Ghettobotics nonshopping list

Kragen Javier Sitaker, 02020-11-04 (updated 02020-12-21) (22 minutes)

So, I have some basic gifted, bought, or lent electronics equipment: a multimeter with a broken lead and I think broken current measurement, a low-temperature hot glue gun, an audio cable, an amplified battery-powered speaker, a non-temperature-controlled soldering iron that leaks some mains current into the workpiece, a Blue Pill, an ST-Link, a couple laptops, a netbook, some USB chargers and cables, a USB power pack, some tin snips, some yogurt cups, some acetic transparent silicone caulk, a couple of cellphones running Android, some quartz-halogen lightbulbs, some butane lighters, a butane blowtorch, some box cutters, a vise-grip, nail polish, ethanol, some nail-polish diluent, some Q-tips, needlenose pliers, a scrap terry-cloth towel, some electrical tape, a carpentry tape measure, a hammer, some chopsticks, an electric screwdriver, some hex tips for it, and some popsicle sticks. My current objective is to bootstrap from this to a reasonable electronics lab, buying a minimal additional amount and replacing as many as possible of the bought items with custom items, made out of garbage.

Of these, I think it's reasonable to list the following as reusable electronics-workshop equipment:

  1. a multimeter
  2. glue gun
  3. soldering iron
  4. Blue Pill
  5. ST-Link
  6. box cutters
  7. needlenose pliers

Recent repairs and improvisations

Although listing my numerous follies would probably help people to avoid repeating them, these are just the successes.

Soldering sponge

The soldering iron came with a stamped-steel stand to hold the tip off your workbench. But to clean the tip of a soldering iron (and also to reduce the temperature of a non-temperature-controlled iron like this one) you need a tip cleaner. Typically this is a wet cellulose sponge. But all the sponges here are polyurethane, which would not work; it would melt onto the tip instead of cleaning it.

So I cut a square of some 70 mm from the terry-cloth towel with the tin snips, wet it, and put it under the soldering iron stand. The towel turns out to be cotton, so this works well.

Multimeter lead repair

I repaired the broken multimeter test lead by using some modeling wax (paraffin + rosin) as flux. To solder the wire, I carved back the plastic around the broken wire with the box cutter, stripped the wire with the flush cutters, fluxed the wires, tinned them with some solder from generous joints in a discarded VCR, and touched them together, held in my hand.

To reinsulate it, I should have used hot glue, but instead I used transparent silicone caulk; this sticks inadequately to the plastic, isn't stiff enough to provide adequate strain relief, took a day to cure, and may turn the wire into copper acetate.

To stiffen it, I squirted hot glue all over it, wrapping it all the way around the silicone in places, since the EVA sticks inadequately to the silicone. The hot glue didn't stick adequately to the plastic, making it a strain concentrator instead of a strain relief; this turned out to be because a thin layer of silicone was in between.

I melted the hot glue with a butane stove lighter, scraped the molten hot glue out of the way with a popsicle stick coated in silicone, and then removed the intrusive silicone. Then I used nail polish diluent and a Q-tip to scrub the plastic, then applied more hot glue to complete the repair.

Solder braid

I cut a meter from the power cord of a pressure washer I had found disassembled and discarded on the sidewalk a few months ago; after slitting the end of the outer insulation with a box cutter, I was able to pull one of the inner insulated conductors out to tear the outer insulation lengthwise, just using my hand. It contained two stranded copper wires made of dozens of 100-μm-wide strands. (I measured using the cellphone as microscope; see below.). I separated out three hanks, trimmed them with the flush cutters, tied them into an overhand knot, and began braiding. After I had 220 mm of braid, I cut it at both ends with the flush cutters. I then heated it with the soldering iron to coat it with the modeling wax as flux.

This solder braid makes it enormously easier to scavenge soldered parts with large or numerous pins.

Cellphone as microscope and flashlight

My cellphone can focus down to about 70 mm. This body is so old that it can no longer focus its eyes closer than about 210 mm. The cellphone is a little less clear than these eyes; they can read 3.8-mm-tall text from 1600 mm, while the camera can only read it from about 900 mm. (This works out to about 3.8/6/1600 ≈ 400 μrad ≈ 1'20" of resolution for the eyes, or 700 μrad or 2'30" for the camera; double these numbers for resolution in terms of distance between lines.) So even without any additional optical magnification the cellphone can provide about 1.7× magnification of small objects, and of course it’s easy to expand the resulting blurry pictures on the screen to be much larger, which I measure as 20× the linear size of the original object; this doesn’t make any new details visible but does make the same details more obvious.

