Project Curacao2 – Update – Time for a Repair Trip!

In late October  of 2017, Project Curacao2 quit charging the batteries from the Solar Panel system and then slowly died.  Project Curacao2 had been down and operating solo down in the Caribbean  for over 5 months.

We could tell this was happening from the curves that we got from the machine as it trended down.    After thinking about what must have happened, we determined it most likely would be one of two problems.

1. A broken wire from the solar panels to the chargers.   This had happened almost a year into Project Curacao.
2. We noticed that this happened when the sun angles hitting the Solar Panels began generating a lot of power, getting the batteries to be fully charged.   With the number of Solar Panels we have on Project Curacao2, we suspect that the batteries became fully charged and SunControl stopped charging them.  This caused the panel voltages to increase to 7.0V (The Open Voltage value for the large solar panels we used – salvaged from Project Curacao).   This in turn (given the large amount of current) blew out our SunControl charger.  We had exceeded the maximum voltage on the board.   That’s the theory anyway.

In December of 2017, we had Dr. Geoff Howard open the Project Curacao2 main box and check out the voltages coming form the panels and they were present, so we had to rule out #1.

How to Fix Problem #2

Our working theory is #2, so we have now built a new SunControl based board with a protective circuit to prevent over voltages.   We used 1N4735A 6.2 V Zener Diodes.  A Zener diode is just like a normal diode (current only will flow one way), except that there is a much lower “breakdown” which in this particular diode is 6.2V.   The specification on the SunControl board is a maximum of 6.5V, so there is a comfortable margin.   The diodes are connected directly across the solar panels as in the picture below.   We used 20 of the one Watt diodes so we could discharge a lot of current and keep those solar panels below 6.5V.   It’s overkill, but we didn’t want to cut it close.   3600 miles is along ways away.

It works like this.  While the batteries are being charged, the voltage of the solar panels sits around 5 volts or so.  As the batteries get charged up, the voltage of the solar panels gradually increases to about 6V and when the batteries are fully charged,  the controller shuts off the Solar Panels which sends these panels towards their VOC (Voltage Open Circuit) voltage of 7V. However, when the Solar Panels get above 6.2 V. the Zener diodes start conducting lots of current which keeps the voltage from going above ~6.2V (see the picture below of our SunControl board with the protective diodes and a full battery – the voltage never goes above 6.14V).

We are also bringing an anemometer to replace one that was broken during a tower repair.

What is Project Curacao2?

Project Curacao2 is a redesign and rebuild of the Project Curacao environmental monitoring system that was running down on the island nation of Curacao in 2014 and 2015.   We took it apart when we got down to the island and debugged the failure.  It turns out what finally killed Project Curacao was the main LiPo battery on the Raspberry Pi had died.   The Arduino watchdog time was happily running away, but couldn’t wake up the Raspberry Pi.   We replaced the battery and the Raspberry Pi booted right up.   We brought the SD Card and a variety of parts back to the Labs for analysis.

The project Project Curacao2 consists of two LoRa transmitters based on a Mini Pro LP and a Raspberry Pi Zero powered base unit.   Lots and lots of sensors!   All solar powered.

• Part 1 - Introduction to Project Curacao2 and the LoRa WXLink Feeder Stations
• Part 2 - First LoRa WXLink Feeder Stations Report
• Part 3 - Testing the LoRa Range
• Part 4 – Initial LoRa Range Results
• Part 5 – Final LoRa Range Results
• Part 6 - Raspberry Pi Zero Base Unit Build
• Part 7 - Flame On!   Initial Base Unit Tests
• Part 8 - Solar Power Management System
• Part 9 - Ready for Deployment and Status 
• Part 10 - PC2 in the Caribbean!  Repairs and Testing
• Part 11 - It Lives! Deployment and Initial Results
• Part 12 - One Month of Results from Paradise

Below is a comparison shot of the original Project Curacao box (on the left) with the new Project Curacao2 box on the right.  You can see from this picture the advances we have made in three years.   The new Project Curacao2 box is wired together almost entirely using Grove Connectors.   We have also designed and integrated new solar power controllers, wiring schemes and, in a big design decision, removed the Arduino Battery WatchDog.   We found the reliability of our new solar power control scheme and power reduction techniques made the single computer Project Curacao2 possible.

 

Five Month Results

We have two issues.  Both results of not quite thinking through some of the environmental aspects of the  installation.  Both of the issues are angle of the sun to the solar panel issues, but for different reasons.

 Issue One

To the left are pictures (post deployment about two weeks – Thank you Jeff Maass) of the two LoRa transmitters.  Note the solar panels are down to flat against the box.  They should be angled up towards the sun, which they were when we left.   What happened?  The tropical sun.   We used bolts and tightened them at the proper angle.  Then we used super glue on the plastic and covered the whole joint with silicon caulking.   However, clearly, the thermal cycles on the plastic and the bolts (we did not use thread locker) has gradually let gravity takes it’s toll and they are now flat against the box.  This has resulted in less solar power to the transmitters although they are running most of the time with just occasional outages (we will generate a graph of the two transmitters power and transmissions frequency in due time).   They are facing south and as the year progresses, they will get more sun as the sun moves south.   This can be easily fixed by moving the panels to the correct angle and then putting a physical screw or wedge to make it physically permanent.  Right now there is a gentleman heading down to Curacao in November 2017 that will add a wedge under each set of panels to catch the sun.

Issue Two

The main Raspberry Pi base unit is mounted on the side of the Pj2T building it is basically facing south (which is correct).  We added solar panels on the sides to catch the morning and evening sun (our SunControl controller and the multi panel solar connector deals with multiple solar panels delivering multiple amounts of power very well).   However, we neglected to take into account just how far north the sun is in June.   So we aren’t quite generating enough power to keep the Raspberry Pi running 24/7.   As the sun moves south, this will gradually improve.   We noted this problem while we are down there (easily fixed by adding or moving one of the solar panels to the top of the unit) but we just didn’t have time to deal with the issue as we were packing up the house to leave.  This has resulted in the unit turning on about 6pm (local time) or on some days at about 6:30am (local time) when the sun hits the right side panel.   Again, we will fix this on our next trip.   The following graph shows the boot up time (local Curacao time on the vertical axis).   We are starting to see a trend of higher solar power (earlier boot up times) as the sun slowly moves.   Since sometimes we have clouds, even in Paradise, it is not a smooth transition, but it is definitely improving.  By the way, it was really cool being able to log into the machine down in Curacao and add the code to generate this graph remotely.

We are having the next guy down to the island in November 2017 adding a solar panel on top of the base Project Curacao2 base unit which should dramatically approve the run time of the Raspberry Pi.

We can see from the graph below that the solar power and run time is getting gradually better.  Some days it runs all day and all night.

The next graph shows the solar panel  and battery voltages for the last 10 days.  Again, improvement.   You can see the days that the Raspberry Pi Zero is running all day (the voltage is reported during the time the Pi is on).    Note that since the LiPo battery is not fully charged the solar voltage never gets much above 5V.   The running time is definitely getting better and better.

 

 

And finally, here is the graph showing the current generated by the solar panels (and how much is going into the batteries).   When the current coming from the panels really spikes high (~1500mA), the panels are in full sun and the limiting factor is how much current our solar power charger and controller SunControl can process (~1500mA).   All this information is coming from the SunControl board.   Love the data.

Conclusion

We are heading down to Curacao next week and will be working on the box on August 1, 2018.   We will report what we find.