Ed Nisley's Blog: Shop notes, electronics, firmware, machinery, 3D printing, laser cuttery, and curiosities. Contents: 100% human thinking, 0% AI slop.
Category: Science
If you measure something often enough, it becomes science
The collection of LEDs that I’ve been abusing with 100 mA pulses at 20% duty cycle got lined up in parallel, with three LEDs in series, driven from a bench power supply set to limit the current to about 180 mA:
Series-parallel LED test fixture
I sweetened the mix by adding a few other LEDs that had served their time in hell, then took some data by clipping the Tek Hall effect current probe around each of the white wires in turn:
Color
LED1
LED2
LED3
Divisions
Current – mA
Total voltage
Red
1.98
1.98
1.94
4.7
23.5
5.90
Red
1.95
1.95
2.00
3.4
17.0
5.90
Red
1.97
1.97
1.97
5.1
25.5
5.91
Yellow
1.97
1.97
1.96
4.5
22.5
5.90
Yellow
1.97
1.96
1.97
4.6
23.0
5.90
Red
1.98
1.98
1.94
4.6
23.0
5.90
Red
1.95
1.95
2.00
3.3
16.5
5.90
Red (new!)
1.99
1.96
1.94
6.4
32.0
5.89
Despite the decimals, don’t trust anything beyond the first two digits.
The LEDs started out with 6.03 V across them at that current, then settled down to 5.92 V after a few minutes of warming up.
The “new” Red string replaces a trio of old LEDs incinerated by a DVM probe fumble. They have a much higher current at the same voltage; the older LEDs have been abused enough to pass a lower current.
As an experiment, I swapped the LED with the 2.00 V drop in the string with the lowest current (line 7) and the LED with the 1.94 V drop in a string with higher current (line 6), only to find that the current followed the LEDs. Evidently, those LEDs were the limiting factor, even though their forward drops weren’t the same in their new strings.
So it seems binning based on forward drop doesn’t help much. Perhaps just line up a bunch of three-LED strings (forcing all of them to see the same forward drop), measure their current, and reject the highest and lowest strings to get a decent match among the remainder?
A brace of “Fashion” USB video cameras arrived from halfway around the planet. According to the eBay description and the legend around the lens, they’re “5.0 Megapixel”:
Fashion USB camera – case front
The reality, of course, is that for five bucks delivered you get 640×480 VGA resolution at the hardware level and their Windows driver interpolates the other 4.7 megapixels. VGA resolution will be good enough for my simple needs, particularly because the lens has a mechanical focus adjustment; the double-headed arrow symbolizes the focus action.
But the case seemed entirely too bulky and awkward. A few minutes with a #0 Philips screwdriver extracted the actual camera hardware, which turns out to be a double-sided PCB with a lens assembly on the front:
Fashion USB video – case vs camera
The PCB has asymmetric tabs that ensure correct orientation in the case:
Fashion USB camera – wired PCB rear
In order to build an OpenSCAD model for a more compact case, we need the dimensions of that PCB and those tabs…
Start with a picture of the back of the PCB against white paper, taken from a few feet to flatten the perspective:
img_3300 – Camera PCB on white paper
Load it into The GIMP, zoom in, and pull a horizontal guide line down to about the middle of the image:
Camera PCB – horizontal guide – scaled
Rotate to align the two screws horizontally (they need not be centered on the guide, just lined up horizontally):
Camera PCB – rotated to match horizontal guide – scaled
Use the Magic Scissors to select the PCB border (it’s the nearly invisible ragged dotted outline):
Camera PCB – scissors selection – scaled
Flip to Quick Mask mode and clean up the selection as needed:
Camera PCB – quick mask cleanup – scaled
Flip back to normal view, invert the selection (to select the background, not the PCB), and delete the background to isolate the PCB:
Camera PCB – isolated – scaled
Tight-crop the PCB and flatten the image to get a white background:
Camera PCB – isolated – scaled
Fetch some digital graph paper from your favorite online source. The Multi-color (Light Blue / Light Blue / Light Grey) Multi-weight (1.0×0.6×0.3 pt) grid (1 / 2 / 10) works best for me, but do what you like. Get a full Letter / A4 size sheet, because it’ll come in handy for other projects.
