The Solar Bucket

The solar bucket (an alternative description being a plastic box full of wires) was designed to learn about solar energy before doing anything expensive.  The objective was not to make a "yes/no" descision about putting solar panels on the roof, rather to learn.  The process started around November 2007 and finished in June 2009.  The total cost was around £150.  The good thing about the Solar Bucket was the experience, however, because of ongoing development, the yield data was of variable quality.

Experience and Future

The lessons learned during the exercise are described on this page:

Lessons and Experience

At the time of writing (October 2010), it is planned to repeat the exercise.  The general plan is to use a solar panel to power a small computer, this will allow the panel's energy yield to be used and also provied some reliable data collection.

Data

The data collected during the project can be found at the end of the link, at the time of (the data is incomplete at the time of writing):

Monthly Summaries

Initially, the description of the weather conditions was obtained from the local radio station, however, Metar reports from Shoreham (EGKA) were used.

Evolution and Operation

From the start, my interest was not on how a solar panel would perform on a sunny day, rather how it would function during the winter and under overcast skies.  The first step was simply to connect an old computer case fan to a 1.5 watt amorphous silicon panel and see what happened.  During December 2007/January 2008, enough energy was generated to turn the fan for between zero and three hours per day.

A feature of wind and solar power is that energy is not necessarily generated when someone wants to consume it, i.e. the sun does not shine at night, thus some form of storage or alternative source is required.  This raised the question how much energy could be captured during the day.  Thus the Solar Bucket  became a solar panel connected to a lead acid battery.  During the day, the panel charged the battery, around 18:00 in the evening, the panel was disconnected, approximately an hour later after the voltage has more-or-less stabilised, a voltage measurement was taken, then the battery was discharged overnight using a variety of loads (e.g. computer case fan, 12V LED light and 100 ohm resistors).  In the morning a similar process took place, after the voltage had stabilised, a voltage measurement was taken.  A process not beyond improvement, but one which was instructive.

Design Objectives

Apart from simplicity and low cost, the design objectives for the Solar Bucket were:

Minimal Observations

A complex observing routine was unlikely to be maintained over a prolonged period, the regime involved changing switch settings followed an hour later by a simple meter readings, one in the morning and one at night.

Simplicity vs Complexity

It was tempting to add sensors and link these to a computer or data logger.  However, this would have delayed the project, Simplicity beat complexity.  With hindsight, as most of the problems which emerged were due to battery management rather than panel performance, a more sophisticated approach would have yielded better data and minimised battery degradation.

Separate the Charging and Discharging Phases

The original plan had been to separate the charging and discharging phases and treat each as an independent process.  In part this was successful, however, on clear sky days, it was sometimes necessary to disconnect the panel to prevent overcharging, thus under recording the amount of energy captured, similarly, it was not always possible to discharge the battery accurately.

Whilst the design evolved since is conception, more importantly the management and observing process evolved.

Design

The circuit diagram at one stage in the Solar Bucket's evolution is shown below.

Calibration

The battery was calibrated by charging it to its maximum capacity, then discharging it in stages, after each stage the voltage was allowed to stabilise.  An estimate of the current drained during each stage was used to create an idealised relationship between the stable voltage across the terminals and the state of charge.  This is shown in the graph below:

The general rule was to keep the stabilised voltage between 12.1 and 12.9V  in order to limit the degradation of the battery over time.

Evolution

The original intention has been to collect data during one trip around the sun and avoid radical changes and experimentation.  However, some evolution did take place.

Battery Management

Apart from the relationship between the sky conditions and yield, the big lesson was the need for effective battery management.  For a lead acid battery to have a long and fulfilled life, the load placed on it should be a small fraction of its capacity.  The rule used for the Solar Bucket was that the load should only rarely exceed 5% of rated charge capacity.  Thus no device was used to discharge the battery which drew more than 200 mA was used.  Also the stabilised voltage should remain within very narrow limits, in general I tried to maintain this between 12.1 and 12.9 volts.  However, a battery behaves something like a capacitor, when it is charging the voltage across the terminals is greater than 12.9 volts (typically 13.5 - 14.4), after disconnection, the voltage drifts downwards.  During discharge it is lower than 12.1 volts (typically around 11.7 volts) and after disconnection drifts upwards.

Ideally, when charging, when the voltage reaches something like 13.5 volts, it should be disconnected and remain disonnected until it has been partially discharged.  Similarly during discharge when the voltage has dropped below some level, say 11.7 volts it should be disconnected until it can be recharged.  Without some from of reliable management, overcharing and excessive discharging are inevitable togetether with the associated reduction of the battery's capacity.  During 2009, some experiments were made with charge controller and and a self-built low voltage disonnect.

Attempt to do something useful

The initial load used to discharge the battery was a 12 volt computer case fan which gave a peak load of around 150mA, this was replaced with a 100 ohm 10 watt resistor which drew a steady 120 ma.  Neither load did anything usefull.  During most of 2009, a 12 volt LED which drew around 80 mA was used, this provided a reliable and effective lighting for a dark stairwell

Size of components

The size of the battery should be such that the energy produced by the panel around the summer solstice should be approx. 30% of the battery capacity.  This would prevent loss of data due to overcharging and provide a margin of error when discharging the battery.  The original 1.5W panel was too small to provide useful data during the winter months and the replacement 4.7 watt unit too big during the summertime.

With hindsight, the Solar Bucket should have been put together as a system, rather than two components.  The measure of effectiveness should have been ability of that system to power a useful load.