The Smell of Molten Projects in the Morning

Ed Nisley's Blog: Shop notes, electronics, firmware, machinery, 3D printing, laser cuttery, and curiosities. Contents: 100% human thinking, 0% AI slop.

Category: Electronics Workbench

Electrical & Electronic gadgets

Samsung Dishwasher Drying Fan Replacement

The Samsung dishwasher (model DW80K7050US/AA 03) that Came With The House fails immediately after entering the Dry part of the cycle: a relay in the control PCB under the door goes doink, all the LEDS go off then on again, the countdown timer stops changing, and that repeats as long as you like.

After considerable listening & pondering, I decided one event early in the Dry cycle involved starting a fan to vent the steam out of the interior. The wiring diagram shows the fan has a Fault wire: perhaps the fan has failed.

The maintenance manual shows different fans in three different places, although the control board has a connector for only one. By process of elimination, I found the fan atop the cabinet:

Samsung dishwasher – top view

The cable from the fan in the vented compartment on the left burrows under the gray duct, around its back side, and plugs into the small white connector on the right. You must ease the cable from a row of hooks guiding it around the back of the duct, which requires slightly lifting the duct.

Unhook the two metal straps, remove four screws from the black vent, and lift it off the top to reveal the duct outlet pores:

Samsung dishwasher – fan duct – overview

Remove four more screws, lift the fan duct assembly just a little bit, and pry open three latches around the fan compartment with a consumer electronics case-cracking tool:

Samsung dishwasher – fan housing

The new fan (on the right) looks very much like the OEM fan (on the left), even though it’s the $15 version rather than the $150 version you might buy from similar randomly named sellers if you were so inclined:

Samsung dishwasher – OEM vs new fan

Detach the old fan & its cable, drop the new fan in place, snake its cable, plug its plug, and install All The Things in reverse order.

Unfortunately, after shoving the dishwasher back into its cubby, the new fan didn’t change the failure at all.

I hitched the old fan up to the bench supply and it spun just like it should. Wiring the Fault wire to a 5 V supply through a resistor shows it’s the usual tachometer signal pulsing as the rotor spins.

Which means the next step requires more pondering and PCB probing. The failure is too consistent to be a Heisenbug, but maybe something shook loose in there.

2026-06-12

Outlet Strip Bench Mount

A spate of tidying-up led to mounting an outlet strip along the back of a bench:

Outlet Bench Mount - installed — Outlet Bench Mount – installed

Rather than drill holes into the top of the bench for those screws, they fit into M4 brass inserts heat-staked into the brackets:

Outlet Bench Mount - show view — Outlet Bench Mount – show view

The holes for those inserts aren’t centered side-to-side on the brackets, because the screw holes aren’t centered on the bent-steel angles forming the outlet strip endplates.

The bottom arm on the brackets probably isn’t necessary, but they kept the outlet strip from crawling away while I match-drilled two holes for the screws into the side of the benchtop.

For obvious reasons, the brackets print on their sides:

Outlet Bench Mount - build view — Outlet Bench Mount – build view

Another outlet strip from a different manufacturer is, of course, different, but changing three parameters in the OpenSCAD program summons a different bracket from the vasty digital deep:

Outlet Bench Mount – different brand

Parametric modeling and a 3D printer are exactly the right hammers for the job …

The OpenSCAD source code as a GitHub Gist:

	// Shower soap dish
	// Ed Nisley – KE4ZNU
	// 2026-06-04

	include <BOSL2/std.scad>

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

	/* [Hidden] */

	HoleWindage = 0.2;
	Protrusion = 0.1;
	NumSides = 334;

	Gap = 10.0/2;

	$fn=NumSides;

	ID = 0;
	OD = 1;
	LENGTH = 2;

	BenchThick = 21.0; // workbench top

	ScrewOD = 4.0; // into edge of bench
	Insert = [4.0,5.5,10.0]; // robust M4 insert

	WallThick = 10.0;
	BaseThick = 10.0;

	OutletBase = [15.0,40.0];

	HoleOffset = 6.5; // from outside edge of bracket
	HoleOC = 24.0;


	MountOA = [OutletBase.x,OutletBase.y,BenchThick + Insert[LENGTH] + 1.0 + BaseThick];

	//———-
	// Build it

	module Mount() {

	difference() {

	cuboid(MountOA,rounding=1.0,anchor=BOTTOM + BACK);

	up(BaseThick)
	fwd(WallThick)
	cuboid([2*MountOA.x,MountOA.y,BenchThick],anchor=BOTTOM + BACK);

	up(BaseThick + BenchThick/2) back(Protrusion)
	ycyl(OutletBase.y,d=ScrewOD,circum=true,$fn=6,anchor=BACK);

	for (j=[-1,1])
	fwd(MountOA.y/2 + j*HoleOC/2)
	right(HoleOffset – MountOA.x/2)
	up(MountOA.z + Protrusion)
	cyl(Insert[LENGTH],d=Insert[OD],circum=true,$fn=6,anchor=TOP);

