Posts Tagged Sewing
A bag of G4 COB LEDs arrived from halfway around the planet:
Those are “5 W” and “4 W” cool white modules, respectively, with another set of 4 W warm white looking pretty much the same. There’s no provision for heatsinking, which makes the wattage seem suspect; halogen G4 bulbs run around 20 W, for whatever that’s worth.
The silicone overlay becomes nearly transparent when seen through an ordinary desktop document scanner:
Highlighting the PCB copper pours shows 18 LEDs arranged in three series groups of six LEDs in parallel:
The “smart IC” touted in the writeup turns out to be a bridge rectifier for AC or DC power:
The SMD resistors on all 15 modules measure 27.6 Ω, more or less, and seem randomly oriented face-up or face-down. I assume that one is face-down; maybe it’s just unlabeled on both sides.
Back of the envelope: there’s no way it will dissipate 5 W. The bridge drops 1.4 V = 2×0.7, the LEDs drop maybe 9 V, leaving the resistor with 1.6 V to pass all of 60 mA, so call it 700 mW.
With 12 VDC applied to the pins, the bridge drops 1.6 V, the LEDs 8.2 V, and the resistor 2.2 V, with 80 mA through the whole affair dissipating just under 1 W.
Cranking the supply until the current hits 200 mA puts 15.7 V across the pins for a total dissipation of 3.1 W, burning 1.7 W in the LEDs and 1.1 W in the resistor.
Cranking the supply to 21.3 V drives 410 mA, dissipates just under 9 W total, produces a curl of rosin smoke from the PCB, and maybe delaminates the silicone around some of the LEDs.
OK, now I have a crash test dummy.
Given complete control over the application, I’ll strip everything off the PCB and bond it to a heatsink of some sort. With 6 LEDs in parallel, 120 mA (6 × 20 mA) total current might be reasonable and 200 mA (6 × 30 mA) probably won’t kill the things outright. Plus, I have spares.
An external 18 Ω resistor should suffice. Perhaps a pair of 6 Ω SMD resistors on the PCB, with fine-tuning through an external resistor. Call it 250 mW apiece: don’t use little bitty SMD resistors.
With the Juki TL-2010Q all lit up, it seemed reasonable to apply the same technique to the Kenmore 158 sewing machine a few feet away:
In an ideal world, I’d match the COB LED module to the opening under the machine’s arm, but module length isn’t a free variable, so it sticks out a bit on both sides.
They run from a 12 VDC 18 W power supply with an adjustable boost converter producing 18 V for the nominally 21 V LEDs:
I replaced the coaxial power plug with a DE-9 connector:
Thpse 1/4 inch QD connectors on the AC power are marginally OK in this situation, as they’re tucked under the sewing table out of harm’s way. The other end of the AC line cord burrows into the sewing machine’s guts and isn’t easily removed, so this was the least-awful place for a connection.
The LED connector pinout:
The black cable comes from my lifetime supply of lovely supple flexible 28-ish AWG 9-conductor serial cables with molded-on male connectors.
I used some silver-plated / Teflon-insulated coaxial cable for the COB LED wiring. It burrows into the guts of the machine through a gap above the presser foot lift lever, then joins up with similar cables from the other LEDs routed through the (grossly oversized) heatsink fins:
The cables meet the repurposed serial cable inside the arm, following the original route of the 120 VAC wires formerly lighting the glowworm incandescent bulb in the endcap:
What’s not obvious in that picture: the cables pass under two stamped steel guides and through two stamped steel clamps, each secured to the frame by a cheese head screw in a tapped hole. They definitely don’t make ’em like they used to!
A 2.0 Ω ballast resistor produced the right amount of light, dropping 780 mV to run the LEDs at 390 mA and burning 300 mW. This supply produces 12.0 V at that current, so the COB LEDs run at 11.2 V and dissipate only 4.4 W.
The lower output voltage (compared to the supply on the Juki) is probably the result of the higher load from the SMD LEDs lighting up the area around the needle. We cranked up their voltage to match the COB LEDs, so they’re surely conducting more than the original (guesstimated) 50 mA apiece = 300 mA total. I have no convenient (pronounced “easy”) way to measure either their current or voltage; when the light’s good, it’s all good.
The other Kenmore 158 machines will eventually get the same treatment, but not right now.
The COB LED module claims to run at 12 V and 6 W, so it expects to draw 500 mA. First pass measurements showed 500 mA happened at 11.6 V:
The 12 VDC supply actually produced 12.1 V at 500 mA, so a 1 Ω 1/2 W resistor should produce the right current:
Which it did, but the Customer Base judged 6 W to be far too much light. A 2.7 Ω resistor seemed too dim, so we settled on 2.2 Ω:
For the record, a 2.2 Ω resistor drops 980 mV and dissipates 440 mW, probably too close to its 500 mW rating. The supply produces 12.2 VDC at 450 mA, so the LEDs run at 11.2 V and dissipate 5 W; the heatsink remains pleasantly warm to the touch.
The hot melt glue anchoring the pin header won’t win any prizes, but it sticks like glue to the Kapton tape and, in any event, there’s not much to go wrong in there.
A cardboard cover hides the ugly details:
And then It Just Works™:
As evidenced by the glove fingertips, she does a lot of sewing and I’m glad I can shed some light on the subject …
The wires to my earlier LED lights on Mary’s Kenmore 158 produced one absolute requirement: the Juki TL-2010Q lights must not have any external wiring. Some experimentation showed putting the COB LED module across the rear of the arm, just over the opening, would spill enough light to the front:
Juki’s teeny OEM SMD LED in the endcap, just above the far side of the needle, casts a dim glow over her left hand. Although they deem it sufficient, I’ll fix that in the near future.
