So I got this old school-physics meter on a fleamarket, with the intent to make a clock of it. Yes, a weird sort of display with the pointer going slowly from 0 to 1 during the day and snapping back at midnight. Or something else along those lines.
The meter scale is detachable with little black box in the connectors path so it can be converted to a Voltmeter, AC/DC and so on. (Unfortunatly these were not included). I do not know the sensitivity of the meter when bypassing this "scaling", yet, but think it will be in the many tens of mA.
Now, the project questions - just bouncing ideas around, sanity check.
Arduino, a RealTimeClock chip, power supply (Wallplug or batteries) This will be mounted underneath or otherwise hidden.
My first thought was to remove the actual meter and just put a small stepper there. No power needed for the stepper to hold position (the needle weighs close to nothing).
This might spoil the neat look, so I am speculating on driving the meter with the PWM (The inertia of the meter would totally mask the PWM frequency - besides there is some circuitry in the box - presumably to potect the meter against students wiring mistakes - that would smooth the pulses). I suspect the power requirement would drain any battery within the day
You need to find out the basic sensitivity of the meter - I've driven an old VU meter from PWM with just a load resistor.
I think the meter was a 100uA FSD.
Of course, for a clock, you're stuck with just 256 increments per day.
Bigger battery 8)
Solar cell to charge the battery during the day?
Wouldn't stepper hold position when unpowered? Have the atmega go into power down sleep mode, wake once a second with RTC pulse, check the time, enable power to stepper if time to move (relay?) then turn off and go back to sleep.
Forget about electronics for a moment and consider the simple physics of visual mechanics.
The meter movement is probably around 120 degrees. The movement on a normal (single revolution) 24 hour clock is 24x360 =8640 degrees, based on the hour hand making 1 revolution and the minute hand making 24 revolutions in one day.
So at best your meter resolution is going to be about 8640 / 120 = 72 minutes per degree.
I would guess that at best you will be able to visually resolve about 2 degrees, which is about 2 hours.
An interesting clock concept perhaps, but not really a visually practical device.
Now consider the electronics.
Normal PWM output is 8 bit or 256 which at around 2 bits per degree will be more than adequate since you will only be able to resolve around 2 degrees of movement (4 bits)
To drive the meter you will need to convert the PWM to a smooth analogue value, the simplest of which is an RC filter.
Timer 1 (16 bit) will provide very fine resolution (about two seconds per increment). You will have to alter the configuration; the Arduino core uses it as an 8 bit timer.
Thank you all for the points made. I am going for leaving the original coilmeter untouched.
A first quick experiment has a PWM output via a 3.3Kohm going stright to the meter. The test program shows a full minute for each sweep. A 60th of a sweep was quite visible, about 3mm. Interesting that the needle had a small but visible overshot on each jump. With the 3k3 resistor it required PWM value 228 for full scale. That translates to ... 1.3 mA. Small compared to the Arduinos power. Hmmm. Batteries might be possible. Project question: The final version will not use the full board, just an appropiate AtMel processor with minimal components - is there a "low power" version?
There is a reason for the meter being glass encased apart from keeping prying student fingers away - the needle will waver at the slightest draft. I am doing a quick fix now so it will run unattended (on a wallplug) for an hour sweep.
Update: Yep, there it sits on top of the stereo. Arduino and wall plug supply quite visible. Only counting minutes so far, and without a RTC. Project on hold for a couple of weeks as I am busy elsewhere next weekend.
Processors with a "P" on the end of the part number (e.g. ATMEGA328P-PU) are "pico-power" and can be operated from 1.8V to 5.5V but should be run at lower clock speeds when the voltage is reduced.
Processors with a "V" on the end (e.g. ATTINY84V-10PU) are low-power and can be operated over the same voltage range but must be run at lower clock speeds (e.g. 10 MHz maximum).
AVR processors run very nicely directly from batteries. I've had a great deal of success with ATTINY85V and ATTINY84V processors using just 2 AA alkaline batteries.