	}
	}

	//———-
	// Build it

	if (Layout == "Show") {

	left(Gap + MountOA.x/2)
	Mount();

	right(Gap + MountOA.x/2)
	xflip() Mount(); // mirror for the other end of the outlet strip

	}

	if (Layout == "Build") {

	left(MountOA.z/2)
	up(MountOA.x/2)
	yrot(90)
	Mount();

	fwd(1.5*MountOA.y)
	left(MountOA.z/2 – BenchThick/2 – Insert[LENGTH]/2)
	zrot(180)
	up(MountOA.x/2)
	yrot(-90)
	xflip() Mount(); // mirror for the other end of the outlet strip


	}

view raw Outlet Bench Mount.scad hosted with ❤ by GitHub

2026-06-10

Aneng AN8009 Resistance Offset

Before measuring a wire resistance in the laser cutter, I checked the resistance of the two test leads on the Aneng AN8009 meter (“Check your zero!”) to show an unsteady reading around dozen ohms.

Poking around inside showed the internal fuse apparently making poor contact with its holder, as poking it changed the random values:

Aneng 8009 low-current fuse

Two tiny drops of Caig DeoxIT stabilized the reading around 1 Ω across several different combinations of test probes, so I declared victory. There is surely an offset calibration buried in the firmware, but it’s no longer a trimpot available to service technicians.

The ceramic fuse has an internal resistance of about an ohm, but swapping it for a replacement fuse with 0.2 Ω resistance didn’t materially change the results. It’s worth noting those glass fuses are slightly longer than they should be, surely due to their leads, and required slightly bending the fuseholder clips.

2026-05-27
Prusa MK4 Camera Lighting
Although the Raspberry Pi camera has a good view of the Prusa MK4’s extruder, there’s not much light under there:

RPi Camera Mount – image

There’s also not much room for a lighting fixture on the printer where it must mount, so I modified a trio of nominally 12 V / 4 W COB LED panels:

Prusa MK4 – Extruder sidelight – COB LEDs

Their “4 W” rating seems aspirational, at best, as a 12 VDC supply pushes only 75 mA through the panel, so they tick along at 900 mW. If you expect cheap eBay / Amazon components to live up to their specs, dream on.

The modifications:
- Unsolder the pins
- Crunch off the surprisingly precise 27.4 Ω SMD resistor
- Clean up the rubble
- Wire the panels directly in series, ignoring their bridge rectifiers
The 15 LEDs on each panel are arranged in five parallel chains of three LEDs for a total forward drop of 8.3 V, so putting three panels in series works with the MK4’s 24 V power supply.

Stick them onto the MK4 power supply case with foam tape and wire them directly to the 24 V terminals:

Prusa MK4 – Extruder sidelight – installed

There’s very little clearance between the machine frame and the X Axis carriage on the threaded rod. Putting the LEDs in a 3D printed case and routing the wires lower on the column would be nice touches:

Prusa MK4 – Extruder sidelight – front view

The panels start at 30 mA when cold and drop to 25 mA as they warm up in the 63 °F = 17 °C Basement Shop. Each panel dissipates 250 mW: bright enough for the task, dim enough to avoid overpowering the camera’s limited dynamic range, and definitely within whatever power rating they should have.

Looking over the camera’s shoulder in normal shop lighting suggests it’s about right:

Prusa MK4 – Extruder sidelight – camera overview

A staged scene with the shop lights turned off:

Prusa MK4 – Extruder sidelight – low-light view

Call it Good Enough™ for the purpose.
2026-05-20
RPi Camera RTSP Setup
The rpicam.cfg file holding the parameters for the Raspberry Pi watching the Prusa MK 4 printer:
```
bitrate=2000000 
framerate=15
timeout=0 
nopreview= 
codec=libav 
libav-format=mpegts 
width=1280
height=720
rotation=180
roi=0.00,0.00,1.0,1.0
hdr=auto
```
The RPi camera for the wren nest is just taped to the window, but has a configuration providing a bigger picture:
```
bitrate=2000000 
framerate=15
timeout=0 
nopreview= 
codec=libav 
libav-format=mpegts 
width=1920 
height=1080 
roi=0.00,0.00,1.0,1.0
hdr=auto
```
The useful pieces:
- bitrate sets the average data rate, which may be too high for comfort outside your immediate LAN
- framerate need not be as high as you think
- nopreview prevents a preview picture while starting
- width and height do the obvious thing, but don’t try to be too clever
- roi picks the image from a specific part of the camera sensor, so you can adjust the image layout if you have a rigidly fixed camera
- hdr doesn’t do anything for cheap RPi cameras
Putting all the fiddly config in a file reduces the command line invocation to a mere jawbreaker:
```
rpicam-vid --config rpicam.cfg -o - | cvlc stream:///dev/stdin --sout '#rtp{sdp=rtsp://:5886/wrens}' &
```
Although you’d want to set that up to run automagically when the RPi starts up, for now I just fire it off as needed through an SSH session, with the ampersand letting it run after that terminal session closes.