The machine’s power supply and drive motor live inside a plastic cover on the rear of the machine, just to the left of where the LED lights will attach to the arm:
For future reference, a detailed look at the PCB:
The yellow-and-blue pair come from the AC power line switch. The brown-and-blue pair carry +120 VDC from the bridge rectifier (left of their connector) to the motor driver. The white-and-blue pair carry filtered 120 VAC from the PCB to the bulky transformer below the motor.
I snipped the white-and-blue pair, added Y connections, and threaded the wires through the vent slots to the 12 VDC power supply:
If I had to do it again, I’d cut the white-and-blue pair an inch further away from the transformer, so as to move the butt splice connectors around the corner of the frame, rather than across the back of the transformer frame. The flanged screw boss pretty well fills the space left of the transformer and made it difficult to arrange the new connectors.
The 12 VDC 18 W LED supply attaches to the 120 VAC lines with 1/4 inch quick-disconnects, making it possible, if not easy, to completely remove the cover and LED power supply. You’d install dummy plugs in the vacant QD sockets to keep the AC out of harm’s way.
There’s just enough space to the right of the PCB enclosure to route the LED wires around-and-down to meet the wire nuts. They’re not the most elegant connectors you’ve ever seen, but wire nuts are impossible to confuse with the QD connectors on the AC line.
With that in hand, the power supply almost looks like it grew under the spool flange:
In an ideal world, the label would be right-side-up, but ya can’t have everything. The wires had to be where they are, primarily to avoid snagging on fabric passing through the machine.
The green-and-black PET braid covers the AC wires to make them a little less exposed, but it’s surely unnecessary. I gently singed the braid ends to prevent unraveling.
The COB LED supply wires emerge through a slot filed in the cover:
Next step: LED brightness tweakage.
Some years ago, I put the LED power supply for one of the Kenmore 158 machines atop a plastic project box with an adjustable boost supply inside:
The LEDs connected through a coaxial power jack on the far side of the box, held in place with a generous blob of epoxy:
A closer look:
I’m adding a light bar, similar to the one now going onto the Juki TL-2010Q, which needs a direct connection to the 12 VDC supply. Rather than add another coaxial jack, I ripped out the existing jack and installed a DE-9 connector (serial ports being a fading memory by now), giving me an opportunity to test the epoxy joint:
Which required grabbing the connector with a pair of pliers and twisting / bending / abusing until it popped free. I don’t know how much grip the scored lines added to the joint, but the connector definitely didn’t give up without a fight; it wasn’t going to fall off on its own.
To be fair, the epoxy had a better grip on the coaxial jack than on the plastic plate, perhaps because the bottom of the jack had all manner of nooks and pins intended for PCB mounting. Ya use what ya got, sez I.
The new connector looks exactly like it should and, because it’s held in place by a pair of screws, should last forever, too:
More about all that, later …
Because the COB LEDs dissipate 6W, far more power than I’m comfortable dumping into a 3D printed structure, I redefined a length of aluminum shelf bracket extrusion to be a heatsink and epoxied the module’s aluminum back plate thereto:
Unlike the flexible LED strips, the COB LED modules have no internal ballast resistors and expect to run from a constant-current supply. Some preliminary testing showed we’d want less than the maximum possible light output, so a constant-voltage supply and a few ohms of ballast would suffice:
With all that in hand, the heatsink extrusion cried out for smooth endcaps to control the wires and prevent snagging:
The central hole in the left cap passes 24 AWG silicone wires from the power supply, with 28 AWG silicone wires snaking down through the L-shaped rectangular cutouts along the extrusion to the LED module’s solder pads.
The model includes built-in support:
Assuming the curved ends didn’t need support / anchors holding them down turned out to be completely incorrect:
Fortunately, those delicate potato chips lived to tell the tale and, after a few design iterations, everything came out right:
The “connector”, such as it is, serves to make the light bar testable / removable and the ballast resistor tweakable, without going nuts over the details. The left side is an ordinary pin header strip held in place with hot melt glue atop the obligatory Kapton tape, because the heatsink doesn’t get hot enough to bother the glue. The right side is a pair of two-pin header sockets, also intended for PCB use. The incoming power connects to one set and the ballast resistor to the other, thusly:
The diagram is flipped top-to-bottom from the picture, but you get the idea. Quick, easy, durable, and butt-ugly, I’d say.
The next step was to mount it on the sewing machine and steal some power, but that’s a story for another day.
The relevant dimensions for the aluminum extrusion:
The OpenSCAD source code as a GitHub Gist:
As part of converting the halogen desk lamp to LEDs, I replaced the hulking iron transformer with a flatter counterweight:
Under normal circumstances, you’d use something like steel or lead sheets, but Tiny Bandsaw™ can’t cut any appreciable thickness of steel and I gave away my entire lead stockpile, so I sawed disks from a pile of non-stick pancake griddles and drilled suitable mounting holes:
Another disk (from a formal aluminum sheet!) goes into the lamp head, with a trio of 3W COB LEDs epoxied in place:
The other side of the disk sports a heatsink harvested from a PC, also epoxied in place:
Realizing the head required only a little filing to accommodate the heatsink sealed both their fates.
A test firing showed the heatsink needed more airflow, which didn’t come as much of a surprise, so I milled slots in the lamp head:
Deburring the holes, blackening the sides with a Sharpie, and tucking a bit of black window screen behind the opening made the vents look entirely professional.
The small dome in the base originally cleared the transformer and now holds the entire 10 W LED driver, along with all the wiring, atop the counterweight sheets:
A cork pad covers the base for a bit of non-skid action:
I couldn’t convince myself filling in those sectors would improve anything, so I didn’t.
And then It Just Worked:
All without a trace of solid modeling or G-Code …