2019-09-16 23:31 - General
I've been a TiVo user for almost exactly a decade. It's pretty great, but as is it's tied to cable service, because a few years back (and before my most recent (and I'm pretty sure only) upgrade) they split their product line so that the device only supports cable or OTA broadcast, not both. More things are going online only (I'm looking at you, "CBS All Access" locking the new Star Trek behind a $6/mo charge that certainly isn't worth it for one show!) and I'm hearing more and more about "cord cutting". Might it be for me?
So I got an antenna. Broadcast TV is legitimately free, after all. Free over-the-air TV plus a streaming package or two should be more content than I need, for less than I'm paying now, right?. This is a "Channel Master FLATenna" which is probably just like all the similar "sheet of paper" looking modern digital antennas, but was reviewed well and only costs $10 (plus $10 shipping). I live in Manhattan, so signal coverage should be great, right?
I put it at the very top of my window at first, which was mostly fine, but I couldn't get ABC. I summoned distant memories from my childhood of playing with TV antennas, and started playing with positions. I first tried the wall above the TV because it's an easy location and 90 degrees offset from the window. Then I tried a couple others, including an awkward compound angle hanging across the corner and pointed up, and lots of other similar variations. Blue painters tape and shipping bubbles certainly isn't an awesome permanent solution, but for now it's been useful to be able to move it around for easy experimentation.
Of course with this small indoor antenna, across various locations I can receive more channels than I can get in any one of them. This is the best compromise I've found so far. Here I get all the big national stations (CBS, NBC, FOX, and ABC), plus some more local stations, plus one of the two local PBSes (WLIW yes, WNET no). I spent quite a bit of the weekend popping the antenna into various places, then running scans and checking the signal strength across the dial. Of course the evangelical stations are the only ones that come in super strong regardless of location.
And since we're in the digital broadcast era, each of them come with a couple to several substations. My favorite discovery, which I was previously unaware of, is Quest TV. With shows like Modern Marvels and Factory Made it should be a good stand-in for losing Discovery and Science.
Along the way I've also discovered that my TV has basic PVR functionality built in! It's disabled in the USA (why?) but that can be fixed. It's basic and seems to only work while the TV is on, but it helps evaluate things.
It will require being happy with more SD content, but I can live with that. There appears to be a rather inconsistent handling of 4:3 content, though. Some of it is letterboxed, so should be zoomed. Some is pillar-barred, so should probably be left as is. Some others appear to be anamorphic and require stretching to 16:9. And my Samsung TV at least can't figure this out automatically.
Another pleasantly surprising detail is the availability of program guide data. I've never hooked this TV to the internet. Especially since it sometimes blanks out the video preview, the guide data is clearly coming over the air. I assume there's a standard tied in to digital broadcasting. I'm curious about it but haven't managed to find any clear information about this online yet.
Right now I'm looking at 39 non-junk channels, which is pretty good. What remains is to experiment with some sort of DVR for broadcast TV, and streaming solutions.
2019-08-01 22:24 - Making
After I built the bedsofa, I rearranged the bed and sleeping area around it. It had to go up against a wall to avoid tipping over when I lean on it. But the head of the bed was previously in open space. When I turned the bed, this made the light there, hanging from the wall, end up at the foot. Not a great way to illuminate the area, like for reading.
For a while, I pointed it at the ceiling and lived with the indirect light. The existing fixture won't work well in the new space that I've got. I've looked for fixtures that would work better, and found nothing pleasing. But I also have this strip of LED lights, which I intended to install in my 3D printer. (And, of course, never quite got around to doing.) It puts out a ton of efficient light, even just a small part of the reel I've got. The picture above is the "prototype", just taped to the wall. It's intended to run from 12V (and I picked that because the 3D printer has ample 12V power available).
