Dirt-simple time switch with one hour period
klugesmith, Mon May 07 2018, 08:47PMHere's one easy way to make a switch open and close once every hour, with duty cycle of about 50%.
We decided to start with a battery-powered quartz regulated wall clock.
This is a good one: it's rectangular, lightweight, and has no clear plastic cover.
A few bent paper clips suspend it like a seesaw, or teeter totter, in a roughly horizontal plane.
A coin resting on the minute hand serves as an eccentric weight, making the whole clock rock back and forth each hour.
The motion operates an ordinary mercury tilt switch, whose fluid provides some hysteresis.
Electrical connections to moving subassembly are through the paper clips.
Characterization report to follow.
Re: Dirt-simple time switch with one hour period
Dr. Slack, Tue May 08 2018, 08:57AM
very good, impressed.
Now how to do it for 30 minutes?
Dr. Slack, Tue May 08 2018, 08:57AM
very good, impressed.
Now how to do it for 30 minutes?
Re: Dirt-simple time switch with one hour period
Ash Small, Tue May 08 2018, 02:14PM
Gears
EDIT: The C of G of any gears used needs to be on the centreline.
Ash Small, Tue May 08 2018, 02:14PM
Dr. Slack wrote ...
very good, impressed.
Now how to do it for 30 minutes?
very good, impressed.
Now how to do it for 30 minutes?
Gears
EDIT: The C of G of any gears used needs to be on the centreline.
Re: Dirt-simple time switch with one hour period
klugesmith, Thu May 10 2018, 04:52AM
An EL-USB-5 event data logger soon showed that the switch had issues.
First chart is for the unit roughly as pictured, on a desktop at work. Fair amount of vibration in the environment, eccentric weight too small, maybe not enough damping. Could a tipping event have so much overshoot that the rebound tips it back?
Second chart shows the switch in a stable, low vibration place, with a quarter dollar in place of the penny. Good times are good, but bad times have some bit errors!
Before the third chart I removed the hour hand. I guess its tipping moment was enough to perturb the 1-hour-cycle's casually adjusted trim.
klugesmith, Thu May 10 2018, 04:52AM
An EL-USB-5 event data logger soon showed that the switch had issues.
First chart is for the unit roughly as pictured, on a desktop at work. Fair amount of vibration in the environment, eccentric weight too small, maybe not enough damping. Could a tipping event have so much overshoot that the rebound tips it back?
Second chart shows the switch in a stable, low vibration place, with a quarter dollar in place of the penny. Good times are good, but bad times have some bit errors!
Before the third chart I removed the hour hand. I guess its tipping moment was enough to perturb the 1-hour-cycle's casually adjusted trim.
Re: Dirt-simple time switch with one hour period
klugesmith, Fri May 18 2018, 04:57PM
The switch has been put to use, after a minor improvement.
It was made for investigating how much juice is left in discarded battery cells.
We log voltage under load, and see how it bounces back when the load is periodically removed.
So far the subjects have been old alkaline AA cells. Load is a PR-something flashlight lamp drawing about 0.4 A at 1.5 V. Indicated voltages need to be divided by 12.5, because I bypassed the input attenuator in a EL-USB-3 which is nominally 0 to 30 volts. At the USB connector shell it's 0 to 2.4 volts.
The first voltage logs look like this:
I figured the paper clip contacts were a bit scratchy. Bypassed them with some thin wire (30 AWG), which makes it harder to separate the lever and fulcrum parts. Now the logs look more like what we expect:
Here's my first run on a battery at a conventional end-of-life point. It's from a wall-mounted clock whose second hand was missing a few steps every time it passed 9.
klugesmith, Fri May 18 2018, 04:57PM
The switch has been put to use, after a minor improvement.
It was made for investigating how much juice is left in discarded battery cells.
We log voltage under load, and see how it bounces back when the load is periodically removed.
So far the subjects have been old alkaline AA cells. Load is a PR-something flashlight lamp drawing about 0.4 A at 1.5 V. Indicated voltages need to be divided by 12.5, because I bypassed the input attenuator in a EL-USB-3 which is nominally 0 to 30 volts. At the USB connector shell it's 0 to 2.4 volts.
The first voltage logs look like this:
I figured the paper clip contacts were a bit scratchy. Bypassed them with some thin wire (30 AWG), which makes it harder to separate the lever and fulcrum parts. Now the logs look more like what we expect:
Here's my first run on a battery at a conventional end-of-life point. It's from a wall-mounted clock whose second hand was missing a few steps every time it passed 9.
Re: Dirt-simple time switch with one hour period
klugesmith, Thu May 24 2018, 11:54PM
We see that batteries need to be almost bled dry, before the open circuit voltage drops significantly.
This chart shows the end of one AA cell, then substitution of a different used cell. Wednesday morning the switch got disturbed and stuck in Open state.
Before this goes on much longer, I should figure out how to make the Y axis show actual voltage, or 10x instead of 12.5x actual.
If it's done with a 5:4 voltage divider, which is always loading the cell under test, how about making it 1000 ohms? As opposed to 10K, or 100 ohms?
klugesmith, Thu May 24 2018, 11:54PM
We see that batteries need to be almost bled dry, before the open circuit voltage drops significantly.
This chart shows the end of one AA cell, then substitution of a different used cell. Wednesday morning the switch got disturbed and stuck in Open state.
Before this goes on much longer, I should figure out how to make the Y axis show actual voltage, or 10x instead of 12.5x actual.
If it's done with a 5:4 voltage divider, which is always loading the cell under test, how about making it 1000 ohms? As opposed to 10K, or 100 ohms?
Re: Dirt-simple time switch with one hour period
Patrick, Mon Jun 04 2018, 04:20AM
is the need for a de-bounce switch part of this issue.
Patrick, Mon Jun 04 2018, 04:20AM
is the need for a de-bounce switch part of this issue.
Print this page