power line capacitors
Monday, December 24 2012
Since the ruptured expansion tank fiasco the other day, I've been having trouble getting fluid to circulate in the the solar loop. Even in the winter there is heat to be collected, and on a sunny day without circulation, temperatures in the panels can rise to over 150 degrees even at this time of year. Today I found myself trying all the usual tricks to get circulation working again. Usually the problem is air trapped in the line, and I can almost always fix that by adding fluid at the top of the system (on the solar deck) using a funnel. Today, though, this didn't work, so I found myself trying various things in the basement. These all involved closing some valves while opening others. But, with the exception of a tiny trickle of scalding hot water that managed to come down from the panel, it was all a waste of time. My feeling is that the bubbles in the pipes are too big this time for the usual techniques to flush out. Ideally I'd have drain cocks at all the high points into which I could introduce fluid, but it would be a mess to try to add them now.
Today Gretchen called from the City and we discussed Marie (aka "the Baby"). To me, she seemed to be doing a lot better than she had been. I even saw her grooming herself today, and her appetite seems to have returned. It seems our housecall vet had been correct after all. So I called the Hurley Vet and canceled the Baby's euthanasia appointment. And so, like the Energizer Bunny (if, that is, you can picture an Energizer Bunny that has been run over by a truck and then carefully glued back together), the Baby lives on.
To debug my new fuel-level-reading slave Arduino board, today I added a four-pin serial port (adhering to yet another standard that I am imposing on myself, this one following the example set by whoever designed the four-pin serial port for the Pogoplug). When I hooked this serial port up to my computer using a TTL-to-RS232 level shifter, I could see that the poor little slave was trying to boot up, but it was accidentally resetting itself dozens of times per second. Whenever it starts up, you see, it sends the message "just booted" out through its serial line. But what I was seeing was text that looked like this: "just bjust bootejusjust boojustjjjustjust bootedjustjusjust." What was causing it to behave like this? After confirming that the problem was the board I'd made and not the firmware, I tried adding the little bits that I'd left out. These little bits were so small and insignificant that I didn't think they were required, though they're always shown in the documentation. These included the tiny 22 picofarad capacitors one is supposed to connect to the timing crystal and the larger capacitors one is supposed to put across the power rails. I added those little picofarad capacitors first, but they had no effect. Then I added two largish electrolytic capacitors to the power rails, whereupon the slave booted up perfectly and remained up without rebooting itself at all. I've seen improvements in reliablity from adding power rail capacitors1, but none as dramatic as this. And once I had that problem fixed, I noticed that I hadn't actually been grounding the Atmega168 on the correct pin (I'd been using 21 instead of 22), though somehow it had managed to work anyway. That's a little like driving a car around for a couple hours only to discover that the starter motor and not the internal-combustion engine has been its source of mechanical energy.
1Back in 1984, after building a replacement power supply for my primitive VIC-20 computer, I noticed that the computer would lock up every time a pump in the basement would kick on. This was evidently being caused by a small power sag. I fixed this by adding a number of large electrolytic capacitors both inside the VIC-20 and inside its custom power supply.
For linking purposes this article's URL is:feedback
previous | next