bar where it’s easy to hit, as that’s the whole point of the exercise.

And then It Just Works™:

Remmina – fullscreen LightBurn

Because this is Windows, one user can sign onto the box from either the local keyboard or the RDP connection, but not both.

Being an Old School type of guy, I reflexively save my work before trotting either upstairs or downstairs and signing on wherever I end up, but it’s the same file in the same program on the same hardware.

The performance over the LAN and through Remmina is good enough to make the fullscreen session feels exactly like running LightBurn locally. In truth, LightBurn is not a particularly resource-heavy program.

Then I deleted both Linux installations from the LightBurn license portal …

2024-12-10
Vole Traps: End of Season

A highly effective way to bait a rat trap for garden voles:

Rat trap – still baited

The trap is a Victor M205 (in a 12-pack as M326) with a big yellow plastic bait pedal. The bait is pieces of walnut, secured to the pedal with generous strands of hot melt glue. The trick involves mechanically capturing the walnut by slobbering glue over & around it, forcing the vole to pull & tug while gnawing the last bit of goodness.

Which generally ends badly:

Rat trap – gnawed bait

I do not begrudge the critters a fancy last meal; it’s gotta be better than their usual diet of carrots / radishes / turnips.

Voles have no qualms about eating the bait from a sprung trap with a dead compadre a few millimeters away:

Rat trap – empty bait

They will sometimes eat the walnuts and their dead compadre.

The plastic pedals work much better than the old-style metal pedals at holding the steel arm wire. The wire slides freely on the plastic, in contrast to the previous high-friction metal-on-metal latch.

Some of the traps were entirely too sensitive and required slightly bending the tip of the arm wire upward to increase the friction on the plastic plate. Always always always handle armed traps by the wooden edges beside the kill bar, so when it accidentally snaps your fingers are nowhere near the business end.

After I figured out how to properly bait the traps and we set out half a dozen traps in the most attractive crops, Mary’s garden produced 54 dead voles over the course of 90 days, sometimes in groups of three or four at a time. While this did not prevent all the crop damage, it definitely reduced the problem.

Next year we’ll start early and probably reach triple digits by midsummer.

The same technique with Victor M035 mouse traps (in 12-packs as M035-12) is brutally effective on house mice.

2024-12-06

Handi-Quilter HQ Sixteen: Front Handlebar Angled Mount

So as to not bury the lede, I remounted the front handlebar unit of Mary’s Handi-Quilter HQ Sixteen long-arm sewing machine so she can see the control panel with its small LCD:

HQ Sixteen - remounted handlebars in use — HQ Sixteen – remounted handlebars in use

The new and old white LEDs produce distinctly different colors and intensities on the practice quilt fabric.

The original HQ Sixteen design bolted squarely atop the arm:

HQ Sixteen - original front handlebar mount — HQ Sixteen – original front handlebar mount

The control surface is, admittedly, angled slightly forward, but Mary was unable to see the lower few lines of the LCD without standing on tiptoe.

Begin with a crude tracing of the mating surfaces:

Import the image into Inkscape and lay some shapes on it:

Front handlebar base layout - Inkscape — Front handlebar base layout – Inkscape

Import the SVG into LightBurn and cut templates to verify the hole positions:

HQ Sixteen - handlebar bolt templates — HQ Sixteen – handlebar bolt templates

Obviously that took more than one try.

Rationalize the outlines, clean things up, and organize the shapes into useful named layers:

Front handlebar base layout - Inkscape layers — Front handlebar base layout – Inkscape layers

Save as an Inkscape SVG, import into OpenSCAD, and extrude the layers defining all those shapes into a solid model:

Handlebar Base Mount - solid model — Handlebar Base Mount – solid model

That’s the most recent iteration; earlier ones appear in various pix.

I had intended to use either square nuts or heat-set inserts, but it turned out to be easier to just slam BOSL2 threaded nuts into the front plate and be done with it:

Handlebar Base Mount - solid model - hex nuts — Handlebar Base Mount – solid model – hex nuts

The trick is to sink the nuts around a hole sized slightly larger than the screw’s nominal diameter, letting the threads fill empty space.

The handlebar base is mounted symmetrically along the machine arm centerline aligned with the two screws on the right. The rear block is offset to the left to clear the machine cover on the right, so the hull() wrapped around the two looks weird.

The front plate stands proud of the rest by dint of incorporating only a small slice of its back face into the hull() filling the gaps between the two. It’s not particularly stylin’, but it’s pretty close.

Finding the correct angle for the front plate required a couple of iterations, but they all built successfully:

HQ Sixteen - handlebar mount - on platform — HQ Sixteen – handlebar mount – on platform

Putting the threaded holes vertical created nicely formed threads that accepted the screws without hassle.

The block screws firmly to the arm and the handlebar unit screws to the block:

HQ Sixteen - remounted handlebars - side — HQ Sixteen – remounted handlebars – side

The display now faces front:

HQ Sixteen - remounted handlebars - front — HQ Sixteen – remounted handlebars – front

I eventually replaced those black oxide screws with shiny stainless ones, just for pretty.

