- Joined
- Jan 10, 2019
- Messages
- 1,220
I built a wooden gear clock from plans, and it works great - mechanically. The designer added an automatic rewind feature so there is no weight to reset ever day or so. He chose a DC-powered gear motor attached to a mercury tilt switch, and this is where my issues with the design started.
The gear motor is loud, like, really annoying for a clock you want to enjoy, at least for me. I set out to fix that and wondered if I could change it to a 115 VAC-powered motor, since those are far quieter, maybe even powering it through the same mercury tilt switch, since the current is so small. The clock was modified and indeed, it was much quieter... but this revealed two new issues. One was that the mercury tilt switch sometimes didn't make electrical contact even when the mercury ran down to the contact pins - replacing it fixed that. The other issue was how the circuit worked different with the two motors. With the DC-powered motor, when power is removed, it would take a moment to "wind down" to a stop. This served as a sort of mechanical hysteresis, ensuring that the mercury switch tilted back a bit more, enough to ensure an open circuit. With the AC-powered switch, however, when the tilt switch become an open circuit, the motor stops instantly. This meant that the mercury was just barely clear of the contacts, so any tilting back towards it would cause it to "burp" a bit again, just enough to open the circuit, over and over again. The result is that even though the motor is much quieter, it bugs me that it's getting bumped every 15-30 seconds, unlike the (noisy) DC motor that would coast down with the mercury further away, so it would only run maybe once every 5-6 minutes.
So to this thread, I wondered how to solve this. Sure, an Arduino solution would certainly work, but then I'd have to apply 5V to that circuit in addition to 115 VAC to the motor. Add in programming and, meh. A 555 timer chip could also work, but the circuit starts getting complicated for what's supposed to be a simple and elegant project. I realized that the most elegant solution, requiring practically no parts, no programming, and no 5V supply, would be to modify - or make - a mercury switch with hysteresis. They're all the same design, a short glass tube with a bubble of mercury free to roll the length of the tube. At one end are two bare wire contacts. When the mercury switch is tilted in that direction, the mercury rolls down onto the contacts, and Bob's your uncle. As noted above, though, tilting it back very slowly can result in a crazy-sensitive switch, where tipping it just a hair one way opens the circuit, and a hair the other way closes it... way too frequent in this application.
I realized that if the tube could be curved slightly (an inverted letter "U" is an exaggeration), that could allow the mercury to back away from the contacts and run down the curved tube, ending up further away. Then as the clock runs, it starts tipping the switch back toward the contacts again, where the mercury has to make it up and over the bump before contact is made - that's the theory.
There are a couple ways to try it, the simplest being to heat up the existing glass envelop and press an indentation into the middle. Not sure how well that would go, and I worry about it exploding or cracking. That said, the contacts in the end of the envelop had the molten glass squeezed around them, so maybe reheating the tube would be okay. I'll probably try it, wearing safety gear and heating it over something to catch the mercury if the tube cracks. Another way would be to scribe a line around the middle, slightly off-axis, cleanly crack one end off, then rotate it slightly and glue it back on, ending up with something looking light an inverted "V". If it all goes sideways, I can try making my own, reusing the mercury, putting it into a plastic or silicone tube, curved appropriately. Pics as this progresses.
The gear motor is loud, like, really annoying for a clock you want to enjoy, at least for me. I set out to fix that and wondered if I could change it to a 115 VAC-powered motor, since those are far quieter, maybe even powering it through the same mercury tilt switch, since the current is so small. The clock was modified and indeed, it was much quieter... but this revealed two new issues. One was that the mercury tilt switch sometimes didn't make electrical contact even when the mercury ran down to the contact pins - replacing it fixed that. The other issue was how the circuit worked different with the two motors. With the DC-powered motor, when power is removed, it would take a moment to "wind down" to a stop. This served as a sort of mechanical hysteresis, ensuring that the mercury switch tilted back a bit more, enough to ensure an open circuit. With the AC-powered switch, however, when the tilt switch become an open circuit, the motor stops instantly. This meant that the mercury was just barely clear of the contacts, so any tilting back towards it would cause it to "burp" a bit again, just enough to open the circuit, over and over again. The result is that even though the motor is much quieter, it bugs me that it's getting bumped every 15-30 seconds, unlike the (noisy) DC motor that would coast down with the mercury further away, so it would only run maybe once every 5-6 minutes.
So to this thread, I wondered how to solve this. Sure, an Arduino solution would certainly work, but then I'd have to apply 5V to that circuit in addition to 115 VAC to the motor. Add in programming and, meh. A 555 timer chip could also work, but the circuit starts getting complicated for what's supposed to be a simple and elegant project. I realized that the most elegant solution, requiring practically no parts, no programming, and no 5V supply, would be to modify - or make - a mercury switch with hysteresis. They're all the same design, a short glass tube with a bubble of mercury free to roll the length of the tube. At one end are two bare wire contacts. When the mercury switch is tilted in that direction, the mercury rolls down onto the contacts, and Bob's your uncle. As noted above, though, tilting it back very slowly can result in a crazy-sensitive switch, where tipping it just a hair one way opens the circuit, and a hair the other way closes it... way too frequent in this application.
I realized that if the tube could be curved slightly (an inverted letter "U" is an exaggeration), that could allow the mercury to back away from the contacts and run down the curved tube, ending up further away. Then as the clock runs, it starts tipping the switch back toward the contacts again, where the mercury has to make it up and over the bump before contact is made - that's the theory.
There are a couple ways to try it, the simplest being to heat up the existing glass envelop and press an indentation into the middle. Not sure how well that would go, and I worry about it exploding or cracking. That said, the contacts in the end of the envelop had the molten glass squeezed around them, so maybe reheating the tube would be okay. I'll probably try it, wearing safety gear and heating it over something to catch the mercury if the tube cracks. Another way would be to scribe a line around the middle, slightly off-axis, cleanly crack one end off, then rotate it slightly and glue it back on, ending up with something looking light an inverted "V". If it all goes sideways, I can try making my own, reusing the mercury, putting it into a plastic or silicone tube, curved appropriately. Pics as this progresses.
Last edited:
