Rolling Ball Clocks #2
The following describes some of the troubleshooting process that was required to make these clocks perform 'like clocks'. I expect them to run continuously, reliably, and accurately just like 'real' clocks.
All done -- well, mostly --and ready for testing.
Everything works but there are a few places where the balls stick and a couple places where they bounce out of the tracks.
This drive unit powers things just fine but there are a couple places where the drive gear cogs bind on the first wheel gear teeth.
The balls have a tendency to hang up on the corners of the supply track (bottom-left in picture ). They also 'crowd' the hole on the first wheel which can cause the drive to balk.
I tried to sand down the bottom of the track, at the corner joints, to stop the binding. I didn't like the way this looked.
Finally I decided the whole bottom track needed to be replaced. It needed to be longer with modified angles to change where the balls were presented to the hole in the wheel. When making the new track piece I took care to keep the joints smooth to prevent binding.
Replaced 8-cog drive wheel with larger 9-cog drive.
Notes:
After long hours of pondering -- and long struggles with carefully drilling holes in the disks for the dowls (cogs) -- I realized that the dividing head could be swiveled on its base. This allowed me to mount it on my drill stand. (Whew)
The dividing head has a ratio of 40:1 between the index wheel and the chuck. 40 turns of the index wheel will rotate the chuck 360 degrees.
I need 9 teeth in the gear so I will turn the indexing wheel 40/9 turns for each tooth.
(here is a calculator on the web )
Gear for clock 2
9 teeth - 360 degrees/9 = 40 degrees / tooth
40/9 turns on the index wheel
Based on the calculator:
Plate 18
4 turns plus 8 holes
From right to left:
The larger drive wheel makes this fit a little more snug -- but still plenty of room.
I do not attach this drive unit to the base. This allows it to float up and 'skip' if there is a bind somewhere.
One track had to be completely replaced.
Transfer control components from breadboard to permanent chassis.
For the first clock I built the controller completely from scratch. Power supply, stepper motor driver, and controller. For these clocks I cheated a bit. I didn't think I needed the powerful motors I had on hand so I found some on Amazon. These come in a 5-pack and include the driver module on a small board. All five for less than $20.00! (now I have spares) How could I go wrong?
I also found a power 'brick' that supplied 3000 mV at 5.0V. AND it included a dongle that it plugged into with a screw-down terminal strip for the + and - supply leads. This is plenty of power to drive both the Raspberry Pi and the stepper motor.
Here is the controller assembly mounted on a wooden chassis.
As you can see, I am using a Raspberry Pi Version 4 to drive the system. This may be hugely overkill for this purpose but for the small price it provides a huge amount of expandability.
By using a micro-computer running Linux to drive the clock, I can program in a self-calibration to automatically keep the clock on accurate time. The software checks the clock every minute and makes any needed adjustements to the speed of the motor.
In the above image the Raspberry Pi 4 is on top. The dongle for the power supply brick is bottom left. The stepper motor driver board is on bottom right.
Here is an example of the cool things you can add because of the Raspberry Pi. The Pi runs an Apache web server which can host a web page that can be used to turn on and turn off the clock from any device that has a web browser (as long as it's on the same local WiFi network with the clock.)
Run the clocks and watch for the following:
