The solar sensor used here is a small circuit board from Red Rock Energy (redrok.com) that is housed in a weatherproof enclosure from Leviton (leviton.com) fastened to an extension of one of the hangers on the most outboard collector, the one that is in the sun earliest in the morning.
Things had been going along quite well since this was installed in this form about two weeks ago. Tracking seemed to be good most of the time from early morning until late in the afternoon.
Of course I was not out there every minute of every day watching how it was doing, but would go out every hour or so on a sunny day to observe how well the collectors were positioned. If they are accurately focussed, the concentrated light beam which is easily visible, will be right down the center of each collector tube. The collectors are all locked together with the metal rod which links them and they are adjusted as part of the initial setup relative to this one which holds the sensor.
What I was noticing was that on some visits the collectors were not well focussed and were off slightly to one side or the other. I would at first try to make a small adjustment to the sensor, but then later that day, or the next day, they would be off again and it did not seem to be a constant amount, or it would be off in the other direction. Once or twice when I tried to make an adjustment, the sensor did not seem to be well behaved at all, and would race off in the opposite of the direction I had wanted. This was bizarre.
Here is what the sensor housing looked like before today. You will notice that there is a message from the manufacturer telling you what to do to open the housing. The letters are raised embossed letters cast into the clear cover when it is molded. Of course it would have been better if there was no lettering on the cover at all and that it would have a smooth clear surface, but that was not the case and I had decided to use it like this.
What appears to have been happening is that the lettering was catching the sun and forming concentrations of light that were fooling the sensor.
You can see how close the lettering is to the two diodes that do the actual tracking. When I took this picture, the sun was above and behind so its rays would pass through or very nearly through the letters. I couldn't take a picture from inside the housing but I could look up through it and the letters had many little facets of light, like tiny diamonds that shone from the sunlight they caught. It wasn't difficult to imagine the sensor being distracted.
Here is the Leviton product I am using as a sensor housing. It is what they call a "Rain Tight While-In-Use Weather Resistant Cover" model 5997-CL. These are intended for use outside to provide a pretty much weather tight shield around a power cord that might be plugged into an outdoor receptacle, for example.
There is a lot to like about this cover. It is inexpensive (about $18). The plastic cover is very clear optically. It is pretty nearly water tight when closed although it could be made a little tighter with the addition of some additional gasket around the cover latch.
But there is the light catching faceted lettering right in the part of the cover that I want the sensor to look through.
I decided to see if I couldn't salvage the cover for use in this solar collector application by first sanding out the lettering and then through a series of progressively finer sandpapers and final polishing try to remove the raised lettering. The cover seemed to be acrylic material so this should be possible. I decided to give the process a try.
Here is what it looked like after careful sanding with 220 grit sandpaper for about five minutes. At this point I thought that this was not going to work and that I had possibly ruined a sensor housing.
I continued with the 220 grit sanding in a circular motion changing direction periodically and then side to side and up and down, trying to vary my pattern often, until the lettering was completely gone. The cover top surface was now completely cloudy like a frosted bulb.
I cleaned it carefully with a clean cloth to remove any abrasive and stepped to 400 grit paper, again varying my sanding pattern. Clean again and step to 600 grit and then to 1200 grit. It still looked very frosted, but very evenly so.
I then used a grinder on which I had mounted a felt buffing disk and loaded the disk with a light coating of plastic polish. After the first pass, I knew that this was going to work and after a few more cleanings and progressively lighter touches to the wheel, again varying the pattern, the cover looked great. It looked just as if it was made without the lettering!
I took the opportunity of having it disassembled to replace the connector which mates with the Red Rock module. Red Rock provides the matching connector but mine had been out in the weather for the last four years and I decided that I'd like to replace it to eliminate a possible source of trouble.
I remounted the sensor housing and the sensor this morning and all seems well so far. I have a pretty good feeling that this was the cause of the wandering behavior, but time will tell.
Here is a final view of the de-lettered sensor housing in place.
This morning also I installed a water flow meter so that I would know the flow rate of the water through the collectors in order to do a thermal calculation which I will present shortly.
Thanks again for your interest in my project.