The nine LEDs under the display now do a great job of lighting up the front of the machine’s arm, rather than the fabric at the needle, but fixing that will be a whole ‘nother project.

The handlebar grips with their control buttons now tilt at a somewhat inconvenient angle, which is also a whole ‘nother project.

Early reports from the user community are overwhelmingly positive.

The OpenSCAD source code and the SVG layout as a GitHub Gist:

Sorry, something went wrong. Reload?

Sorry, we cannot display this file.

Sorry, this file is invalid so it cannot be displayed.

view raw Front Handlebar Layout.svg hosted with ❤ by GitHub

	// Handiquilter HQ Sixteen front handlebar base mount
	// Ed Nisley – KE4ZNU
	// 2024-11-22

	include <BOSL2/std.scad>
	include <BOSL2/threading.scad>

	Layout = "Show"; // [Build,Show,Block,Holes]

	HandlebarOffset = [0,-30.0,14.0]; // pure empirical values
	HandlebarAngle = [60,0,0];

	FrameBlockThick = 35.0; // how much meat they need
	HandlebarThick = 12.0;

	/* [Hidden] */

	Holes = [[-19.0,0,0],[0,0,0],[0,12.5,0]]; // Must match SVG hole coordinates

	FrameCenter = [-45,-65]; // coordinates of corner hole center
	HoleCenter = [-40,-20];


	Protrusion = 0.1;

	module AdapterBlock() {

	union() {
	hull() {
	linear_extrude(height=FrameBlockThick,convexity=10)
	translate(FrameCenter)
	import("Front Handlebar Layout.svg",layer="Machine Frame");

	translate(HandlebarOffset)
	rotate(HandlebarAngle)
	linear_extrude(height=0.05*HandlebarThick,convexity=10)
	translate(HoleCenter)
	import("Front Handlebar Layout.svg",layer="Handlebar Base");
	}

	translate(HandlebarOffset)
	rotate(HandlebarAngle)
	linear_extrude(height=HandlebarThick,convexity=10)
	translate(HoleCenter)
	import("Front Handlebar Layout.svg",layer="Handlebar Base");
	}
	}

	module AdapterHoles() {

	linear_extrude(height=FrameBlockThick,convexity=10)
	translate(FrameCenter)
	import("Front Handlebar Layout.svg",layer="Machine Holes",convexity=2);

	translate([0,0,FrameBlockThick – 7.0])
	linear_extrude(height=7.0 + Protrusion,convexity=10)
	translate(FrameCenter)
	import("Front Handlebar Layout.svg",layer="Machine Counterbore",convexity=2);

	translate(HandlebarOffset) // cut clearance for nut threads
	rotate(HandlebarAngle)
	linear_extrude(height=HandlebarThick + Protrusion,convexity=10)
	translate(HoleCenter)
	import("Front Handlebar Layout.svg",layer="Handlebar Holes",convexity=2);

	}

	module Adapter() {

	union() {
	difference() {
	AdapterBlock();
	AdapterHoles();
	}

	# translate(HandlebarOffset) // add threads inside holes
	for (c = Holes)
	rotate(HandlebarAngle)
	translate(c)
	threaded_nut(10.0,6.2,HandlebarThick,1.0, // flat size, root dia, height, pitch
	bevel=false,ibevel=false,anchor=BOTTOM);
	}
	}

	// Build things

	if (Layout == "Block")
	AdapterBlock();

	if (Layout == "Holes")
	# AdapterHoles();

	if (Layout == "Show")
	Adapter();

	if (Layout == "Build")
	rotate([180,0,0] – HandlebarAngle)
	Adapter();

view raw Handlebar Base Mount.scad hosted with ❤ by GitHub

2024-12-02

Prusa MK4 Input Shaper vs. Resonance Test Box

Although the laser ramp test fixture looked good, Brent wondered what a real test box would reveal about the Prusa MK4’s Input Shaper resonance control.

Loading the STL into PrusaSlicer, adding a text label to remind me which way it printed, then slicing with my PETG-CF profile shows the “Actual Speed”, which seems to take acceleration into consideration:

PrusaSlicer preview – actual speed

The colors in the legend don’t quite match the colors on the model, but the greenish layers with the jolts trundle along in the mid-20 mm/s range and the blue-ish straight-through layers at 30-ish mm/s.

Eryone PETG-CF has a somewhat fuzzy appearance that seems not characteristic of other brands, so I’ll try something else when these spools run out:

MK4 Resonance Test Box – overview

The right side of the box (as oriented on the platform) got all the layer retractions and came out festooned with PETG hairs:

MK4 Resonance Test Box – right side

You can check my labels by tracking the small retraction zit sticking up from the top layer; I got it wrong the first time. Open the images in a new tab to see more pixels.

The front:

MK4 Resonance Test Box – front side

The left:

MK4 Resonance Test Box – left side

And the rear:

MK4 Resonance Test Box – rear side

You can barely see the shadow of the “Rear” text on the surface, even though the wall is two threads thick and the text is indented by 0.2 mm, about half the thread width.