It is of course common to use cellphones as flashlights or to use an LED on them to illuminate an object being photographed.

Attempts to use cheap pocket Fresnel magnifiers to improve the resolving power of the cellphone have so far met with failure. I’m considering buying one of those “60×” clip-on “cellphone microscopes”.

Closeup image quality is improved substantially by supporting the cellphone on top of a table on a a 75-to-76-mm-tall triangular prism, with triangle sides of 55, 55, and 60 mm, folded from a 75-mm-wide strip cut from a cardboard box discarded by the supermarket using box cutters, measured with the tape measure, marked with a ballpoint pen, and taped into shape using electrical tape. The triangle firmly supports the cellphone’s center of gravity and the place where you have to tap to take a photo, without obstructing its camera, “flash”, light sensors, or power button. This reduces non-parallelism, eliminates motion blur, and keeps the imaging range near optimal (as close as possible without losing focus.) So this probably gets to better than 1.7× resolving power, maybe 3×.

Both of the cellphones I happen to have here are more or less equivalent for this.

password2 points out that there’s a program called scrcpy which might be useful for this, allowing me to disable the screensaver and display the cellphone’s screen on a laptop or netbook connected over USB or TCP/IP. This would afford greater magnification, because those screens are bigger, though of course not greater detail, and probably with more latency.

My equipment wishlist

In addition to the stuff above that I already have, I want:

  1. helping hands
  2. a bench power supply
  3. magic tweezers (i.e., with an integrated LCR meter)
  4. a second multimeter (with working current measurement and a buzzer)
  5. alligator-clip wires
  6. breadboards
  7. an oscilloscope
  8. a temperature-controlled hot-air setup
  9. a temperature-controlled soldering iron
  10. a 3-D printer
  11. a custom PCB fabrication machine such as a mill
  12. an autonomous solar-powered personal computer
  13. a stock of consumable parts (see below)
  14. some way to organize parts
  15. a workbench
  16. a portable tool chest for carrying all this around with a place for every item
  17. a precision multimeter
  18. a printer
  19. a dremel
  20. a Kelvin-sensing power supply
  21. thermometers
  22. a Kelvin-sensing ohmmeter
  23. a thermal imager
  24. a function generator
  25. a caliper
  26. a protocol analyzer
  27. a microscope, and
  28. micro-scale waldos.

These are not in priority order.

Desired consumable parts

I want:

  1. solder
  2. solder braid
  3. flux
  4. resistors
  5. capacitors
  6. LEDs
  7. inductors
  8. microcontrollers
  9. connectors
  10. breadboard jumper wires
  11. transistors
  12. triacs
  13. batteries
  14. supercaps
  15. crystals
  16. op-amps
  17. linear regulators
  18. switchers
  19. buttons and other switches
  20. photodiodes
  21. phototransistors
  22. Darlingtons
  23. potentiometers
  24. other linear and rotational sensors
  25. MOSFETs
  26. motors
  27. motor drivers
  28. MOSFET drivers
  29. speakers
  30. microphones
  31. temperature sensors
  32. heating elements
  33. diodes
  34. zeners
  35. voltage references
  36. level converters
  37. humidity sensors
  38. pumps
  39. valves
  40. displays
  41. cameras
  42. accelerometers
  43. gyros, and
  44. touch sensors.

These are not in priority order either.

Breadboards and jumper wires

A standard solderless breadboard is extremely useful for prototyping and testing circuits. But the original breadboards were actual boards, made of wood, used for the same purpose; to connect two or more components, you would put their leads under a washer, dished outwards, and screw the washer to the board to pinch their leads under its edge, making a tight connection. This is feasible but far less convenient than the now-standard approach.

In my childhood, I would take government welfare cheese boxes and poke electronic components through the non-corrugated, non-plastic-coated cardboard. Panel-mount components I would poke through all the way, then screw them down to the cardboard as normal; for through-hole components, I would poke just the leads through. Then I could connect the leads inside the box using alligator-clip jumpers. All of this was made easier by the use of parts that hadn’t been used yet, so the leads were still long, and the near complete nonexistence of surface-mount parts. This approach is convenient; construction paper is of the appropriate weight, and so are the inner and outer walls of single-wall corrugated cardboard, which can be soaked apart in water.

