My Custom Power Supply (is complete!)

2018-10-18 18:29 - Making

You might not truly be an electronics nerd until you build your own power supply. Either way, I've finally passed that threshold. As I've mentioned previously (and previouslier), I've been working on mine — very slowly, off and on — for most of a year. The bare start came with a guide posted to Hackaday about using nichrome wire to heat and bend acrylic plastic in straight lines, to make cases.

The top half of my power supply case, freshly bent up. The bottom half of my power supply case, which needed some work.

I previously failed to bend the top correctly, and broke it in the process. I took more time, and made sure the folds-to-be were more thoroughly melted before pushing this time, and it came out almost perfect. There's one line in the top left where I melted the plastic enough to make a visible mark in the non-bent bit. (This is a fold down of the side. The holes on the top attach the side piece through this small folded piece.) I might have been able to avoid this, but it's only a small cosmetic issue, and more visible in pictures than in person.

The bottom needed work. I trimmed off bits of the sides to get the front lip to fold in, and later I ended up needing to cut off most of that lip. Some detail didn't go quite right in the planning and design. Worse, I got it left-right backwards. I chose plastic with a matte finish, but it is matte only on one side. I put the wrong side up (out) when cutting, so to get the matte surface on the outside, everything was flipped. This worked out mostly okay, but the (perhaps unnecessary) fan is blocked more than I would have hoped, and I had to extend some wires inside to make everything reach where it's supposed to.

The worst issue with the case is that I had barely experimented with the plastic folding, so I had to guess in a few places and didn't get what I'd have really liked. Some of the parts are located too close to the edge, causing a fold to warp or stretch inelegantly. I had to mostly guess about where the various surfaces would end up relative to each other to get things like screw holes to align. On the inner half, I made the screw holes elongated to cope with this, which ended up helping a lot later on!

The back of the rectifier board, close up. The rectifier board, wired to the transformer, with the case.

The custom circuitry is really just this rectifier board. The diodes are tacked on the back — their leads are too thick to fit through the holes. The other side is stuffed full of capacitors, and another rectifier for the lower voltage half of the circuit, hooked to (I think) the transformer's center tap. That transformer was salvaged from a failed UPS. It's supposed to handle 1200 watts, so although I don't know its precise specs, I'm comfortable with what use I'm putting it to, here.

The transformer and rectifier, in place. Close up of the front side of the rectifier.

Things are getting more assembled here. The transformer is in place, and the rectifier board is nestled right next to it. On this side, it's brimming with capacitors (and one inductor) to smooth out the rectified AC voltage. There's hot melt glue everywhere. The fan is tucked behind the transformer, and at the left you can see the back of the plug/switch/fuse. I've been careful to heat shrink and/or hot glue exposed conductors. The bottom has slots for air intake, and you can see the extra cuts I've made, where things didn't fit quite as planned. After this point there was a significant delay as I decided to pause, to get black nylon (non conductive!) screws to assemble with. Then did it again when, only after receiving them, I realized the ones I ordered were too short to work for this.

Both halves of the power supply, nearly assembled. I 3D printed some brackets, to get the USB connectors to line up with the holes in the case.

The left picture above is near the final stage of assembly. In addition to two main front end units, I've also got a USB supply in here. It and the fan are driven from the lower voltage middle tap of the transformer. I tried my best to plan the mounting holes (on the top) and connector holes (on the side), but I was way off. At least one layer of plastic moved around in late design tweaks. So I ended up with these four little 3D printed brackets. Each is screwed into the USB supply board, dropping it quite a bit lower and moving it back a smidge, and those get screwed into the top of the case. This unit displays the voltage (not so useful unless you're drawing so much current it sags?) and the current being drawn. And it constantly flips back and forth. I was concerned that the blinking would be annoying. I tried to address that by sandwiching in a smoke-colored only-mostly-transparent piece of acrylic between it and the world. Due to these brackets, it ends up only being visible from just the right angle, so the blinking display will not be an issue!

The power supply, plugged in and operational. The back of the power supply. The side, with the USB connections that took special effort to line up. The top, with the USB supply showing half an amp of draw to run the fan.

Here's the final product. There's two RD Tech 5005 (50 volt, 5 amp max) units on front. Each is wired to a pair of both binding posts and banana jacks, so I can hook pretty much anything up. The fan exhausts out the back, next to the power plug, with switch and fuse integrated. On the right side are the USB connectors which took those brackets to line up, here a fan is plugged into one of the two ports. If you look down from the top, you can see the USB supply's display, in this case showing half an amp being drawn by the fan.

