Microcontrollers and YOU!

It's been a while since i posted a project but i think its about time we make use of the tools we have.

I've always wanted to make a simple bot and it should be no surprise that i'm in the solar boat club, so i figured i would combine the two.

I have a small 5W panel in my lab; its too small to directly power any significant equipment and too small to let set on a desk and do nothing.

Usually panels like these are used to passively charge lead acid/whatever batteries. They provide around 12V and usually more than enough current to charge a decently sized battery in a few hours.

However, i wanted to do something different, i wanted to use the panel to power a small robot that had a backup power source. If i made the systems efficient, i wouldn't need that much power to make it move about.

The obvious choice for a backup power source is a battery, although this is true, batteries get a little pesky because you need charge controllers for them. I could design one, but they're not the most thrilling things in the world to build and design.

A better alternative is a capacitor bank. If i stick to my rule of only using minimal power to get about, a capacitor bank makes sense (especially considering how easy they are to charge)

The biggest problem with capacitor banks are the less than spectacular energy density they offer. Energy density is how much energy a medium stores per unit volume. A battery will give you much more energy per unit volume than a capacitor, thus for equal amounts of energy, you need a larger bank (significantly larger). Enter the super-capacitor

Super Caps:
Super-capacitors offer much superior energy density compared to an electrolytic capacitor. The capacitors I am using for this project are 5.5V 1Farad capacitors from: http://www.all-battery.com/. Which you can find here: http://www.all-battery.com/55v1capacitancecoinsupercapacitor-1-1.aspx

Props to all-battery.com for sourcing such cheap and powerful capacitors.

Lets compare the energy densities of our capacitors:

We recall that energy stored in a capacitor is given by the following formula:

We can see that the super-cap has a 100 fold increase in energy density over the electrolytic cap.

How does it do it?
Super-caps are really a hybrid between a traditional capacitor (Separation of plates) and a chemical battery. The key is with super-caps, to use nano-scale barriers to separate our charge, increasing surface area. We also use the space for the dielectric to hold charge (where the icky chemical part comes in). These two things combined help increase our energy density. There is one downside, super-caps sacrifice the ability to store large voltages (because of the nano-sized gaps), have increased leakage currents, and take longer to charge than conventional capacitors. However the ability to store much more energy is worth it.

So why super caps again?
Even at 100 joules per cubic centimeter, a typical alkaline battery dwarfs it at 1100 joules per cubic centimeter. However, the ease of charge and use of a super capacitor bank makes them very attractive. Also another thing to note. Capacitors deliver instantaneous current at rates much higher than a battery. Although a capacitor has lower energy density, the power density is much greater for a capacitor, it can discharge its entire energy contents in a fraction of a second while batteries take seconds to discharge completely. (This of an explosion versus a slow burning match)

The Goods
Each super-cap holds 12 Joules of energy at 5 volts. A typical AA holds about 1kj of energy. So to match the capacity of a AA battery i'd need on the order of 100 super-caps. (Not on my budget) Luckily, we don't need 1kj of energy for our robot, we only need enough energy to move the robot (not far).

For this bot we have 30 super-caps:

They came packed in a tray very well via first class mail. My order got here in a few business days (not bad for free shipping)

Just like electrolytic capacitors, super-caps are polarized. a word of caution with these however. Because of how thin the are, the manufacturer didn't leave enough of the marking band on to show the polarity of the caps (and the leads are equal length). However, it should be noticed that the side with the dimple is + and the flat side is -. Thus in my picture above, the capacitor is facing with the + terminal upwards.

The Pack

For this project we are putting our 30 1Farad capacitors in parallel for a 30F pack, this yields us 375 joules. A little more than 1/3 of a AA battery. Still quite a capacitor pack.

In this bank i have the capacitors in an alternating pattern to put them all in parallel like so:

+|--|++|-

Once you hook all the + together and all the - together, the pack is in parallel.

The backside of the proto-board looks like this: (to give you an idea of how i wired it)

The Idea:
Although 5V isn't a huge system voltage, its convenient for our micro's and motors. at 5v our motors are self-current limiting. They don't draw as much power at 5V as they do at say 12V, this is nice for current limiting. The biggest reason is that these caps are rated at 5V and i'd prefer one large parallel pack rather than a combination of series and parallel.

So, our source of energy is our solar panel:

Solar panels provide power by converting photons into electrons (kinda) Each panel has a few ratings. The open circuit voltage (Voltage between bare terminals) the Short circuit current (Current through a short) and the maximum power point.

For maximum power you want to keep the load of the solar panel at this voltage current point. Since i don't have money for a maximum power point system ill do one better, a switching power supply. Although not a MPPT system, with my switching power supply i can adjust the load on the capacitor bank to adjust the load on the panels, thus with some math and some good logic i can easily implement a sudo-MPPT system using a cheap little switching power supply chip.

I will be using the LM2674 buck converter IC to convert the panel voltage(anywhere from 0 to 18V) into 5V. The advantage of a switching power supply is that it works for a wide range of input voltages and is very efficient.

http://www.national.com/pf/LM/LM2674.html#Overview

(Thanks to nat semi for the samples!)

So that covers the power system, i'll be using a 5 watt panel to charge a 30F capacitor bank via a switching power supply.

The logic:

For the logic on this project i will be using a PIC18F. What model i don't know, but what i would like to do:

-Analog input for sensors (LDRs, color sensors etc)
-Serial In/Out for radio/infra red communications
-PWM out for motor drive

A potential candidate is the PIC18F26J50 i recently sampled a few and after checking them out i'll make my decision.

I also might be introducing the use of C to program. For robots C is ideal because we have to use complex data structures but we don't care how long it takes to process them.

Whats next?

I have to begin fabricating the robot's chassis and i will begin testing the power systems. I have to implement a charge controller on the Cap pack and begin designing the mechanics. So stay tuned!

Also any feedback is appreciated!