A disadvantage to this approach is that it doesn’t scale to more than a few dozen connections.

At the time, I also had a Radio Shack “300 in 1” electronics kit, which had about 50 components mounted on a piece of cardboard next to some extension springs. By thumbing a spring to the side, you opened up space between its coils to insert one or more wires into.

In some YouChube video, Espacio de César demonstrated the construction of a now-conventional solderless breadboard from a pile of 2.54-mm-pitch DIP sockets; he snipped out the middle of the sockets, stacked five socket sides together, and wired them up in the conventional way. I suspect that even a very modest effort in this direction would yield useful results: three 8-pin molex female connectors or socket sides stacked up would be sufficient for a fairly wide variety of circuits with discrete through-hole components.

Other candidate sources of such female connectors include connectors on cables and jumpers for configuring boards by connecting two such adjacent pins.

The usual way to use the now-conventional breadboard is with jumper wires made of hookup wire with nothing on the ends, but you can also solder traditional male pins onto the ends of wires in order to make it easier to plug and unplug them.

The absolute minimum hardware for that kind of convenient plugging and unplugging of pins is female-female jumper wires: basically a single-pin female molex connector soldered onto each end of a jumper wire. You engulf a lead of a component with each pin, thus setting up a two-terminal net. Nets with three or more terminals can be constructed with Y-shaped female–female–male jumpers, which additionally have a third wire attached with a male pin on it; these can be daisy-chained to make nets of any degree of complexity.

These female–female jumpers seem to be more or less the standard that the Arduino world is converging on: each component is on its own little PCB, with some 2.54-mm-pitch pins sticking off the edge. (I’ve even seen such a PCB for an ULN2803, which is a 2.54-mm-pitch DIP with eight Darlingtons on it.) You can either plug it into your breadboard or you can hook up each of the pins directly to some other pin on some other board with a female–female wire. Often, 1970s-style rainbow cable is involved between the two ends of the cable.

If I rip apart a largish motor, dead or otherwise, I can get many meters of thick magnet wire out of it, which is directly suitable for use as breadboard-style hookup wire once you sand its ends. For very-low-frequency or very-low-current circuits, twisty ties from bread bags can also be used; and jumper wires from single-sided PCBs can sometimes work for short distances.

So, connectors are the most immediate, urgent necessity.

Tool chest

How can the lab fit into a portable toolchest? Suppose it’s 15 kg, if it has wheels, and maybe 200 mm × 400 mm × 500 mm, an acceptable size to carry on the bus, stuff under your seat, or carry through a doorway. (This ATX tower PC in front of me is 200 mm × 450 mm × 450 mm, a very similar size, though a marginally more inconvenient one.)

The toolchest needs to be able to contain the 7 pieces of equipment I have, plus the 27 other pieces of equipment I want, and the 44 kinds of consumable parts. In a very crude averaging sense this allocates about 190 g to each of these 79 “items”, including the toolchest itself, and just over 500 mℓ per “item”. Because any item larger than these averages must be compensated for by items additively smaller, the vast majority will need to be smaller than that; I think it makes sense to “budget” 40 g and 100 mℓ per “normal item” in order to have space left over for the few that can’t squeeze in that way.

As some motivating examples, this multimeter — a lightweight handheld one — weighs 152 g without its leads, 202 g with its leads, and occupies 70 mm × 25 mm × 125 mm without its leads, 219 mℓ. The flush cutters weigh 64 g; the soldering iron weighs 174 g with its stand and “sponge”. The iron is some 230 mm long and, with the cord, 60 mm in diameter, thus occupying some 650 mℓ. The flush cutters occupy 150 mm × 60 mm × 10 mm when closed, thus some 90 mℓ, the only item to come in under the 100 mℓ “normal item” threshold, though they exceed the weight threshold by more than half.

So the vast majority of “items” need to be much smaller than this, and probably these three in particular need to be replaced by smaller and lighter items.