The top clearly doesn't line up perfectly with something else, so it's bowed a bit. Assembling the case, and getting all the screws to line up and mate with the nuts inside was very difficult, so I'm not going to open it up and try to fix this unless something worse than cosmetics goes wrong. Look closely around the power plug, and you can see that I bent the edge out from beside it, because the hole was too close to the edge being bent. A bunch of extra hot melt glue helps the plug stay put, as a result. The same sort of (unavoidable, in this case) thing happened at the front: the supplies each have a USB connector beside them, which can be used to log data and set settings via a computer. The hole for the connector must be right next to the mounting screw holes, and so each of them ended up puckered along the folded edge.

I'm really happy with the angled front, which makes the displays very readable from a normal sitting position. And I'm extremely happy that I've finally taken this project through to completion! Learning how to melt and fold acrylic opens up all sorts of opportunities.

Power Supply Hurdles

2018-09-01 16:29 - Making

My last post was about a successful step in a long delayed project. I've continued making progress, but mostly progress towards hurdles I haven't quite passed yet. I did cut out my case designs in acrylic plastic:

Laser cutting the bottom case part. The bottom case part, protective layers removed. The top case part, protective layers removed.

My case design involves a bottom part, which is also the back (after a bend). Plus the front part, which is also the sides and top after a few more bends. I didn't quite think about making all these bends, when doing the design. I folded the sides in first, and that went pretty well. Except at the left side, near those big openings in the middle. Acrylic is brittle, and I put some hairline cracks in the very narrow part across the middle and the slightly wider part near the bottom. Mostly cosmetic, so that's okay. Except that left folding the top down, which went even worse.

The case front, shattered after a bad attempt to bend the top down.

In addition to adding weak points, all these holes leave fewer places to hold on. I ended up shattering this larger front piece of the case completely, while making that last bend. Quite the bummer, but at least it leaves me the opportunity to double check a few measurements. I've got one spare piece of plastic to use for a second try. And I know to be careful when I attempt the (simpler) back piece as well.

With that physical part of the project on hold, I returned to the electrical part. There's power in, which goes through a (huge) transformer, salvaged from a bad UPS. This drops line voltage to around fifty volts, which is rectified by a circuit also cobbled together with parts from that bad UPS and some old bad computer power supplies. This goes into a pair of RDTech DPS modules, which handle voltage and current limiting for the adjustable outputs. The transformer has another tap, which produces around 18 volts. I'm going to separately use that to power a fan and a USB supply module (which was visible in the corner of the fresh-cut before-bending case front picture).

An end-to-end test set up of my power supply. Ripple voltage at the rectifier output. Power being drawn during this test.

Here's all the key parts of that set up on the bench as a test. The good news is that it works. The bad news is it doesn't work very well.

The full bridge rectifier has just a pair of 66µF capacitors smoothing it out. At first that seemed okay enough. Since I did some very early tests I've gotten an electronic load, at bottom right of the first picture above, which I can use to do more thorough testing. In that picture, it's drawing twelve volts at two amps, well below the 50V / 5A theoretical upper limits (of the DPS modules). But even at this low power level, it's causing around nine volts of ripple at the rectifier output. Far too much! I know that adding an inductor can create a much more stable rectified output. It drops the maximum voltage level, but this much ripple will too, so I'm going to experiment with inductors I can salvage, for the more reliable of the two lower voltage options. (I don't really need that whole 50 volt output range, anyway.)

Plastic Bending

2018-08-21 21:03 - Making

My first (failed) plastic bending system.

I've had a power supply project in the works for a good while. When I saw a post about bending acrylic plastic to make custom cases for electronics, I knew I'd use that technique. I didn't want to build such a big contraption though, and I delayed quite a bit. I had the system pictured above assembled some time ago, but never operated it.

More recently, I tried harder. I doubled up the length of wire, to half the resistance and thus double the power (at the same voltage). I ended up hooking a laptop power brick to the unit that will be the heart of the supply I'm building, and I finally got the bending system working! Note the spring at the far end. Over the twelve inches or so that I'm heating (in the working set-up), there's a noticeable expansion of the hot wire. The spring helps take up that slack, and keep the wire strung in a nice straight line, near the plastic to be heated.

My refined, working, acrylic bending system.

I added some L aluminum brackets to (maybe) better contain and focus the heat, and (definitely) help me hold the plastic in place correctly. I'm only energizing a short piece of the wire I've got in place. I didn't know at all what sort of voltage or current would be necessary when I started. I correctly figured that I could always connect less of it, if I put too much.

