Using Gravity to Follow the Sun

How do smart phones and tablets orient screens? Inexpensive sensors inside readily define “up” and can also measure tilt, register motion and detect taps. I’ve been combining an Arduino with digital accelerometers (http://www.adafruit.com/products/2019) to see if together they can point a solar collector when the sun doesn’t shine.  This should enable them to point at the horizon just before sunrise, track the sun behind clouds, and know when they’re facing down so they can sleep overnight or during storms, protecting solar panels and mirrors from hail, dirt, frost, freezing rain and snow.

My approach, outlined in the figure below, tries to create a gravity direction scale that matches a time scale so that the controller can reset its internal clock when the direction of the collector and earth coincide. The earth establishes time for solar collectors that rotate on axes parallel to the earth’s polar axis. A tracking controller simply has to synchronize its internal time base with the earth that consistently rotates one revolution per day. Only longitude matters and conventional “clock” time with time zones does not. Concentrating solar collectors that use mirrors to direct sunlight into a receiver work best when the energy, often intensified 1,000 times, is centered inside the receiver. When sensor pairs, with an equal number of thermocouples on opposite sides of the receiver aperture, are balanced (when the temperature on both sides are equal), the tracking structure and concentrator optics are exactly aligned with rays from the sun.

The elliptical orbit of the earth around the sun makes solar noon, when the sun is highest in the sky, vary slowly throughout the year but only by up to a few seconds per day. Complex celestial mechanics equations can correct for this perturbation but require significant computing power and expensive sensors with minutes of arc resolution. “Open loop” approaches that use encoders and time to point at the sun also require expert installation service with exacting setup routines. Even a power failure or perturbation of a sensor can require a service call.

One goal for this work is to enable a simple controller to program itself, using sunlight sensors to direct the collector at the sun so that intensified sunlight hits sensors mounted on the edges of the receiver. One set on a left-right axis controls east-west motion and another set controls up down motion. “Closed-loop” tracking that use sun sensors have had maintenance issues when one side gets dirtier than the other (e.g. from bird poop).  This approach uses thermocouples that face the mirrors (downward) when operating and routinely get cooked above 1,000 degrees as the intensified sunlight enters and leaves the receiver aperture (that should keep them clean – it works for self-cleaning ovens!). In any event, since they work on temperature differences, differential soiling should not affect them very much.

In coming weeks I’ll be playing with both digital and analog accelerometers to see which are easier to use and what programming techniques are appropriate. I’d appreciate feedback if anyone has suggestions on how to proceed. I have spreadsheets of data from sensors as they rotate and am trying to develop code that accommodates latitude so that anyone can simply erect a solar collector, tilting it at the latitude angle, along a line of longitude and have the controller take it from there. Bump the collector with a tractor? It'll reprogram  itself!

Programming Flow Diagram for Tracking the Sun Using an Accelerometer