Posts Tagged ‘total solar eclipse’

Functional Eclipse Computer

August 8, 2017

I have progressed on the eclipse computer project past a prototype to something that mostly works. There a few minor bugs in the code, but it is usable as-is. I’ve got fixes for them, but need to test before committing. The system still doesn’t use its buzzer for an audible prompt, and I haven’t written anything yet to help take a sequence of pictures showing the progression of the eclipse. I may not get to that for a day or two to deal with travel related issues. I plan on taking it, and a bunch of camera gear, to the Nashville area to view the eclipse.Eclipse ComputerThe picture shows what is nearly the final hardware, but older software. This video shows the current software. The hardware changes between the two are all to help keep everything from moving around in the case, and to keep the barrel connector that supplies power to the upper breadboard close to the board. The connect was moving away from the board too easily and causing a reduction in the supplied voltage.

All of these problems were solved by applying solid copper wires in the right spots. I use solid copper with breadboards a lot because I can cut the wires to the length I need and the conductor is stiff enough to be inserted into the breadboard without tinning or the addition of some connector. I used a few more of these wires to hold down the barrel connector and to apply some pressure against the top of the case. It worked out quite well.

I changed the display from a 16×2 LCD to a 20×4 one just for the additional text. The video shows how I’ve made use of the space. The 20×4 display is a bit dark and needs a backlight to be readable unless it is in bright direct sunlight. The 16×2 display didn’t have this problem; it is more like a common digital watch display in how it handles light. I made and adjusted an automatic backlight control program that gets brightness measurements from a TSL2591 at its minimum gain setting and uses one of the Raspberry Pi’s PWM outputs. It seems to make the display readable enough in bright light and keeps it from being brighter than it needs to be most of the time. I’d rather have a 20×4 that is more like the 16×2, but I haven’t got the time.

I tested the computer running on 8 Eneloop NiMH 2000mAH batteries. For most of the test, the computer was indoors and used the minimum backlight setting. It recorded around 600mW power consumption under these conditions. In brighter light, power consumption got as high as 850mW. Working out a new times of totality can add about 400mW, but I wrote the code to limit how often that occurs.

The batteries kept the computer running so long that I couldn’t finish a battery life test in one day. The combined runtime before exhausting the battery charge was around 25 hours. That was much longer than I anticipated when I decided on 8 AAs. I’m still going to use 8 because I can, and the backlight will make the runtime a bit shorter, maybe 16 or 17 hours, about twice what I need. Also, a smaller battery pack would have more room to move about, and I already wrote low battery detection code based on 8 batteries in series.

I’ve got the Raspberry Pi Zero running Gentoo Linux. I made modifications to the configuration used by OpenRC to boot up the system so that it starts the program that provides information on the LCD and the separate backlight control program. A simple Bash script keeps re-running the software until it terminates without error, or the script is killed. No GUI is installed. I’m going to see if I can get it to set the clocks on my cameras, and bring up a network connection, when the corresponding device is plugged into USB. Nothing critical, but it would be nice. Hopefully plugging in a camera won’t cause it to reboot.

Eclipse Computer Prototype

July 30, 2017

I’m going to see the upcoming total solar eclipse on August 21, 2017 somewhere north of Nashville. I’d like to get some good photographs of the event, which makes anticipating when certain eclipse events will occur very helpful. For this purpose, I put together a custom computer system to provide me with some information about the eclipse based on my location.

The prototype eclipse computer

The prototype eclipse computer

The computer is based around a Raspberry Pi Zero running Gentoo Linux. I used a GPS receiver from Adafruit along with GPSD to query the location, and NTPD to synchronize the system’s clock with the atomic clocks of GPS. I also used a 16×2 LCD that is readable in bright light, and with the help of its backlight, is readable in the dark. To power it, I’m using Adafruit’s Verter product; it takes 3 to 12 volts in, and produces 5.2 volts to run everything. It makes for a flexible power supply that doesn’t require me to buy a special battery that I might not use much in the future. I plan to use 8 AA batteries.

The software figures out when totality, the part of the eclipse when the moon’s umbra blocks the sun from view, begins and ends using some data published by NASA. The data shows the irregular shape of the moon’s umbra projected onto the irregular shape of the Earth at one second intervals. Using this, I made the program search for which of the umbra shapes contains the location provided by GPSD. The resulting totality time information, plus the current time and location, cycle across the LCD. Its results don’t exactly match many of the interactive maps available online, but I think they may be using a simplified, maybe circular, umbra shape and may not account for the Earth’s terrain.

At present, the program doesn’t take any user input. I’m considering changing that, but I want it to be useful without input. I’m going to put it inside a sturdy case that is water resistant enough that it should survive a strong downpour, just in case I have to deal with less than ideal weather. The top of the case is transparent, so the current version of the system can be useful without opening the case.

What it doesn’t do right now is show when the very beginning and ending of the eclipse occurs. I’m having some trouble figuring out how to do this. I’d also like it to show the azimuth and elevation of the sun for the current time, the beginning and ending of the eclipse, and mid-totality. I hope to make a time-lapse video of the event, and want to keep the sun in the frame, but do not want to disturb the camera once it starts. I did make an attempt at computing the values based on an algorithm published by NOAA, but what I made doesn’t produce correct azimuth values. Unfortunately, that is the more important value of the two for this eclipse.

I have published my source code as two repositories on Github. The first is a library I wrote intended to provide a C++ happy high-level style interface to using low-level hardware. I called it DUDS, for Distributed Update of Data from Something. I’m not very good at names. The name shows what I’d like to do with the library in the future, but I’ve got to build up other functionality first. It is already useful for this eclipse computer, so I hurried a bit to publish the code.

The second repository is for the eclipse computer program. It also includes a program to test finding totality times that does not use DUDS, and takes longitude and latitude values from standard input.

I’ll be doing more development and testing for at least a couple weeks.


False Steps

The Space Race as it might have been

keithlugcom.wordpress.com/

You Control The Action!

High Frontier

the space colony simulation game

Simple Climate

Straightforwardly explaining climate change, so you can read, react and then get on with your life.