In this shot you can see the successful fruits of my tests. If I hold the plastic still above the orange hot wire, after a little while it heats to the point of becoming quite malleable. Quickly and carefully bend at that point, and the plastic will quickly cool back to being solid. I got three ninety degree bends in the test piece. I'll do a few more tests to figure out how the inside/outside ends up: what's the radius of the bend? If I try to make a box, where do I need to put the bends to know what the outside dimension will be (so it lines up with the other half of the box!)?

Either way, it feels good to have this long delayed part of the project behind me!

Moviepass is dead? (And I killed it?)

2018-08-06 23:20 - General

I found out about MoviePass mid/late last year. I probably saw an average of one movie per year until then. In general I'm patient, so I'm in no rush to see a movie in the theater, and the ticket price is more than I'm willing to pay for little benefit (from my perspective). But MoviePass allowed me to see effectively all the movies I could ever want, in the theater, for ten bucks a month. So I used it. A lot.


DateMovieTicket Cost
Sat, May 5Avengers: Infinity War$12.00*
Sun, May 6You Were Never Really Here$15.00
Sun, May 13Tully$12.00*
Sun, May 20Deadpool 2$12.00*
Sat, May 26Solo$12.00*
Sun, May 27Book Club$15.69
Thu, June 6Upgrade$17.99
Sat, June 9Ocean's 8$16.99
Sun, June 10Won't You Be My Neighbor$15.00
Fri, June 15Hotel Artemis$15.99
Sat, June 16American Animals$17.40
Sun, June 17Nancy$12.00
Sat, June 23Tag$15.99
Sun, June 24Incredibles 2$7.99**
Sat, June 30Jurassic World$16.99
Sun, July 1Hereditary$12.00*
Wed, July 4Hover$12.00
Fri, July 6Sicario: Day of the Soldado$12.00*
Tue, July 10Sorry to Bother You$15.00
Fri, July 13Three Identical Strangers$12.00*
Sat, July 14Leave No Trace$16.99
Sun, July 15Ant-Man and the Wasp$12.00*
Thu, July 19Skyscraper$17.99
Sun, July 22Hotel Transylvania 3$8.49**
Sat, July 28The Equalizer 2$12.00*

* These were E-Tickets, which MoviePass seems to get a discount on.

** These were matinees. Why did MoviePass never encourage users to go to the much cheaper matinees?

These are the ticket stubs I have lying around. I might have missed one here or there. But the long and short of it is that for these three months, while I paid MoviePass $29.85 ($9.95 × 3) they paid several local movie theaters $379.16. This is so obviously a losing proposition for them.

Death of MoviePass?

Early last month, MoviePass introduced surge pricing. With little predictability, some movies would not be included but instead come with an extra charge around $4. When this happened, I felt like I might be canceling soon. You can see an extra strong focus on the (one) local theater with E-ticketing in my watching habits over July — they are immune to surge pricing.

Then in late July MoviePass ran out of money and had a service interruption. Somewhere in there, they started forcing users to tacke pictures of their ticket stubs. (To enforce the one-viewing-per-movie rule that they added in April.) Now they've introduced more, stricter limits: from "unlimited" (one per day) to three movies per month. And very unpredictable availability (black-outs). This already started a bit. Rumors seem to be perhaps that A) every showing before 5PM (why?) and B) basically every showing of every major movie (at least for the first few weeks) will simply be unavailable. Rumors also abound about availability disappearing in the time it takes to travel to the theater.


I'm not happy. Honestly, I expected MoviePass to be short lived when I first subscribed. But that was just over nine months ago. It's survived this long, but the writing is clearly on the wall now. I'll probably keep it and extract what value I can. I'm probably getting $10/mo from it.

On the other hand, this past year has re-introduced me to the theater experience. Given a theater full of them, you're guaranteed to run into strangers that have no problem talking, turning on their bright phone screens, and generally making the theater experience less than it could be. Worst is the fixed time slots. You can see a heavy focus on weekend viewings above. I find it's quite rare for a weekday showing to both not be overlapping work hours, and also not force an uncomfortably early or late dinner. So many weekday movies start at right around six o'clock. Eating early enough to finish and then get to the theater is far too early. Eating after forces a nearly 9PM meal — I don't want to be stuck in the theater with hunger distracting me from the movie!

So I feel very lukewarm now. If I walk to a theater and then can't get into the movie; if I find I can't get a seat from limited availability after black-outs; if I just get tired investing my time in finding a compatible showing time — I won't be too sad if MoviePass ends up in my past.