Open it up (converting at 300 dpi), turn it into a layer atop the PCB image, use the color-select tool to select the white background between the grid lines, then delete the selection to leave just the grid with transparency:
Camera PCB with grid overlay – unscaled
We want one minor grid square to be 1×1 mm on the PCB image, sooo…
Accurately measure a large feature on the real physical object: 27.2 mm across the tabs
Drag the grid to align a major line with one edge of the PCB
Count the number of minor square across to the other side of the image: 29.5
Scale the grid overlay layer by image/physical size: 1.085 = 29.5/27.2
Drag the grid so it’s neatly centered on the object (or has a major grid intersection somewhere useful)
That produces a calibrated overlay:
Camera PCB with grid overlay
Then it’s just a matter of reading off the coordinates, with each minor grid square representing 1.0 mm in the real world, and writing some OpenSCAD code…
Then wound them with grossly excessive amounts of wire (the up-armored core on the right appeared earlier):
Slit Ferrite Toroid current sensors
The smaller toroid is an FT37-43 that barely covers the active area of an SS49-style Hall effect sensor, but experience with the FT50 toroid suggests that’ll be entirely enough:
slit FT37 toroid trial fit to SS48-style Hall effect sensor
Data on the uncut toroids:
Property
FT50-61
FT37-43
Outer diameter (OD) – inch
0.50
0.375
Inner diameter (ID) – inch
0.281
0.187
Length – inch
0.188
0.125
Cross section area – cm2
0.133
0.133
Mean path length (MPL) – cm
3.02
2.15
Volume – cm3
0.401
0.163
Relative Permeability (μr )
125
850
Saturation flux G @ 10 Oe
2350
2750
Inductance factor (AL) – nH/turn2
68.0
420
Those overstuffed windings improved the sensitivity, but increased the winding resistance far beyond what’s reasonable.
Data on the slit toroids:
Toroid ID
FT50-61
FT37-43
FT50-61
Measured air gap – cm
0.15
0.15
0.17
Winding data
Turns
120
80
25
Wire gauge – AWG
28
32
26
Winding resistance – mΩ
530
920
100
Predicted B field – G/A
872
660
163
Hall effect sensor @ 1.9 mV/G
Predicted output – mV/mA
1.7
1.3
0.31
Actual output – mV/mA
1.9
1.9
0.37
Actual/predicted ratio – %
+12
+46
+19
The last few lines in that table show the transimpedance (transresistance, really, but …) based on the winding current to Hall sensor output voltage ratio (in either mV/mA or V/A, both dimensionally equivalent to ohms), which is why the toroid’s internal magnetic flux doesn’t matter as long as it’s well below saturation.
Gnawing the 80 turn winding off the FT37-43 toroid and rewinding it with 15 turns of 24 AWG wire dropped the winding resistance to 23 mΩ and the transimpedance to 0.36 mV/mA:
FT37-43 with 15 turns 24 AWG – Hall sensor
However, applying a voltage gain of about 28 (after removing the sensor’s VCC/2 bias) will produce a 0-to-5 V output from 500 mA input, which seems reasonable.
Then generate the sphere (well, two spheres, one for each dent) and offset it to scoop out the dent:
for (i=[-1,1]) {
translate([i*(DentSphereRadius + HandleThick/2 - DentDepth),0,StringHeight])
sphere(r=DentSphereRadius);
HandleThick controls exactly what you’d expect. StringHeight sets the location of the hole punched through the handle for a string, which is also the center of the dents.
The spheres have many facets, but only a few show up in the dent. I like the way the model looks, even if the facets don’t come through clearly in the plastic:
Quilting circle template – handle dent closeup – solid model
It Just Works and the exact math produces a better result than by-guess-and-by-gosh positioning.
The sphere radius will come out crazy large for very shallow dents. Here’s the helmet plate for my Bicycle Helmet Mirror Mount, which has an indentation (roughly) matching the curve on the side of my bike helmet:
Helmet mirror mount – plate
Here’s the sphere that makes the dent, at a somewhat different zoom scale:
Helmet mirror mount – plate with sphere
Don’t worry: trust the math, because It Just Works.
You find equations like that in Thomas Glover’s invaluable Pocket Ref. If you don’t have a copy, fix that problem right now; I don’t get a cut from the purchase, but you’ll decide you owe me anyway. Small, unmarked bills. Lots and lots of small unmarked bills…
The humidity in the basement safe started rising this month:
Basement Safe – 2013-07-28
The bag of new silica gel weighed 575 g, so it adsorbed about 67 g of water as the humidity rose from bone dry to 24%. Last month it had soaked up 31 g, so the safe admits nearly an ounce of water each month with 50% RH in the basement. It takes five months to accumulate 60-ish g of water during the winter.
According to the Sorbent Systems charts, silica gel’s equilibrium capacity at 24% is about 12% of the gel’s weight, which would work out to 60 g. That’s close enough, methinks, given the graph resolution; the humidity changes slowly enough that it’s sorta-kinda equilibrated in there… 67 g works out to 13.4% of the dry weight, which is in the same ballpark.
I made up three more bags of dry gel (500 g + 7 or 8 g tare), tossed one in the safe, one in the 6 gallon plastic bucket of 3D printer filament, and one in an empty 6 gallon bucket for comparison. Some 6 dot (10-through-60%) humidity indicator cards are on their way, seeing as how I don’t have nearly enough dataloggers to keep up with the demand…
I put the XY coordinate origin in the middle of the platform, so that laying objects out for printing doesn’t require knowing how large the platform will be: as long as the printer is Big Enough, you (well, I) can print without further attention.