As far as I can tell, the MK4 Input Shaper compensation does a great job of suppressing resonance or wobble in all directions.

Looks good to me!

2024-11-29
White LED QC Escape

Judging from the dates codes on the ICs inside, Mary’s HandiQuilter Sixteen long-arm machine is about two decades old and many of the white LEDs in the front handlebars have gone dark:

HQ Sixteen – dead handlebar LEDs

The vertiginous view looks upward into the handlebar at the top of the machine (more on this later). The PCBs run strings of three series LEDs from a 16 VDC supply with a 390 Ω ballast resistor (oddly enough, on the ground end of the string), so one failed LED takes down all three.

I decided to replace all the LEDs, on the principle they’re surely dimmer than they used to be and to take advantage of a decade or so of improvement in white LEDs (yes, I have old stock).

After discovering that the HandiQuilter engineers violated the Principle of Least Surprise by orienting adjacent LED strings in opposite directions, I found one of the strings still didn’t light up.

Pop quiz: which one of these LEDs caused the problem?

5 mm LEDs – swapped polarity

To the best of my knowledge, all 5 mm round LED packages mark the cathode lead with a flat edge. It’s easy to remember, as the cathode side of the schematic symbol has a bar: straight bar = straight edge.

Inside, the LED chip’s cathode lead is bonded to the reflective cup, with the anode lead wire-bonded to the top.

Took me a while to see what was wrong, too.

For whatever it’s worth, the backward LED works fine.

2024-11-28
Quilting Ruler Rack: Expedient Base

Mary is at least the third owner of a steel rack, originally intended to hold packages of retail stuff, which now holds (much of) her collection of quilting rulers:

Quilting Ruler Rack Base – overview

Obviously, it was never intended to hold heavy acrylic sheets, but it worked surprisingly well, right up to the point where too many of the rulers collected on two adjacent columns of pegs and overbalanced the whole affair atop her while she attempted to remove a ruler.

Subsequent accident recreation showed the rack toppled when the weight of the rulers on the two adjacent columns of hooks moved the center of mass outward, just inside the line between those feet, whereupon the slightest tug on a ruler pulled it over.

Measurements revealed the four legs do not sit on a square contact patch, are not parallel to the radii from the center point, and are not uniformly distant from the center. Rather than committing to a finished product, I made a cardboard prototype to verify a bigger base would solve the problem and I could capture all those feet.

You don’t have such a rack, so the exact dimensions don’t matter, but the LightBurn layout looks like this:

Quilting Ruler Rack Base

The disk is two cross-laid sheets for stiffness, with marks burned on the top to help align the feet more-or-less around the center point.

The oblong rings fit around the feet to capture them, so cut eight or twelve to make four stacks a bit taller than the wire diameter.

The H shape then glues atop the rings to hold the feet in place. They’re not removable, but a razor knife will eventually solve that problem.

I slobbered hot melt glue across the cardboard disks to hold them together, glued and aligned the rings where the feet dented the disks, stood the rack in the rings, and glued the H plates.

About an hour elapsed from the sound of the crash to the rack once again standing quietly beside the fabric cabinets.

We’ll run this for a while and eventually replace it with a plywood disk and screwed-in-place clamps for the feet, which will surely call for wood surface preparation / stain / seal treatment.

2024-11-27
Prusa MK4 Input Shaper: Accelerometer Tuneup
After adding bling to the Prusa MK4, I touched up the belt tensions and re-measured the axis resonances with the Prusa Accelerometer gadget to update the Input Shaper settings.

The Prusa belt tension guide pretty much explains that subject, with their Belt Tuner making up for my utter tone deafness. FWIW, if the Belt Tuner produces inconsistent results differing by an octave, either up or down from the correct value, the belt is way too loose: give the axis belt tension screw a turn or two to drag the results into the right time zone, then fine-tune from there.

While it is possible to reach both tensioning screws without too much trouble, they’re definitely not convenient.

The accelerometer fits on the hot end:

Prusa MK4 Accelerometer – on hot end

Then under the steel sheet, where it’s clamped by the platform magnets:

Prusa MK4 Accelerometer – on platform

The MK4 firmware measures the resonant frequencies while prompting you to put the accelerometer in the proper locations, then computes the best shaper values.

For reference, the stock OEM values:
- X = MZV 50 Hz
- Y = MZV 40 Hz
Just after I got the accelerometer and without doing anything to prep the MK4, these results popped out:
- X = MZV 56 Hz
- Y = MZV 42 Hz
Now, with bling and properly tensioned belts:
- X = MZV 59 Hz
- Y = MZV 45 Hz
The most recent values were also the most stable, once again pointing out the value of careful assembly and maintenance.

With that in mind, though, I built the laser ramp focus fixture shortly after doing the first recalibration and it has no visible ripples on any of its walls:

Ramp Test Fixture – corner detail

That’s a square corner perpendicular to the sloped top surface at the default 45 mm/s. It’s not as difficult a test as some you’ll see, but it suffices for my simple needs. The MK4 definitely behaves better around corners than the Makergear M2.
2024-11-26