In bed, I actually want very subdued lighting. A half meter of that LED strip running at 12V is really brighter than I want. I've got plenty of spare AC adapters lying about, but they're all too high or too low. But I also have some spare DC/DC buck converters! So now I've got a 12V AC adapter, into a buck converter that's infinitely variable, with its output powering the lights. I can tune it to the exact brightness I want. Perfect.
I recently found this ultra-cheap extruded aluminum channel. Pop an LED strip into it and instant light fixture. I got this V shaped version which mounts at a 45 degree angle. I intend to point it towards the ceiling for primarily indirect light.
2019-07-30 18:27 - Making
I got a 3D printer just over 3 years ago. Some months back it stopped working. I've got an OrangePi running OctoPrint to control it. I did a system update of that, and it stopped working. I unhooked everything and tried to figure out why, and failed. And gave up for quite a while.
I've got a thing I want to 3D print, now. So I took another stab. First day, same result: I was trying to set up everything from scratch, but I couldn't get the thing to boot. After switching out every spare power supply and SD card I had, still no dice. Of course, the whole time I wasn't thinking about the cable that carries the power. Turns out that was it. Once I swapped in a different cable, everything was working again. Set up OctoPrint again from scratch, and it's all great.
I just completed a 10mm cube test print, which worked fine. The machine needs a little TLC, it's squeaking a bit as it moves. But it works! Unfortunately, my desired object is too big to produce on my machine. So I'll need to find another way.
2019-07-25 00:23 - Gaming
As I'm wont to tell most people that will listen, I've got a deep backlog of video games lined up to play. It's been growing deeper recently, as I let some new games jump to the front of the line. I played Stardew Valley recently (it's new to me!) and loved it. A bit earlier, Mini Metro. Some Smash Bros, Baba is You, Steamworld Dig. I was having fun with the Switch.
For the Fourth of July week, I got right down to the backlog though. A PS1 game had been sitting eagerly at the front of the line for some time: Chrono Cross. The sequel to the amazing action RPG Chrono Trigger. I was so underwhelmed that, after thirty hours or so in, I just stopped. This one is a standard turn-based JRPG, which I've played plenty of. But it's not a good one. The battle mechanics were very slow and repetitive, lots of superfluous animations taking up time. A battle mechanic that's a bit clever, but tedious and time consuming. And the story didn't grab me, so I couldn't maintain enthusiasm.
Next on the list was BioShock Infinite. I've played the previous two in the series. Simply put: it's more of the same, with little to recommend it, and continues the downward trend of the series, in my opinion. It's just yet another FPS game, with a strange companion bolted on, who adds little. This one only took a couple weeks, beginning to end.
Which brings me to my third game this month: Assassin's Creed: Brotherhood. Again the third in a series. This time, the first game was the worst. The second got better. The third so far adds little to nothing, it feels like a cookie cutter copy of the second. Like I said I "never" do this, but I'm already thinking of giving up on it (rather than playing to the end), again.
Could I, should I, hope that the next game on the list is better?
2019-04-21 13:07 - Making
I posted about this project just over two months (!) ago, around when I was getting started in earnest. That was starting the cushions, which were the "scariest" part, where I've got the least relevant experience. Since then in my weekends and evenings (on and off), with some time for shipping delays and generally enjoying myself otherwise, I've completed the rest.
This post is the story version of how that all went! It will be a little bit out of order so that the story goes well. If you're truly curious, look at the un-adjusted file names in all the images, the dates and times are burned in there and generally represent when the pictured step happened.
Plans
I just wanted to call out this original paper sketch that I did at the very beginning of the project. I've changed a few things, and it's not to scale, but most of the idea was there from the very beginning.
Cushions
As previously mentioned, I started by experimenting just a bit with the piping. When I was happy that would work, I did them for real: four long single sections to go around the edges. (Here was one of my first big mistakes: they weren't quite long enough. Three of them got extended; one was already stitched in when I realized, so I've got a small gap. Thankfully at the bottom where it doesn't show much.) Next up was the zipper panels. I'm not sure what exactly I did wrong, but they came out very uneven. Not so much of a mistake as an "apprentice mark", here. You can also see the extreme degree of raveling at the edges of the zipper panel, which plagued the sewing part of this project.