The RepRap world puts the XY coordinate origin in the front left corner of the platform, so that the platform size sets the maximum printable coordinates and all printing happens in Quadrant I. This has the (major, to some folks) advantage of using only positive coordinates, while requiring an offset for each different platform.
Yes, depending on which printer software you use, you can (automagically) center objects on your platform; this is often the only way to find objects created with Trimble (formerly Google) Sketchup. I am a huge fan of knowing exactly what’s going to happen before the printing starts, so I position my solid models exactly where I want them, right from the start. For example, this OpenSCAD model of the bike helmet mirror parts laid out for printing:
Helmet mirror mount – 3D model – Show layout
… exactly matches the plastic on the Thing-O-Matic’s platform, with the XY origin right down the middle of the platform:
Helmet mirror mount on build platform – smaller mirror shaft
It’d print exactly the same, albeit with more space around the edges, on the M2’s platform.
Similarly, the Z axis origin sits exactly on the surface of the platform. That way, the Z axis coordinate equals the actual height of the current thread extrusion in a measurable way: when you set the Z axis to, say, 2.0 mm, you can measure that exact distance between the extruder nozzle and the platform:
Taper gauge below nozzle
Now, admittedly, I fine-tune that distance by measuring the height of the skirt thread around the printed object, but the principle remains: a thread printed on the platform with Z=0.25 should be exactly 0.25 mm thick.
The start.gcode file handles all that:
;-- Slic3r Start G-Code for M2 starts --
; Ed Nisley KE4NZU - 15 April 2013
M140 S[first_layer_bed_temperature] ; start bed heating
G90 ; absolute coordinates
G21 ; millimeters
M83 ; relative extrusion distance
M84 ; disable stepper current
G4 S3 ; allow Z stage to freefall to the floor
G28 X0 ; home X
G92 X-95 ; set origin to 0 = center of plate
G1 X0 F30000 ; origin = clear clamps on Y
G28 Y0 ; home Y
G92 Y-127 ; set origin to 0 = center of plate
G1 Y-125 F30000 ; set up for prime at front edge
G28 Z0 ; home Z
G92 Z1.0 ; set origin to measured z offset
M190 S[first_layer_bed_temperature] ; wait for bed to finish heating
M109 S[first_layer_temperature] ; set extruder temperature and wait
G1 Z0.0 F2000 ; plug extruder on plate
G1 E10 F300 ; prime to get pressure
G1 Z5 F2000 ; rise above blob
G1 X5 Y-122 F30000 ; move away from blob
G1 Z0.0 F2000 ; dab nozzle to remove outer snot
G4 P1 ; pause to clear
G1 Z0.5 F2000 ; clear bed for travel
;-- Slic3r Start G-Code ends --
The wipe sequence, down near the bottom, positions the extruder at the front center edge of the glass plate, waits for it to reach the extrusion temperature, then extrudes 10 mm of filament to build up pressure behind the nozzle. The blob generally hangs over the edge of the platform and usually doesn’t follow the nozzle during the next short move and dab to clear the mess:
M2 – Wipe blobs on glass platform
I’ve also configured Slic3r to extrude at least 25 mm of filament in at least three passes around the object. After that, the extruder pressure has stabilized and the first layer of the object begins properly.
Which brings up another difference: the first layer printed on the platform is exactly like all the others. It’s not smooshed to get better adhesion or overfilled to make the threads stick together:
Robot cookie cutter – printing first layer
I print the first layer at 25 mm/s to give the plastic time to bond to the platform and use hairspray to make PLA stick to glass like it’s glued down.
While pulling together a talk on OpenSCAD modeling (more on this later), I ran off a batch of calibration and “torture test” objects, with the intent of seeing how my somewhat modified M2 performs. The short answer is that you (well, I) can’t ask for anything better…
That level of as-printed cleanliness is typical: no stringing, no hair, no misplaced globs, no retraction problems. Basically, the plastic shape on the platform matches the mathematical shape on screen.
All of the linear features are with ±0.1 mm of nominal; both the 0.5 and 0.25 mm walls came out at 0.40 mm, because that’s the thread width. Slic3r doggedly puts a thread down the middle of hair-fine walls, which I think is a Good Thing.
The holes came out less than 0.3 mm undersize, which is about what you’d expect because they’re not pre-distorted and have far too many sides. The 1.0 and 0.5 mm diameter holes are present, but just barely visible; those simply aren’t reasonable sizes for this technology.
The bottom view shows a few strings in the bridge test area and more detail of the overhang:
M2 – Calibration Block – bottom
Grouping the overhangs like that produced a flat surface that tended to curl upward, so the final slopes don’t match the design. In round numbers, the M2 can handle something like a 60° overhang reasonably well.
The top view shows the shape in the box looks fine, but with some curls in the main structure. The arch closed over a few random strands, so it’s rougher than I’d like:
M2 – 3DHacker object – top
The spires are lumpy and there’s more striation than I’d like, but this lies well outside the realm of stuff that I build. If I were doing it for real, I’d add some support structures here & there.
The Smell of Molten Projects in the Morning 