|
|
|
overwhelmed by leaks
Monday, February 9 2026
This morning after a little tinkering and more back and forth with ChatGPT, I managed to get my I2C bootloader working on the Atmega644p. It seems ChatGPT had given me the wrong fuses to set for a 4 kilobyte bootloader, which caused the microcontroller to never enter the bootloader. The evidence for this was in the logs and the microcontroller behavior, which I explained in detail to ChatGPT, ultimately leading it to come up with better fuse directions. Not long after that, I managed to get my bootloader working on a third AVR microcontroller, the Atmega2560. It was feeling like I was on a roll.
Late this afternoon, I attempted to take Charlotte on a walk on the Stick Trail using a path through the snow that Gretchen had tromped down over the course of several walks. It was the first time I'd been there in weeks, and the walking wasn't too hard so long as I stuck with that path. But a brief foray onto the untrodded Gullies Path (it seems Crazy Dave hasn't been using his trail system since the big snow) wasn't pleasant. By then, Charlotte had abandoned me, so I trudged back home.
I decided then was the time for me to do something I'd been procrastinating on the hotwater replumbing project: the installation of that PEX manifold I'd built last week. By then I'd realized I didn't have enough space to attach it to the small amount of exposed one-inch outflow pipe above the hot water tank. I could go up into the inter-joist ceiling space a little (I was probably going to have to do that), but it seemed likely I was going to have to reorganize the things already attached there, particularly a side-loop that allows hot water to be just-in-time heated by a tankless electric heater. To get my manifold in place, I was going to have to remove T-fitting that connected the inch-thick pipe from the hot water heater to the inch-thick hot water supply while providing a half-inch side path for the electric system. There had been enough flex in the pipes that I'd managed to shoehorn this piece in close to twenty years ago, so I could probably also remove it. But I was worried about what would happen up in the ceiling if flames from the torch got up in there. So I opened up a little ceiling drywall and then cut pieces of sheet metal from galvanized cans (the kind that canned peaches had come in) and placed them around the pipe up in the ceiling as heat shields.
Getting removing that T-fitting wasn't easy and required (as it probably had required to install) a block of wood as a fulcrum and a pry bar to force the plumbing overhead to flex enough to get the pipe coming down into the fitting up and out of it. Then, once I had it removed, I had to use a series of improvised contrivances (pieces of wood) to push the pipe coming in from above up and to the side so I could get the massive one-inch T fitting at the root of the new manifold in place. Unlike the T-fitting I'd just removed, this one had one-inch pipe joining from all three directions so that more flow would be available to all the manifold's branches should they all demand hot water at once.
Once I had the manifold in place, I then I had to improvise the outflow connection for the just-in-time heater, whose half-inch destination had been lost when I'd removed that first fitting. I decided to run it instead to one of the PEX manifold branches, which meant removing the barbed PEX fitting on it, and using the now dead-end that had been going to the removed fitting as a new PEX connection. Hooking all this up required a couple half-inch street elbows, four inches of half-inch pipe, and a rarely-used 45 degree half-inch fitting. It was difficult to get solder to flow into one of the joints that was also a little hard to reach, but I hoped I would have my usual luck with soldering copper pipes: if I make any attempt at all and the pipe is hot enough, the solder is leak-free.
But the moment I turned on the water, it started shooting out of that hard-to reach joint. I had to drain the pipes and try again, this time drenching the joint in flux and then using copious amounts of solder. After that, the joint stopped leaking, but now I could see leaking happening in the joints above and perhaps below that big all-inch T at the root of the manifold. I'd been a little skeptical of those solders as well, as they hadn't really taken any solder when I tried to solder them, though they were pre-tinned from that T I'd removed.
Around this time, Gretchen returned from dinner, which had been take-away from the Garden Café eaten at Lynn & Gregg's house. I'd gone out so many times last week that I'd begged out of going to their place for dinner tonight. But I'd also wanted to get some of this hot water replumbing project behind me.
The next time I turned on the water, I could see leaks in multiple places on my manifold, meaning some of those solder joints hadn't been heated enough when I'd built it. But more worrying was a leak high on the inch-thick copper pipe going up into the ceiling, and it was from a fitting that predates all my plumbing work. Fixing that was going to require sending heat up into unknown contents of the inter-joist bay overhead. But it might also be impossible. At this point I wondered if I was going to be able to fix this mess I had gotten myself into. But unlike messes in other domains, this one directly impacted our life in our house. Until I'd fixed these leaks, we were not going to have hot water, and, due to the way some faucet mixers work, it was possible we wouldn't have cold water either. I came upstairs to report this news to Gretchen and to warn her that I was going to be in a bad mood and we wouldn't have water until I fixed the problem. "Have you eaten anything?" she asked, since I'm frequently crankiest when I am hungry. I'd had 120 mg of pseudoephedrine this morning and recently chugged some of a Mighty Mango smoothing, and that wasn't much. But I was in no mood to eat either.
While flowing more solder into that tricky leak up in the boiler room ceiling, I managed to kind a little fire on the insulation facing up there, but fortunately I had a squirt bottle ready for just this emergency and quickly extinguished the flames. When I next tested the plumbing, I was down to only two leaks. I was making progress!
The last of the leaks was a real bitch. It was 3/4 inch junction and no matter how much solder and flux I ran into it, it leaked the next time I turned on the water. Even rotating the fittings in place, something that usually smears molten solder across any unsoldered patches, did no good. On my last attempt to fix it this evening, my tank of MAPP gas reached its end, and I had no extra gas in the laboratory. I was forced to call it quits with one leak remaining. I was able to get cold water working again in the house by closing some valves, and, since we rarely need hot water, that was good enough until I could resume my project tomorrow. I didn't consider this a huge defeat given that I only had one leak left and it wasn't in a difficult-to-reach part of the plumbing.
Given all my success today with my bootloader, and given all the plumbing work I'd done, I decided I owed myself the right to drink. So I made myself back-to-back stranded loons, the drink I invented that consists of scotch poured into snow.
For linking purposes this article's URL is:
http://asecular.com/blog.php?260209 feedback
previous | next |