As hinted in the piping failure, the next step was stitching up the shells for the cushions. This went almost completely as intended. A couple pieces were a bit small, not really enough spare material to hem. I think this is because (or, worse because) upholstery fabric is stronger, with thicker threads. Which don't weave so tightly as a result, and unravel easily. The second issue was sewing up the corners: with that quarter-inch-thick piping in the way, those were very difficult stitches to make. More apprentice marks here: a few corners are perfect, but most are uneven in random directions, just a bit.
Due to the unraveling, I took the time to (order, and) sew in binding tape around all the seams inside the cushions. Things would probably be fine without this extra layer, but it adds peace of mind at least. And another mistake: when stitching them in, I got a bit of the side panel across the top of one cushion. So it's a little narrower, and has an extra unnecessary stitch visible. Ah, well.
With only spare material for this project and "junk drawer" items, I put together this hot wire cutter jig. You can see the very rough edge that I hacked out of the foam with a knife, just to get close to the proper size -- my wire cutter has only a few inches of throat. Then, the wire-cut edge: perfectly straight and smooth. With the foam cut out to just the right shape and size, stuffing the cushion shells went very well! Except for one ripped seam I had to repair, a raveled edge which caused a weak stitch. This was repaired with some "fusible bonding web" because I didn't want to re-make the whole thing! You can see one piece of binding material in the corner of this picture, where I knew the edge (and thus stitch) was weak. Apparently I needed a bit more! I ripped out most of this seam, bonded the extra material to it, then re-stitched it.
Structure
With the cushion shells completed, I was confident that I could make this project really work. So I bought a bunch of two-by-fours. These were cut up at the workshop in the office -- a great perk. And then bunched up to roll home by hand cart. Since they were smaller in their cut-up form, this was easier than rolling the raw stock from the store to the office -- thankfully that was only a few blocks.
I was overconfident. The complex angled pieces did not come out right in that first batch of cuts, going only by measurements. I got lucky here in that I planned for eight foot boards, but ended up with ten footers. Each had just enough extra slack that I got my four main five-foot-long boards cut out of only two of them. This left an extra unused board, just enough material to re-cut these pieces.
For this second try, instead of cutting the piece to fixed measurements, I clamped up the rest of the pieces in place and scribed out the exact shape I needed, then cut from that template. These didn't end up perfect, but A) good enough and B) that's a small theme of this project.
At home, the structure is all screwed together. Pocket holes are used liberally where possible, and some simple butt joints as well. The back was assembled first, then pieces were screwed into that. Then the shorter front section onto that. This was the main shape that was templated above, so the angled sections, support for the back rest, were added and the main structure was complete!
With the main structure complete, the next task was to attach the seat backs. These were rough cut from plywood, then marked and clamped in place. The cuts were scribed to the actual structure: nothing came out perfectly square or perfectly to plan, but this let me get the pieces cut to match! The top and bottom cut are angled to match the recline, so they end up with "flat" edges relative to everything else, which is nice.
On recommendation of a coworker that I've been chatting with about the project, especially the sewing bits, I got some batting to wrap the cushion foam in. The smallest unit was quite large. I had been considering some sort of foam layer to soften the edges of the structure with. All that spare batting took the role! In hindsight, I should have chamfered a few of the wooden corners before this, but it's worked out well enough overall.
Upholstery
The upholstery step was also a new skill to execute. It's done with the same exact fabric as the cushions. My original idea was a solid color, to make both the cushions and the upholstery easier: no opportunity for mistakes laying out the pattern. I ended up with a pattern. It went almost completely fine. The entire lower section, beneath the seats, is one wide piece. This went well, except that the sides stretched a bit unevenly, leaving some warp to the pattern there. There's also a little spare material wrapped around the spots I expected to be tricky: in case I completely miss something, this lower layer (rather than the white batting) is what will peek through.
