Project: 24 Volt Slave Clock Driver

Thread Starter

Sensacell

Joined Jun 19, 2012
3,432
I have this beautiful old school clock, built in 1952 and still works great.
it's designed to be a slave, driven synchronously from a master clock in the principals office.
It requires a 24 Volt DC pulse once per-minute to operate, I built an AC Line powered driver for it years ago, but needing the line cord sucks.

My goal is to create a battery powered driver for this clock, so it can hang on the wall without wires.
The mechanism has a solenoid coil of about 680 ohms resistance, which would draw about 35 ma @ 24 volts.
The power required is (24 * 0.035) = 0.84 watts, times duty cycle, (0.2 / 60) * 0.84 = .0028 watts (assuming a 200 ms pulse, every minute)
An alkaline "D" cell is supposed to contain about 20 watt-hours of energy, so it might last 1.5 years- (theoretically?)

I am considering the following architecture:

3 VDC power source- 2 "D" cells- I want it to run for at least a year between battery changes.
Micro-controller running 32.768 Khz crystal for time keeping- creating the once per minute pulse.
Simple boost converter to charge a capacitor to around 30 V.

What I am considering is having the boost converter shut down most of the time, then have the converter kick in after every pulse to re-charge the capacitor. This would conserve power, no wasted quiescent current from the boost converter once the capacitor is fully charged.

For simplicity, I was considering having the micro-controller generate the pulses for the boost converter, operating in "fixed-on-time" mode.
A comparator would monitor the capacitor voltage, the micro would stop making pulses once the capacitor is charged.

Anyone have any ideas on how to do this simpler, better, cheaper?

IMG_5177.JPG IMG_5178.JPG
 

Bernard

Joined Aug 7, 2008
5,784
Maybe experiment with capacitive discharge to see what is the smallest C that will reliably advance the clock
with V of 24 to 36 ? School dist. disposed of around 3000 impulse clocks in the 1970's, wish I had saved one.
 

Marley

Joined Apr 4, 2016
502
I think you have the right idea. I like the idea of pumping up a capacitor with a boost converter and discharging through the solenoid. you will need to experiment.

Reminds me that I have an electric fencer unit here that runs from 2 D cells and manages to produce a kick every 10 seconds or so and lasts quite a few months on a good set of batteries - so possible!
 

crutschow

Joined Mar 14, 2008
34,285
What I am considering is having the boost converter shut down most of the time, then have the converter kick in after every pulse to re-charge the capacitor.
You should have the converter kick in just before the pulse so leakage current from the capacitor would have not have any significant time to drain the charge.

You might experiment with capacitor size and charge voltage.
A somewhat higher voltage and a smaller capacitor could actually require less energy.
 

Bernard

Joined Aug 7, 2008
5,784
Using a 4PDT, 24 V, 613 ohm coil relay as clock, monitored with R & LED on one set of 4 contacts I tried 10 uF @ 33 V, good flash, 5 uF still good, 4.7 uF, good, 2.2 uF , small click -no flash.
Retaining 4.7, ( 4.88 measured ) dropped V in steps. 33 V- good, first small change at 25 V. Charger to 25 V & held for 1 m. good flash. Weak at 22 V no delay, no go at 21 V.
May next try with 2N5060 SCR.
 

marcf

Joined Dec 29, 2014
288
I enjoyed your post a lot.

I used to work on AR-3 impulse clocks.
https://ihbusboy.wordpress.com/clock/impulse/

If you wanted to get a little more involved, this clock has the ability to correct itself every hour. It also has the ability to do a 6 o'clock correction if things really get bad.

The longest running clock that I saw was installed in 1927 and still clicking. (Lots of 1 min pulses)
 

crutschow

Joined Mar 14, 2008
34,285
You might experiment with capacitor size and charge voltage.
A somewhat higher voltage and a smaller capacitor could actually require less energy.
I did some simulations varying the voltage and capacitance value, and that would not appear to be true.
For a decay time to the same estimated solenoid dropout voltage of 9V, the reduced capacitance with a higher voltage required more energy because of the V² factor in the capacitor stored energy.

Of course I didn't factor in the pull-in time of the solenoid and that might be better with a somewhat higher capacitor voltage, allowing for a faster voltage rolloff and even smaller capacitor.
I guess some experimenting with the actual solenoid would be in order. :)
 

Thread Starter

Sensacell

Joined Jun 19, 2012
3,432
Thank you all for the responses!

I will summarize my take-away from all your input.

Charging a cap to a voltage higher than I need might be wasteful of energy- (the V2 factor) I will experiment with some caps and the clock movement to determine the optimal values. Seems like the voltage decay on a smaller capacitor works to my energy-saving advantage here, as the armature / magnetic circuit closes, the current required to pull-in the armature decreases. This seems better by far than just topping-up a very large capacitor, for example.

For the boost converter to be efficient, I will find an inductor with low ohmic resistance and a high saturation current.
Since I have a long time to charge the cap, It's interesting to consider how to optimize for efficiency, not rate of energy transfer.
I will utilize an off-time long enough to ensure the inductor is completely discharged every cycle, even at min cap voltage.

Choosing the inductor and switching parameters so the peak inductor current is relatively low, so the losses in switching elements, inductor and internal battery resistance will be low.

An additional optimization would be to re-start the charge phase late in the cycle, to minimize energy loss from leakage.

I did find this - I cannot sort out from the description if it would drive my movement correctly.
 
Last edited:

Bernard

Joined Aug 7, 2008
5,784
This circuit might be similar to your need. The transformer is used because of the low V rating of the FET. The 4.7 mH choke with added 36 turns, center tapped was used before to produce 160 V DC. I'll try to
re wind with about 40 turns of .008 in. wire. and may try circuit but time is short for next 10 days.
Will have to use LM393, best that I have in stock.Impulse Clock 00000.jpg
 

Thread Starter

Sensacell

Joined Jun 19, 2012
3,432
Here is what I came up with.

The PIC16F1455 chip is tasked with doing everything, even creating drive pulses for the boost converter.
When the boost converter is running, the CPU runs at 4 Mhz from the internal oscillator.
The 3mH coil is a little bulky, but the relatively high inductance makes it simple to bit-bang the boost converter in code.
I use the PIC's internal comparator to check when the capacitor reaches ~24 volts, the voltage divider driving the comparator input is switched - so it doesn't suck charge from the capacitor when it's coasting.

The time keeping is via TIMER1, which runs from an external 32.768 Khz crystal, generating an interrupt every 2 seconds when the timer rolls over.
To make the battery last, I turn off the comparator and voltage reference peripherals when the boost converter is idle.
I am trying now to figure out how to make the MCU sleep between TIMER1 interrupts, that's the final energy saving trick.
 

Attachments

Top