|
|
|
fixing that final leak
Tuesday, February 10 2026
Immediately after getting out of bed, I made myself a travel cup of tea, but had to use the microwave, as our hot-tea-water dispenser was not working. This was because it is supplied by our hot water system, which was still disabled after yesterday's plumbing near-fiasco. I sipped on that tea as I drove out to Lowes, arriving there before 8:30am. There were a lot of employees standing around with nothing yet to do, and they seemed to have all been briefed recently to say hello to customers. I went straight back to the tool section to look at pliers, hoping to find a pair with good leverage and very narrow jaws. As I'd lay in bed last night thinking about that one remaining leak, I figured out why rotating the heated fittings in place had failed to smear solder into whatever "dry patches" (solder-free areas where copper was pressed only against copper) were causing the leaks. What was probably happening was that the little 1.5 inch piece of 3/4 inch pipe interconnecting two fittings at that leak was acting as a pivot while one of the fittings rotated and was not itself moving with respect to the other fitting at the junction. This meant the solder was not being smeared into any dry patches responsible for the leak. To make that smearing happen, I had to reach between the fittings and grab that inner piece of pipe and rotate it. But the gaps between all the fittings were very narrow (which is the way I like things) and to the extent there was any space at all, it was because of a little spreading force applied by the other end of a tricky connection I'd made to connect up the half inch pipe coming from the output of the tankless electric hot water heater. This had resulted in maybe a quarter inch of exposed inner pipe if I maximized where the spreading between fittings at the site of the remaining leak. So I got the two pliers with the narrowest jaws, one of which was a Vise-Grip tool. I then got what I'd come for, a tank of MAPP gas, which now costs $27.
Back at the house, I dawdled for awhile in the living room, drinking tea and playing Spelling Bee until I had the courage to go be frustrated in the basement again. Who knew how many tries fixing that leak would require? In case I needed to rip that part of the manifold apart and redo the soldering, I'd bought another 3/4 to 1/2 inch copper pipe adapter.
Once I had the pipe heated up enough at the leaking junction to once more rotate the fittings with respect to one another, I reached in with a pair of long-reach needle-nosed pliers that I already had (since this now seemed like the best tool for the job). I managed to get it into the gap between fittings, spread them apart further to get a little more to grab onto, and then I grabbed that inner piece of pipe and turned it as best I could. I do not know for sure if it actually moved, but it probably did. Then I drenched that side of the junction first in flux and then in solder. I didn't care how wasteful or messy I was being. Whatever price I was paying for that was just a rounding error for what a plumber would charge to bail me out (a scenario that had seemed like a real possibility yesterday). When I thought I'd done about the best I could, I squirted everything with a squirt gun and tried pressurizing the hot water system. Finally, there were no leaks! I had overcome this adversity!
A sketch I made of what I was doing with the narrow-jawed pliers. Click to enlarge.
I tried to have ChatGPT draw a diagram of how that final leak was happening. Even after I insisted there were no leaks coming from the right side of the junction, there are still drops of blue remaining there and a line from the "Leak at Joint" caption pointing there. Also, despite multiple attempts to get ChatGPT not to have the leak seeming to flow from the top back into the pipe and somehow out of the middle of the fitting, it insisted on making the water do that. Furthermore, the gap where my narrow pliers needed to go is way too wide and there are threaded fittings in places that were sweat-only. Click to enlarge.
I immediately began cleaning up the enormous mess this work had resulted in. There was lots of water on the floor, in collecting buckets, and also atop whatever surface that could contain it. I didn't worry so much about that but instead focused on the disgusting mix of solder balls, solder flux, and drywall powder on top of the hot water tank. That had to be attacked with paint thinner, and before long the top of that water heater was looking almost factory-new. I always save stray globs of solder in these situations, since the plan is to some day use it to cast an object. (When I do that, hopefully impurities like drywall dust will float to the surface.) Then I cleaned up the basement hallway and, much later in the later in the day, I cleaned up the dining room table, which still had things on it from when Ray failed to install a new battery in a MacBook.
This evening, Gretchen made mashed potatoes but also left some unmashed for me, since I still have a lingering childhood aversion to potatoes in that form. This was be eaten with some gravy Gretchen had made last week. I ended up eating both forms of potatoes and was able to enjoy them, though this was mostly because of the deliciousness of the gravy.
Later this evening I was trying to install my fancy I2C bootloader on a small simple Arduino board called that "Pro Mini." It lacks a USB chip, so reprogramming it generally has to be done out-of-circuit. But with my I2C bootloader, it should be able to remain in-circuit, as now reflashing it is just another amazing thing that the ESP8266 master can do. I made the mistake, though, of asking ChatGPT to give me the avrdude command to set the fuses on the Pro Mini to make it compatible with the 8 MHz on-board crystal. When I issued the command, I lost contact with the board. It was only after I told ChatGPT this that it told me that there are two different crystal types, and if one picks the wrong fuse setting, it can make the microcontroller unreachable. Not to worry, though, ChatGPT reassured me. All I had to do to fix it was remove the existing crystal and then apply an 8 MHz square wave to the XTAL-in pin on the Atmega328p. But the crystal was surface-mount and had no exposed pins. Since it was looking like I probably destroyed the board given what my options were, I decided to try using a tiny drill bit to sever a trace on the PC board making a connection between the Atmega328p and the crystal. But even with that trace severed and an 8 MHz signal delicately applied from a working Atmega32A, I was unable to revive the board. I ended up tossing it into my jar of bricked microcontrollers. Perhaps some day I can desolder the Atmega328p and put it in some other circuit.
For linking purposes this article's URL is:
http://asecular.com/blog.php?260210 feedback
previous | next |