Next was the seat backs. They went quite well overall, except due to the recline angle, the pattern is angled next to them. That's fine. The bottom edge is tacked in with a cardboard strip which keeps that edge nice and straight. Some metal tack strips bridge the empty space front to back over a gap where there's no wood structure, holding a clean edge. The material is folded up over that, then wrapped around the structure and stapled from the back, leaving a nice surface on the front and sides.
Clean. The last piece to upholster is the bridge across the top. I took two tries at this, not pictured. The first was only tacked at the bottom and wrapped around, a bit too loose. So I took the folding metal clamps from my upholstery tack set: with the inside tacked across the bottom, this holds the left and right edges in place by being tacked in underneath, folding the material over, then bending the metal down to hold it all in place. With the remaining edge tacked again from the bottom.
The back will be slightly visible, as the wall it will be resting against is only partial, with the sleeping loft looking out from above on both sides. So the back (and bottom) is being covered in a black cambric dust cover fabric. It's got a small corner folded back, for the electrical cord (read ahead!) to stick out of. Like the rest of this project, it's got small issues. Not very visible in the far side of this low-lighting picture is a fair deal of bunching (at the left side when looking at the back). Which of course ended up being the more visible side. I might go back and re-touch that.
Shelves
With the main upholstery done, the shelves were the only significant remaining piece. Like with the seat backs, nothing is perfectly square here. I used "story sticks" to measure the front and back widths, they were a sixteenth to an eighth of an inch different. Then I cut out blanks, with the front edge angled again to match the recline, too deep on purpose. Once they generally fit, with the front edge lined up nicely, the back edge was marked to match the structure and cut square.
Both shelves are held in place with half-wide two by four pieces screwed into the main structure, which they're also screwed into from below. The top shelf was cut a little wide, so I used some scraps of wood to stretch the opening a bit. This made it easier to insert the shelf without stretching the fabric too much. For the bottom and top, it was a tricky operation of putting the shelf in the right place, so the support pieces could be clamped in place. With those clamps holding, verify the placement is good with a level. Then screw in the support pieces, and screw the shelf to the supports from underneath.
The lower shelf is flush with the main structure so its bottom side and supports are invisible. The upper shelf of course is visible from all sides, as are its support pieces. So when finishing, the top shelf was finished on all sides, as were its support pieces to match. Both the two by fours and the plywood were near paper white, which didn't look super nice next to the relatively dark blue fabric. I used some Minwax "Polyshades" which I already had from a project a few years ago. This is a stain and (polyurethane) finish in one. I probably would have done separate if I was buying specific for this project, but this worked out pretty well. I couldn't choose the color, because I was using the spare I already had: "Mission Oak", which is quite dark.
I'd change one thing if I could go back, though: I did the smaller top shelf first, and I brushed on the finish (like the instructions said to!). This gave a very thick and dark coat. For the second shelf, I wiped the finish on on with a rag. This gave much thinner and finer coats, and a lighter color I would have likely preferred. It took eight hours for each coat to dry, but I ended up putting four coats on the lower shelf, to get it to a depth of color to match the already finished top shelf. I'd have done only two or three wiped-on coats on both, if I knew how that would end up ahead of time.
Electrical
This is going up in my sleeping loft. There's only two sides of this loft which have walls this can rest against (for support), and both of them have electrical outlets. Whichever one this goes in front of I'd like to use: I plug in various eBook readers, tablets, game systems, etc. to charge. In my first apartment, I had a nice table/hutch in the kitchen: It had a small shelf above and more importantly an electrical outlet built into the surface, with a cord to plug into the wall. It made a convenient place to plug in a mixer or other appliance, while cooking. Above you can see a crowded/messy picture from when I was near packed up to move out, and a closer in picture of the outlet in the surface. I decided to do something similar to that for this project.