Game Boy Speaker Repair

2018-07-29 13:33 - Making

The Game Boy Color, open, with the new speaker installed.

I recently picked up a Game Boy Color for cheap, due to an issue with the sound. This is a common issue, the speakers wear out or break over time. Replacements are easy to come by, and cheap. Here's mine, with the new speaker in. Turns out the cheap and easy to get replacements aren't quite perfect. The original speaker (with the "Z" on the back, outside the console) comes in a plastic case, with some nubs on the outside to hold it in place. The new one fits in that space, but it's a bit smaller. A few dabs of hot glue should keep it from rattling around. And now my Game Boy Color has sound again!

Mason Jar Soap Dispenser

2018-07-27 17:11 - General

Back when I first moved into my current apartment I went on a home furnishings spree. I got a nice soap dispenser and tumbler set that matched and looked good in place. But over time, and with exposure to water, the soap dispenser wore out quite a bit. I could never quite find something I really liked again.

Until this. I found a dispenser mechanism specifically made to attach to a mason jar. It said "any" mason jar on it, so I bought sight-unseen across the internet some mason jars. Some wide mouthed mason jars. Which didn't fit the lid it came with. Thankfully I managed to trace the hole and cut it right out with a dremel, no fuss. I've always had a minor fascination with mason jars. They feel like a basic part of the at-home, DIY spirit. Now I've got a couple at home!

RFID Reader v2 Starts

2018-06-07 23:14 - Making

The first working prototype of the next version of my RFID reader project.

I've used Arduinos for some time now, for electronics projects. They're very easy to get started with, but a little bit limited. As I called out in the VFD clock project I did a few years ago, STM32 is a nice next step up. Pictured is a Maple Mini clone I still had from that project, easily available for around $5. Even better is the (as it's colloquially known) "Blue Pill board with almost the same functionality at around $2, and it's close cousin the "Black Pill". More speed and RAM and Flash, more peripherals. More exciting!

I designed a multi-headed RFID reader project, on top of an Arduino core. After some delays it's finally seen real usage, and revealed several opportunities for improvement. Mostly around the hardware, but if I'm going to redesign, I want to take the opportunity to revisit the software as well.

On my clock project, I found the SDK very detail-heavy and hard to work with. ST Microelectronics, which makes the STM32 chips, also makes a package called STM32CubeMX, a code generator that makes the SDK easier to consume. But it wants to output projects that specifically work with a small handful of professional (read: $$$) IDE packages. I recently discovered that Atollic TrueSTUDIO, which is in that list, is available for free download! I've spent a fair deal of free time, in small chunks, recently getting these all set up and working, and especially understood.

The plan is to take advantage of FreeRTOS, which the Cube tool can include with just a click, to handle scheduling and some other things to make this next version both faster and more stable. For now at least, I've got a proof of concept, working end-to-end, able to read cards and developed with a capable IDE with breakpoints and value inspection built right in.

Custom Game Case, from Unused Koozie

2018-05-29 17:12 - Making

The pair of source Koozies, with the Game Boy Micro on top. Top sewn up, bottom cut open, sides sewn. Flipped right side out. Tucked in. Flap closed.

Several months ago I got a pair of Koozies, which have sat unused since then. More recently I've pulled out my Game Boy Micro, which has no protective case, unlike most of my portable devices. I realized that the Koozie was the right sort of material, soft and padded, and almost exactly the right size. So I cut one side of the bottom off, sewed closed the top, and sewed the sides in closer, to be the right size. After flipping it right side out and confirming a good fit, I trimmed what was the bottom of the Koozie down, to be a flap which I can tuck in to cover the open side. It worked out quite nicely!

There was a fire, a couple buildings down from mine

2018-05-28 17:49 - General

There was a fire, a couple buildings down from mine, and I had quite the view of it. I actually smelled something first as I was sitting down to work on something, flipping the switch that enabled (among other things including the light I want) my soldering iron, but it was still in sleep mode of course. Not much later, I noticed this out the window:

That's the back side of this building. Seems pretty serious, enough to have broken through the roof. I'm totally fine, but things are a bit smelly.

Here's a story about it from the Post, which says over 100 fire fighters responded, two of which ended up with minor injuries.

My phone only lets me shoot 10 minutes of video at a time, so I had to (spend 20 minutes shooting, then) re-encode to stitch two clips together. But you get to see the blaze at (what seems) its worst through to being put out by New York's Bravest.