# How to Run a Battery Electric Clock on Solar Power--Part II

1,567

25

7

Published

## Introduction: How to Run a Battery Electric Clock on Solar Power--Part II

We achieved much with a single 2 Volt solar cell charging a super capacitor in Part I but to go further we need to add more solar cells which in turn brings in the need for extra control in the charging circuit.

Here in Cornwall in the very south-west of England, days at the time of the winter solstice shorten to around eight hours. The sun is low in the sky and Atlantic depressions queue up to pass over us. Under such conditions our single 2 Volt solar cell will struggle to produce even one Volt at very low current. We need more cells in series to scavenge enough power to charge our super capacitor but an array of several cells in series can, in better light, produce enough voltage to destroy our super capacitor which cannot withstand more than 2.7 Volt. Some form of charging control is vital.

## Step 1: A Possible Control Circuit

First some general thoughts on the control circuit.

The requirements of the control circuit are simple in that we need to charge our super capacitor to around 2 Volts and something in the range of 1.9 to 2.2 Volts would be fine. More than this may be getting too high for the clock mechanism and lower than this means not building up enough charge to run the clock through the hours of darkness. Ideally the control circuit itself should not itself consume unnecessary power as we are trying to scavenge every scrap of energy under low light conditions. Under no circumstances can the voltage be allowed to rise above 2.7 Volts which would destroy the super capacitor.

A conventional solution would be the simple circuit shown in the diagram above which uses an NPN transistor as an emitter follower. The voltage appearing on the emitter is 0.7 Volts lower than that on the base and since the resistor/Zener diode combination places 2.7 Volts on the base we get our 2 Volts output on the emitter to charge our super capacitor. R1 sets the standing current through the Zener diode and R2 represents some sort of load. For those willing to experiment a 2.7Volt Zener diode would be fine but I opted for a slightly cheaper system that allows a little flexibility and this is shown in the next step.

## Step 2: The Actual Control Circuit

The circuit in the first picture above shows a very cheap and simple modification of the Zener diode circuit outlined in the previous step. The Zener diode has been replaced by a chain of silicon diodes which can be small signal or rectifier types--a diode 'bargain bag' is a good source. After passing through a threshold the forward voltage drop across a silicon diode changes only slowly as the current through the diode increases further and we can utilise this to make a crude control circuit. Our chain of diodes is fed from the solar cells via a 10k resistor and the voltage at the junction of the resistor and the top diode is around 2.7 Volts. This voltage is fed to the base of the NPN transistor which is essentially an emitter follower and for such a circuit the voltage at the emitter will be 0.7 Volt less i.e. around 2 Volt which we can feed to our super capacitor in safety.

We can add or subtract diodes to raise or lower the output voltage in roughly 0.5 Volt increments, and raising the 10k resistor to 22k or even higher will lower the output voltage a little for fine tuning and also reduce the quiescent current even further.

The 47 Ohm resistor in the collector circuit is non critical and simply limits the potentially damaging current when you have an empty super capacitor and a solar array in bright sunlight.

The circuit output of both four and five diode circuits for feed voltages up to 12 Volts is shown in the second and third picture above. This shows adequate control and a five diode circuit may be appropriate for a six Volt solar array and a four diode circuit for a twelve Volt solar array.

The fourth picture shows a five diode circuit on a breadboard charging a conventional 1000uF electrolytic capacitor from a twelve Volt source. It is a good idea indeed almost mandatory to try your circuit out on a breadboard with a normal electrolytic capacitor before risking your super capacitor.

(As an aside for those wanting to thieve joules note that this charging circuit can be used for safely charging super capacitors from arrays of flat cells or even arrays of experimental ones. For this situation an important advantage of this charging circuit is that once the super capacitor reaches maximum charge the current stops and in darkness the charge cannot leak back as it is blocked by the base/emitter diode of the transistor and the only power drain is down through R1 and the silicon diode chain. Further enhancement would be to make R1 higher or even replace the transistor with a Darlington pair in which case R1 could be very much higher and the residual current of the circuit would be micro amps.)