The bedsofa will block my outlet, but it will have its own replacement outlet built in. I splurged here and I got an outlet with USB ports also built-in, and a switch to control things. The electrical box I got was designed (as far as I can tell) to fit through half inch drywall, but I'm putting it through three quarter inch plywood, so I made up some shims from some old quarter inch wood stock to fill the gap, which helped everything screw down tightly and line up nicely.
I made a small accident here: I've got a two-gang box. The first gang is a switch/single outlet, while the second gang is two outlets plus two USB charging ports. I wired it up with the intention to have the switch disable everything (and any possible vampire power). But what I actually did is just switch the second outlet. The single outlet with the switch is permanently on. It wasn't intentional, but this is nice and flexible, so I'm keeping it.
The Glamour
Here it is, completely in place. When it was being assembled on my coffee table, err...work bench, I realized it wasn't quite square, it would wobble a bit. I got adjustable screw in feet, to make it possible to leave it perfectly stable and flat. In addition, there's two of these feet sticking out the back of the top. These rest against the wall, so that it remains stable and upright when I sit back on it.
Cost
Look at this as either a passion project or explanation of why furniture is expensive. Or both. Some of this, especially the wood, is down to Manhattan prices. I've rounded things to the nearest dollar after tax, just to make the accounting easier. Either way, I spent:
| Item |
Cost |
Where |
| Structure |
|---|
| Two by fours (six, 10 foot each) |
$58 |
Prince Lumber (Local) |
| Baltic birch plywood (3/4 inch, 5 by 5 foot) |
$57 |
Prince Lumber (Local) |
| Stain/Finish |
$0 |
the "junk" pile |
| Subtotal |
$115 |
|
| Upholstery and Cushions |
|---|
| Fabric (55" by 7 yards) |
$56 |
eBay |
| Zipper |
$13 |
Jet |
| Cambric (dust cover) |
$12 |
Jet |
| Welt cord (too big, too short) |
$7 |
eBay |
| Welt cord (3mm x 50yd) |
$9 |
Jet |
| Cushion foam (3x24x72") |
$40 |
Amazon |
| Dacron Batting (48" x 5yd) |
$24 |
Amazon |
| Binding Tape |
$10 |
Amazon |
| Tack Strip Bundle |
$25 |
Amazon |
| Subtotal |
$196 |
|
| Misc |
|---|
| Furniture levelers |
$11 |
Amazon |
| Screws |
$9 |
Home Depot |
| #000 Steel Wool |
$5 |
Home Depot |
| Subtotal |
$25 |
|
| Electrical |
|---|
| Power strip (for cord only) |
$5 |
Home Depot |
| Electrical box |
$7 |
Home Depot |
| Face plate |
$3 |
Home Depot |
| Switch/outlet |
$13 |
Home Depot |
| Outlet w/ USB |
$25 |
Home Depot |
| Wire nuts |
$0 |
the "junk" pile |
| Electrical cable |
$0 |
the "junk" pile |
| Subtotal |
$53 |
|
| Tools |
|---|
| Countersink bits |
$12 |
Jet |
| Bar Clamp Set |
$22 |
Home Depot |
| Subtotal |
$34 |
|
For a grand total of $389, $423 if you include the tools that I bought specifically for this project. Phew.
I did a lot more impulse buying and splurging than I normally would, while working on this. But not only is it a useful durable piece of furniture that I'll keep, it was also essentially entertainment budget: I got to flex several old and new making skills.
I used most of the screws. Several of the upholstery pieces were mostly used. The zipper was just barely, as was the welt cord. In both cases, these were the most economical choices, anyway. I used only a tiny bit of the batting (around the foam, in the cushions). I've also got almost half of the baltic birch plywood left, which should easily turn into some new project in the future!
