This repo contains source code and description for a voltmeter clock build.
Substantial credit is given to lcamtuf for his excellent blog post write up that inspired this build. The source code is largely based on lcamtuf's original code, with some modifications as described here.
- Parts summary:
- MCU: AVR128DA28
- Crystal: 8 MHz crystal (ECS-80-18-4X-CKM)
- Voltage Regulator: LM2596
- Digital volt meters: 6C2 0-5V Analog Volt Meters
- 9V DC wall wart power supply
- Wood frame: cherry wood with Danish Oil finish
- Basic pushbuttons for hour, minute setting
- Hand soldered protoboard for circuit
This MCU used in this clock is a AVR128DA28 as described in the blog post. The MCU can be programmed with a dedicated programmer, but I used an Arduino UNO to do so. To do this, I used the Arduino IDE with DXCore, and flashing jtag2udpi on the UNO to allow it to act as a programmer. The code can then be flashed via the Arduino IDE.
The circuit diagram for the clock is reproduced below from the original blog post:
The templates in meter_clock_templates.pdf were used to overlay on the meters' original face plates. Templates were simply printed on paper, carefully cut out by hand, and glued onto meter faces with a glue stick. I reccomend doubling the paper layers to avoid seeing the original faceplates, and placing a weight on the templates after the gluing to avoid creases.
I used a random 9V wall wart power supply to power the clock and used an adjustable voltage regulator to step it down to 5V. I used an adjustable regulator as I found supplying exactly 5V to the MCU wasn't sufficient to for volt meters to hit the max of their 5V range when driven by the MCU. Through trial and error, I adjusted the voltage input to the MCU until a volt meter hit it's max 5v position, and this took approximately 5.25V input.
The frame was cut from cherry wood. The front holes for the voltmeters were carefully cut with a jigsaw to necessary dimensions. Voltmeters were crudely hotglued in place. A box joint jig was used to make finger joints.
Compared to lcamtuf's original code the following mdofications were made. AI tools were partially used to aid in these modifications.
- The primary PWM loop driving the voltmeters in the original implementation ran at roughly single digit KHz rates. I found this created an audible high frequency hum from the voltmeters. The PWM driving loop speed was increased to 100 KHz which eliminated the human audible noise.
- By running the PWM loop faster, however, the PWM duty cycle step resolution decreased to 80 steps. This would cause jerky movements on the second hand if used as-is. To smooth the voltmeter motion back out, a sigma-delta dithering approach was used to artificially emulate a higher step resolution. With how fast the duty cycle loop runs, the resultant second-hand motion is smooth.
- Functionality was added so that the second-hand falls back more gradually on minute wrap-arounds (to avoid an audible noise as the hand hits the end stop). This function is skipped on hour wrap-arounds, as this then kind of acts like a chime to make one aware of the top of an hour
Box dimension diagram