## Step 3: Example Clock 1

This clock shown in the first picture is something of a family heirloom and was presented to my parents on their 25th wedding anniversary in 1964. The original Metamec movement has been replaced with a modern one. The second picture shows the rear of the clock before modification and shows that, in this case, the rear is flush and this allowed a piece of quality slate to be glued on with MS polymer glue. First the slate had been drilled to allow wires from the solar cells to be passed through before the six solar cells were glued on to the slate. The solar cells were obtained from the same source as in Part 1 of this article. The use of six cells may seem a little excessive but we have to cope with very low light levels in the middle of winter--it is possible that three cells would have sufficed. The first picture shows the front view of the assembly which all concerned find aesthetically pleasing.

The circuit diagram is shown in the third picture above. In this example the 6 solar cells are wired as a chain of three lots of two cells in parallel and in this case bypass diodes were not included. The positive end of the array is connected to a control circuit explained in the previous step and here we have used a chain of five diodes resulting in the capacitor being charged to around 2.1 Volts maximum. In this example the super capacitor consists of two 10 Farad units wired in parallel as these fit better in the space available.

## Step 4: Example Clock I (cont)

Here we see the actual construction at the rear of the clock and it owes so much to MS polymer glue. Tag boards provide the anchorage for components and wiring junctions and these are glued down with MS polymer glue. On the tag board glued to the clock itself you can clearly see the chain of five diodes, the NPN transistor and the two super capacitors. Electrical connection to the battery terminals was done by wedging the wires in with the length of wooden dowel--the method used in "How to run a Battery Electric Clock on Solar Power--Part I" would have been better. The gold wording on the 'quality' slate reveals that this was sold as a teapot stand!

The second picture shows the clock deployed in a conservatory attached to our daughter's house where it fulfils a need and gets plenty of daylight. The clock is a good timekeeper and access is needed just twice a year at the times of 'spring forward, fall back'

## Step 5: Example Clock II

This is the ultimate stage in our logical progression through the problem. Shown in the second picture is a battery clock which receives a long wave radio time signal from the 60 kHz MSF transmitter at Anthorn in the UK. These signals are transmitted in many populated parts of the world, for instance WWVB in Colorado USA on 60kHz and DCF77 in Frankfurt am Main in Germany on 77.5kHz. You have to be in range of such a transmitter for this stage to work.

The first picture shows the circuit diagram. Note that here we have six solar cells again but they are connected in series and bypass diodes are connected across each solar cell--in retrospect these diodes are probably not necessary. I cannot say whether it is better to have 6 cells in series or two lots of three in parallel but there may be some advantage to the six in series when the light source is extremely low. The regulator circuit copes well even when it receives the full 12 Volts in bright sunlight.

Note that there only four diodes in the base circuit of the regulator transistor Q1 and this is because it is necessary to reduce the voltage of the clock power supply--this sort of clock has considerable electronic content and too much voltage can make the clock operation erratic in contrast to the robustness of an electromechanical movement.

The super capacitor here is a whopping 50 Farad and this will help the clock cope with long dull periods since we are using a reduced voltage and the capacitor will not charge up so far.

The third picture shows construction at the rear of the clock which, in this case, presented some difficulty in that there is little space for the electronics. The circuitry is mounted on Veroboard glued to the rear of the solar cell array which is in turn attached to the clock with screws after drilling--it all depends on the clock and flexibility is needed.

The fourth picture shows the sub assembly in close-up. Five diodes are in view but one of them is shorted out.

The last picture shows the clock deployed in a small porch/conservatory that we have at the rear of our house. The clock is visible from around our garden and, since it is accurate to second, easily accessible as our house master clock. It works around the year coping automatically with 'Spring forward and Fall back' and can stay where it is unattended until it is time to redecorate or move house!

## Step 6: In Conclusion

Our objective has been largely met although it must be admitted that it is not yet possible to operate such a clock all year inside the house away from windows and obvious light penetration.

Cost is obviously a factor but the joy of experimentation may overrule this and as a present for a favoured relative it would be priceless.

There may be applications in commercial premises such as a garage forecourt where the radio version could be mounted in a high position and left there. There is also the possibility that it would work in the shopping centre/mall situation where the long hours of artificial light may keep the clock running but I am not in a position to try this. Wherever deployed it could be a talking point.