2019-04-02 18:52 - Making
My bed sofa project is progressing nicely. I expect to put together a detailed post, once it's complete. Over the weekend I picked up some nice plywood, for the seat backs and center shelves. Cutting those pieces right is challenging. I've got the main frame assembled, at home. I've got tools, at the shop at work. I want to transport as little as possible. And I also have some angles to cut, and hopefully get just right.
I've seen one of these before, probably on This Old House, and that's the article I found when I went looking for the name of the thing. I decided to make one quickly today. Laser cut a shape that allows me to match angles like the commercial equivalents. Tighten the nut to just lock it down, then push it into the angle to squeeze it perfectly flush. Then just scribe the same line onto the real material. (After tracing it onto a piece of paper, to measure more easily, it seems I'm at about 22.5°. I was aiming for 20°, but shifted some of my measurements around. And some of the cuts were probably not perfectly to plan!)
In this case, I'm trying to cut the top and bottom of the seat back, which is an angled piece. So that the cuts are level, flush with the neighboring pieces of the frame. I've got a jig saw at home, which I've used to divide the pieces down to the smallest possible size, for transport. The band saw in the shop has a tilting table. I can match that table to this angle, and get exactly the cut I want. The only challenge will be getting the right height, after cutting this tiny bit off.
I could have spent about ten bucks to buy one of these, but it was much nicer to make it! This was all spare materials except $0.33 for the thumb screw at a local hardware store on the way home! Access to a laser cutter really helped.
2019-02-13 22:39 - Making
I've got nice high ceilings in my apartment. And at one end a raised sleeping loft. There's only three or four feet up there, but it's plenty of room to lie down in. Except the bed is actually just a mattress on the "floor". I've made it work so far.
But, technically, it's just a mattress on the floor. There isn't the height for a "real" bed. I've got a chair next to the bed, but I don't end up using it much. I'd rather sit up in bed. I thought about it for a while, and I had an idea.
It's not at all unusual to take a sofa, add a mattress, and call it a sofa bed. So why can't I take a mattress, add a sofa (back) and call it a bed sofa? I can build in a couple cushions at a gentle recline, a shelf to hold an alarm clock and a few books, and sit up comfortably in bed! Read a book, under the covers, then just lie down when I'm ready to sleep. This would make me happy. And it's a nice project with some opportunity for design (done), carpentry (I can handle that), and upholstery (oooh, new skill to learn!).
The construction is primarily two-by-fours. The seat backs might be plywood, but might be zig zag springs. Which are normally used for seats, but I wonder if I'd like that bit of give to the back. Some plywood or wide solid boards will make the shelves, and the whole surface (except the shelves, I think) will be covered in upholstery fabric. Then I make some custom cushions, with the same fabric over foam. Voila!
It's the cushions that are the exciting bit. I can sew. I think. This will be a special new challenge. I want the edges to have piping like real upholstery, so I need to figure out how to do that. Time for experimentation!
My bed sheets and cover are dark blue. I was looking for some fabric that would go well. Expecting a solid color (because the sewing would be easier, with no pattern to worry about lining up). And clearly from the design I was thinking grey might go well, but I found this blue fabric with a geometric pattern that really appealed to me. So I got several yards of it. Plus some "welt cord". The idea is to sew a tight loop of fabric around the cord, then you've got a matching "pipe" shape to sew into the corner seams, and it looks nice. So I gave it a shot. This was a quick and dirty first attempt, which turned out quite well except for not getting a very straight seam.
Then I sewed that up into a seam. I got one of the pieces inside out, but for this simple test that's fine! This is sub-optimal in a few small ways, and I think I know how to improve it. Can't ask for much more! Hung over the edge of the table to roughly approximate the real shape, it's even serviceably straight, despite putting almost no effort in there at all. A couple more practice runs and I should be ready to try my hand on a real cushion. If that goes well, it's time to buy foam and two by fours!
