The rant subject
Today’s rant is about a Glentronics Basement Watchdog backup sump pump system I’ve installed a while ago. The system overall is relatively simple- a 12V DC pump lives in the pit and is turned on by the controller when the water level rises above a level set by magnetic float sensors. The pump is fed by a big 12V Lead Acid battery that’s always charging. So when power goes out or the main pump fails, I’d have at least some pumping going on for a bit instead of me having to bail out manually. At least that was the idea.
For the first year or two the system sat there blinking happily. I’d test it occasionally by pressing the test button or moving the floats and the pump would kick in. During one of such tests I noticed that with the pump turning on, I’d also get a low battery alarm. That sounded weird after the system just sitting there charging all the time, so I started digging in. First thing I noticed was that the battery is charged in a pulse fashion- the charger turns on for a bit, then turns off. Looking at the current, it would go into the battery for a bit, then out of it! Upon a bit more thinking, I realized that this is how the unit tests for battery’s presence and relative health. The charge current was very small, and about on the order of discharge current. It was actually taking more energy out of the battery than putting into it. A nice discharger so to speak!
Battery voltage was suspicious as well- at most it was sitting at 12V while always charging. Too low for any kind of charging to actually occur. And finally I remembered that in the two or three years of owing the system I’ve never had to add water to the battery, even though the manual warned that it should be done pretty regularly, based on low water indicator. All this sounded like a system that is not charging correctly. So I called Glentronics tech support. The answer that came back was – “you need a new battery”. No matter how I tried to explain that I see no meaningful charging ever happening, the tech just kept insisting on his line. So I ran my own test. I disconnected the battery and charged it with a lab power supply, while measuring things with a WattsUP meter. After a pretty long charge, the battery took a good 20AH+ in and was sitting at a much happier voltage, it also ran a much larger number of pump cycles before hitting low voltage warning. So much for bad battery!
“Take it apart”
Well, clearly it’s time to crack it open.
A very old school design, no micro. Just relay, logic and discretes. Should be pretty reliable once designed and tested well. Well, let’s see how things are connected:
We basically have a diode, a TIP32 BJT and a series resistor between input wallpack and the battery. The system has an oscillator that pulses the big BJT to turn it on, while also driving the green “Charger” LED at the same time. Looking at the voltages, around the circuit, one thing becomes pretty obvious early on. Vin is very load dependent. As soon as the BJT turns on and tries to charge the battery, input supply sags quite a bit. That means while the battery is very low it’ll get a tiny bit of charge in, but as it gets fuller, the voltage left after the whole chain of parts is just too low to push any current into it. A closer look at the power supply that came with a system reveals an unregulated wallpack. There is our problem! Next step is to use a bench supply and see if that makes things better. At a steady 15V we get a nice charge and at the end of it, the output is never above 13.7V- a perfect value for float charge system like this. So the fix is very simple- replace a wallpack transformer with a suitable switching power supply. I used Cincon TR10R150-12A03 that cost me maybe 10$ at the time for a 15V 0.7A switcher.
My battery is now happily sitting at the proper float voltage of about 13.7V and actually requires adding water every once in a while. It seems that once again beancounters screwed up a reasonably good system, to save a few bucks on the wallpack.
Edit: There is now a discussion thread over on eevblog