This subject has probably been done to death now but there are still possibilities for more innovation and there may be material, if not for Part III, then at least Part IIa.

## Recommendations

• ### 3D CAM and CNC Class

686 Enrolled

## 7 Discussions

No! This project has not been done to death and I have been looking to do this for some time. I have now bookmarked this as my starting point for my project.

I ideally want to build a solar powered radio controlled clock to work indoors in a south florida house. So here I have the guts of the solar/capacitor/control unit, now I just need the radio controlled clock movement for that location and the clock frame/base.

Thank you! Great work!

1 reply

Thanks for your interest. I don't think that there will be problems in the latitude of South Florida--plenty of sun!

The radio controlled clock here is still where it was placed in March 2016 and has run faultlessly ever since. Importantly, it has coped with the 'spring forward/fall back' adjustment where the clock has to have the power to wind back eleven hours in October. The clock can stay there now and it should carry on running so long as there is a long wave radio signal and daylight!

Even now, there may be ways of deploying the solar cells in a more efficient way and thus cope with poor light levels or maybe take advantage of artificial light. If I come up with anything I will tag it on to this article.

Circuit looks fine to me. You could maybe simplify it a tad with a zener diode across the solar cell and a series diode to the supercap, allowing everything to be built from the random diode grab bag, but that's not a significant improvement. Solar cells are supposed to be a light dependent constant current source so the resistors are optional. Very much doubt you would fry a supercap with your circuit.

I'm wondering whether the supercap would have longer life than a NiMh battery however. Battery life is usually all about discharge cycles of full depth, and the cycle depth would be very small in this application (assuming you get sun at least once a week say). I had to replace the 'supercap' in my Seiko kinetic watch - it wore out. And NiMh batteries are readily available.

2 replies

Further to my previous reply I have now published 'Part IIa' which does take on board the use of a nickel/metal hydride cell and a little more. Thanks for your input--the reader is now spoilt for choice.

Thanks for your very constructive comments. So much here is dependent on the life of components in a rather strange application. A NiMh cell would be the best solution if it has enough life. The clock power consumption is so very low that once charged up the cell would stay that way just being trickle charged when the light is there and, if it can hold that charge, it could then take the clock through long periods of poor light. Taken to the limit it might be possible to get the clock through an Arctic/Antarctic winter on the strength of the summer midnight sun and you could probably manage with fewer solar cells.

I am rather hoping that the super capacitor is going to have a long life in this mechanically static application and that remains to be seen :-)

If your capacitor cannot absorb more than 2.7V your schematic is wrong, as it is fed with 3.3V because there is a voltage of 0.6V between base and emitter, moreover the intensity will vary due to the bad polarisation of the diodes where the current is too small. A one try can work but you cannot certify it will always work ! Your fisrt try was better with a Zener, you can use a zener between 1 and 2V.

1 reply

Well hand on heart I can say that I have not lost a super capacitor yet.

The key fact here is that in the circuit shown the current through the diodes is very low and indeed well below a milliamp so the forward voltage drop is very different to that observed when the diode is on power supply duty. This is important because at very low light levels we are trying to scavenge current in the milliamp region and we do not want it dissipated in a control circuit.

I have just done a few measurements with an IN4006 rectifier diode and a 6 volt supply with different series resistors.

10K gave 0.66mA of current and a voltage drop of 0.56 Volt

22K gave 0.30 mA of current and a voltage drop o. 0.52 Volt

47K gave 0.13 mA of current and a voltage drop of 0.48 Volt

5 diodes in series will have a combined voltage drop of 2.8, 2.6 and 2.4 respectively and then you can subtract 0.7 volt from those figures.

A 2.7 Volt Zener diode would be fine but would normally be used with a higher standing current and it would have to be determined how the circuit would perform with the sort of starvation current being used in the design.

Again I would urge prospective builders to breadboard the circuit with a normal electrolytic capacitor first.

As an aside there are other possibilities. I have just measured a red LED to have a forward drop of 1.77 and 1.82 Volts with currents of 0.1 and 0.5 mA respectively and, most relevant, a white LED has a forward drop of 2.54 Volts at 0.09 mA current.