2019-02-05 22:58 - Making
The Retro Duo Portable is the portable version of Retro-Bit's clone consoles, which play original game cartridges. A while back I rescued one from the trash. It didn't (quite) work. It could play games, but the screen constantly flickered. I had plenty of early ideas, and for the past few weeks I've spent an hour here and there on evenings and weekends trying to repair it.
One of the first things I did was just open it, to try to see what was going on inside. I don't remember specifically what I tried first, but it was minor. I tested the system again before fully re-assembling, and voila! It worked great! So I reassembled it and ... it wasn't working again. After a couple experiments, I found that if I pushed on the front, between the face buttons and the screen, with modest pressure (in fact, pushing the buttons themselves could trigger the failure as well) then the screen would start flickering. One of the screws that holds the thing together would, itself, apply this same pressure.
With it "working" (and open) I could confirm that really what's happening is a constant trigger of the "contrast" (brightness, really) button's control. In fact, if you hold that button down, it stops. Inside, there is a weak pull-up resistor on a signal line, and pressing the button closes a switch which (strongly) shorts that line to ground. The system triggers on the change of the signal, and somehow only when being pressed, some other signal line was leaking into the brightness control, causing constant and repeated changes of value, which it interpreted as constant pushes of the contrast button.
The first thing I tried was liberal application of tape (kapton and "electrical") to prevent shorts. In the close up picture you can still see a few pieces, which weren't worth removing now that they're in place. I tried all sorts of clever things with the electronics to try to isolate exactly where the failure is coming from, and eliminate it. None of these attempts helped at all. Somewhat in desperation, with no reason to believe it would help — but I was hoping for some sort of structural, rather than electrical, support — I stuck a bit of card inside. Especially in the second close up picture, you can see that there's one spot that it's pinched enough to leave marks on the card. (And the same thing in a second spot, after trimming and rotating the card to fit better, without blocking any of the screws.) Some thin kapton tape (and, I think, a layer of electrical tape on the other side during one try) just wasn't enough. But this card in this location separates or supports something in the right way, and it's all working again! Perhaps with use it will wear down? For now, it's nice to be repaired.
I could buy one of these brand new for around $80. But now, it will always be that one I rescued from the trash, and fixed up.
2019-01-29 00:36 - Making
I've had a Wii at my Mom's for several years. It's still fun to break out Bowling or Dr. Mario once in a while when visiting. This one came used, with rechargeable batteries for the WiiMotes, and a cradle to charge them. Last visit, the cradle stopped charging correctly. It wouldn't get power as expected. Wiggling the plug could help a bit, but not reliably. So I brought it home with me to take a look, which I finally did this weekend. (Delayed writing this post a bit for no good reason.
This little pictured board is one of two inside. This just holds the power connector and switch, and routes some wires out: to the other board which handles battery charging, and to some LEDs which indicate that the thing is switched on. From the symptoms, I suspected a bad power connection. First thing I tried was re-flowing the three pins on the barrel jack. Didn't help. The second thing I tried was wicking the solder away from the pads. I got lucky, the first one I tried was the problem. Look very closely and you might be able to see a bit of curling near the top and bottom (as pictured) of the bare pad near the middle. It's lifted away from the trace, so it doesn't conduct the power to where it needs to go.
It took a bit of investigation to hit this "lucky" find. When my first attempt didn't work, I started figuring out what all the connections do. There's more connections (five) on the power switch than it would seem to need (two!), and more wires leading off than seemed strictly necessary. Eventually I found that this pin of the barrel jack did not have a connection over to any of the other places that it should have.
Here's a view of everything else. There's a second (pair of, pass-through) power connector(s). And the fix: an extra wire added in, to carry the power that the PCB trace isn't carrying anymore. I didn't bring the batteries with me to confirm, but the power indicator LEDs light up reliably now. I re-assembled everything with gobs of hot glue for both insulation and hopefully some more mechanical support of the power connector, so such a failure is a bit less likely to re-occur.