The RTSP port (5886) and stream (wrens) can be anything you like, which comes in handy when squirting streams through port-forwarded firewall pinholes using a router that cannot handle different external and internal port numbers.

Useful background info:
2026-05-18
USB Micro-B Connector: FAIL

While setting up a Raspberry Pi camera, I had occasion to pull out its USB power cable, whereupon grabbing the camera while unscrewing it from the tripod felt unusually sharp:

Micro-B USB – RPi jack

It seems the wall wart’s USB Micro-B connector pulled apart:

Micro-B USB connector – disembowled

Somewhat to my surprise, it was a CanaKit 5 V 2.5 A wall wart, definitely not the cheapest piece of junk ever made by the hand of man. On the other paw, it’s been around for quite a while, so …

Even I will agree that’s not a repairable failure, so I planned to splice in a Micro-B connector from a volunteer chosen from the Box o’ USB Micro-B Cables:

Micro-B USB connector – tiny wires

Nice color code in there, eh?

Each of those conductors appears to be made up of nine springy copper-colored 0.06 mm strands, somewhat smaller than 40 AWG: not what you want on the business end of a 2.5 A wall wart. I had previously measured the cable’s overall resistance with a surprisingly useful Treedix USB Cable Tester and it was on the very high end of the charge-only cable collection.

So I soldered a female USB-A breakout from the Drawer o’ USB Breakouts to the wall wart’s wires, snapped a 3D printed case around it, got a good (0.26 Ω) A-to-Micro-B cable from the Box o’ USB Adapters, and moved on.

Ya gotta have stuff, but that was absurd.

2026-05-07
Branson 200 Ultrasonic Cleaner: Wiring Fix

Our ancient Branson 200 Ultrasonic Cleaner began behaving erratically due to water seeping under the rather casual seal from last year’s fix. Although drying the switches let it start up again, it would run for only a few seconds before shutting down again, which suggested a deeper problem than just the switches.

Take a picture of the PCB’s component side:

Branson 200 Ultrasonic Cleaner – PCB component side

And of the solder side:

Branson 200 Ultrasonic Cleaner – PCB solder side

Transform those pictures to be the nice real rectangles shown above, resize to a common pixel format, mirror the solder side, turn it into a layer atop the component side, then tweak its opacity to make both sides visible at once:

Branson 200 Ultrasonic Cleaner – PCB overlay

Some pondering produces a partial schematic of the left half of the board:

Branson 200 Ultrasonic Cleaner – partial schematic

The 1:1 transformer is constantly powered, so the ON button connects the 120 V (!) half-wave rectified output to the +12V supply bus, with the 750 Ω resistor dropping most of the voltage while the switch is pressed.

The hotwired +12V supply forces the relay closed, which (in some as-yet unidentified way) fires up a +12V power source to hold the relay closed, with the 555 timer driving an MC14060 14-bit divider to count down the time until it turns itself off.

Reminder: this design dates back to the days when a pair of chips and a handful of through-hole components cost less than one of those fancy microcontroller thingies.

Plug the cleaner into an isolation transformer and trace the half-wave rectified signal through ON button to find it got all the way to the contact on the end of the orange wire in the connector, but did not reach the pin header on the PCB.

A closer look at the connector revealed a broken contact on the white wire, which I (rather crudely) soldered together while considering my choices:

Branson 200 Ultrasonic Cleaner – soldered contact

While plugging that wire back in place, this happened:

Branson 200 Ultrasonic Cleaner – another broken contact

Neither of those are the (presumably) similarly failed orange wire, but even I can get a clue from three similar failures.

So I replaced the OEM connector with a JST-XHP 2.54 mm connector from an assortment I got for another project, replaced the chunky 22 AWG wires with flexy 26 AWG silicone wires in the same cheerful rainbow colors, and it began working perfectly again.

The buttons needed another water seal, so I tweaked the previous layout to kiss-cut GITD tape and through-cut colorful vinyl sheets:

Branson 200 Ultrasonic Cleaner – power button cutting

Capped with a transparent cover sheet cut from a pack of PDA screen protectors (remember PDAs?):

Branson 200 Ultrasonic Cleaner – power button cover

In truth, the GITD tape is too thick, so I’ll probably repeat this dance later this year.

FWIW, I was totally ready to buy a new ultrasonic cleaner, but all of them have scathing one-star Amazon reviews, to the extent I decided fixing this cleaner would be much easier than fixing a new one that’s been cheapnified to the point of no return. A common complaint seems to be water leaking into their capacitive switches and killing the circuitry stone cold dead: not an improvement over this one.

2026-05-06