Suppose the items have a Zipf distribution of weight and volume, which is maybe pessimistic but probably not much. Then the first ten weights would be 3 kg (perhaps the 3-D printer), 1.5 kg, 1 kg, 750 g, 600 g, then 500 g, 430 g, 380 g, 340 g, 300 g; the next ten between 150 and 275 g; the next ten between 100 and 150 g; the next ten between 75 and 100 g; the next ten between 60 and 75 g; the next ten between 50 and 60 g; the next ten between 43 and 50 g; and the last nine between 38 and 43 g. In particular this gives the quartiles as about 50, 100, and 150 g. And the top few volumes out of 40 ℓ are 8 ℓ, 4 ℓ, 2.7 ℓ, 2 ℓ, 1.6 ℓ, etc.; and the quartiles are roughly 130 mℓ, 200 mℓ, and 400 mℓ.

This suggests that my “normal item” quotas above were maybe a bit too stingy. A “normal item” can be 100 g and 200 mℓ; every item over the threshold must be compensated for by an item below it. Still, of the three test items, only the flush cutters are under these thresholds.

Suppose the resistors are close to the median and I have, say, 1000 resistors. (An SMD sampler book I was looking at online has 25 resistors of each of 170 values, 4250 resistors in all. It costs about US$8 overseas but about US$50 in Argentina.) Then the resistors average 100 mg each and 200 μℓ each. Right now I have about 50 resistors (five of which are trimpots) and they weigh about 3 grams, so I think this is realistic.

But how can I keep them organized? This is actually far more pressing than the problem described above of how to reduce the multimeter’s weight by half; to build circuits with the hundreds of components I’ve already scavenged from discarded machinery, I must be able to find the components I need quickly, I must be able to connect them together, and then I must be able to probe the resulting assemblage in some way to find out why it’s broken.

Ideally I’d press the “heat” button to heat the smart tweezers’ tips to desoldering temperature, pick up a component with them, and the smart tweezer would chime, indicating that the component had been successfully characterized; I could then press a “print” button with my foot, storing the test results and printing a tiny envelope describing the component and giving the crucial test values. Then I would put the component into the envelope, and file the envelope in the larger envelope for its type.

A common 100-mW 0805 resistor is 2 mm × 1.25 mm × 0.5 mm, totaling 1.25 mm³ or 1.25 μℓ; you can fit 800 of them in a milliliter. So the whole collection of 4250 resistors mentioned earlier would, without the paper, amount to some 6 mℓ.

Resistor envelopes

I’m trying out categorizing physicaly-small resistors by E12-series resistance and putting them into paper envelopes, one for each E12 value (10 12 15, 18 22 29, 33 39 47, 55 68 82). Most of the resistors I’ve salvaged so far are in the 100Ω–100kΩ range, which requires 37 envelopes, much like Merlin Mann. There are a few higher and quite a number lower, but so far I don’t have an accurate way to measure resistances below 100Ω. Hopefully I can rig something up soon.

So far I only have on average two or three resistors of each denomination, many with very short leads. I probably need at least four times the number of resistors I have already, a couple hundred, to make even smallish projects straightforward.

I found my helping-hands to be very helpful for resistor measurement: I can connect one lead of the multimeter to one alligator clip, place the resistor in the other clip where I can see it through the glass, and touch its other lead with the other probe. I got up to about two resistors a minute that way. I suspect that a better ohmmeter and using alligator clips will speed me up further.

I also have a partitioned plastic box, like those for fishing tackle, which is partitioned into E3 resistor values.

Bench power supplies

I have a 12-volt power brick that includes current limiting, as I found when I used it to electrolytically dissolve some copper; for some 45 minutes the output voltage was well below 12 volts. But I need something where I can twist a knob to scan across a range of voltages. The “Tech Ideas” YouTube channel from India points out that there’s typically a TL431 on the low-voltage side of modern isolated switching power supplies, directly hooked up to the feedback optocoupler, and the TL431 is programmed with a voltage divider, so you can replace its fixed voltage divider with a pot, and maybe upgrade the output-side filter caps, and you get typically a 3V–25V adjustable switcher, with whatever current limiting the original supply had.

One difficulty is that you typically want to know what voltage you’re getting, so you probably want to add some voltmeters on the output. Also it’s common to want voltage down to zero; for low-power loads you can do this very easily with a linear output stage consisting of an emitter follower following a potentiometer.

Topics