A little solar powered pendulum that I made a few years back. It uses a very simple circuit with just 2 transistors, a couple of resistors, a diode and some capacitors. Power is supplied by a two calculator solar panels wired in parallel for faster charging. The power is dumped into a coil which repels a magnet (disguised by some old brass gears) hanging from some fishing line. As the magnet swings back towards the coil the EMF generated lights a red LED in the top post and the power from the capacitors is dumped into the coil again giving the pendulum a little kick, forcing it higher.

Video, more pics and the schematic after the jump.

Pendulum swinging past the driving coil

The circuit is based on the following Solarbotics schematic that I found.

A view showing the LED and solar panels

Detail view of the pendulum gears

The messy guts of the thing

Close up of the rats nest (messy) circuit inside the box

Very luckily I came into possession of a free electric kiln. It came with an old temperature control unit that I’ve decided to get back into working order so I can fire my own ceramics. The controller is made by the Industrial Pyrometer Company, which now seems to have become Mitsco. It uses a clever cam-follower system to regulate the kiln temperature and heating rate. The cam wheel has a scale laid out on it with the rings corresponding to 100C increases and the radial bands equaling 2 hour periods (a full rotation takes 24 hours). A sprung arm follows the edge of this cam around and through a system of gears, rotates a potentiometer inside the unit. An R-type thermocouple probe is used to monitor the temperature inside the kiln providing feedback to the control unit, which is compared to the cam-follower position using a simple Op Amp circuit (based on an F709PC chip). A relay is then triggered to turn the kiln on or off.

When I opened the controller up I found it in surprisingly good condition. It was very clean and the only obvious problem was an Electrolytic Capacitor that was oozing goo.

Dodgy capacitor oozing goo

It was a fairly quick job top replace the cap with a nice new one. All the other components looked fine so I left those alone.

The next task was to figure out how to connect the thing up. Using a multimeter to trace the existing wires back to their connections made this fairly straightforward on the external side. Internally took a bit longer as I had to track all the wires back to their various components. I’ve drawn up a nice colourful schematic of what I found.

External wiring block

Pyrometer schematic, rendered in coloured pencils

I hooked up the mains power after installing working out what should be comnnected where and held my breath…luckily nothing exploded and everything seemed to be ticking over fine. I hooked up a light bulb as a test load and used a cigarette lighter to heat the thermocouple probe.  It seems to function as expected, switching the light on and off in relation to the temperature and position of the follower arm.

The relay inside the controller looks a bit too wimpy too switch a kiln on and off, so I’m going to make an external relay box with a nice beefy relay in to handle the actual switching and hook the wimpy relay up to the coil of that one.

There are a couple of variable resistors on the PCB inside the controller that seem to be used to calibrate the thermocouple voltage against the cam-follower arm position. My next task is to hook up the the thermocouple using the new compensating cable that I bought and then put the probe in an oven at a known temperature (about 100C should do it). Then I can fiddle with the resistors on the control board until the relay switches on and off at the right temperature. That’s coming up in part 2.

Inside the controller

Servo with new feedback wire

Today I’ve modified two of my servos to allow access to the output of the variable resistor inside them. This very simple modification opens up a world of possibilities that really should come as standard on all servos. All that’s involved is opening your servo, locating the potentiometer that provides feedback on where the output shaft is and then adding an extra wire onto the center tap. After adding this wire you can read the voltage present using an A/D converter and following some simple calibration, know quite precisely what angle the output shaft is at.

The actual modification is discussed in detail over at Trossen Robotics so I won’t go into that too much.

Here’s a video of what I cooked up using the newly modified servos and an arduino board.

You can see that as I twist the horn of one servo, the other rotates to match it and mirrors the movement very closely. To do this the value from the feedback pot is read using the analogRead() function. As the output of the feedback line only reaches around 2 Volts at maximum (and goes down to around 0.2V at the other end of travel) the AREF pin of the arduino must have a voltage just above this applied to get good A/D resolution.

To scale the A/D readings I connected a simple voltage divider between GND, +5V and the AREF pin. A note here is that when I measured the maximum output from the feedback pot without the servo being connected to the arduino I measured 1.2V for one and 1.3V for the other and made my divider to output around 1.37V. However, when I connected the ground from the servos to the ground of the arduino board, the voltage seen at the outputs moved closer to 2V, which messed up my readings and meant that the A/D converter was reporting a value of 1023 (max) at about a quarter of a rotation of the servo. This was down to the fact that I was using a separate power supply for the servos which was obviously mismatched slightly from the arduino board voltage. So make sure you hook everything together before you measure the servo voltage and work out which resistors to use in your divider. Incidentally, I used values of 5.1Kohms and 4.7Kohms, worked out using this calculator.

The code I used on the arduino was largely based around the example on the Trossen Robotics page. It’s available to download below.

Download the Arduino Code

Here I’d already removed the guts. You can either cut all the wires, or do what I did and use a small screwdriver to lift the metal springs in the terminal block that are holding the wires in. After that remove the central screw as shown below to seperate the PIR section from the light.

To seperate the rotating section of the PIR from the container housing the electronics just needs a good firm pull. The two sections are held together by a couple of plastic clips which should pop out easily.

My PIR housing seemed to have glue in the seam between the two halves, so I cut a shallow channel around it until the two halves were easily seperated. The case was also held together with retaining clips as well as glue, so make sure you don’t cut through these, it’ll make putting it back together much easier.

Here’s what the insides look like when you finally get the sensor housing open.

The important bit here is the relay. It’s part number is 812H-1A-C made by Song Chun. You can see the datasheet here.

What you can see from the datasheet is that the live wire is connected to one side of the relay output contacts and the white wire that goes to the halogen light is on the other side of the contact.

To allow the PIR to function as a general purpose switch we need to desolder the live wire from the relay and reconnect it so that it powers the circuit, then add another wire to the relay.

Below you can see where you should be cutting and desoldering.

And here’s the slightly messy, but effective result. Make sure you test the cut track for continuity with a multimeter, you really don’t want any mains voltages getting onto the output side of your relay.

In the next shot you can just about see where I’ve resoldered the live wire onto the leg of the resistor so it can continue to power the detector circuit. I wrapped it round the leg to make the join a bit stronger. I’ve recycled some ground wire from a piece of flex and soldered that into the hole that the live wire was removed from.

After you’re done soldering, you might want to hook everything up and make sure it’s all working. I hooked up the new relay connections to a multimeter to test that it was turning on and off as it should, and also that there weren’t any stray mains voltages that could cause issues later…such as death.

Once everything’s checked and working you can repackage it in the PIR housing. I put the original terminal block back in to hook up the mains wires and then added a couple of terminal posts through the holes on the front. These were attached to the relay wires and will now allow me to switch things on and off under control of the PIR sensor.

For a start, most BEAM robots can be assembled out of mostly junk parts. Even if you have to buy new parts, there is usually a very low component count for each robot, making each project cheap.

For this project I wanted to make a small flower that responded to light in some way, using only components that I already had to hand. This is what I came up with:

Components used were as follows:

• Panasonic BP-242221 Solar Panel
• Pager motor
• 0.047F 5.5V Capacitor
• 1x 1K Resistor
• 1x 100K Resistor
• 1x 220K Resistor
• 1N4148 Diode (or similar)
• 2x 2N3906 PNP Transistors
• 1x 2N3904 NPN Transistor
• An empty beer can for the petals
• The base of an old PP3 battery for the stand
• The wheel of a small toy car used to mount the flower to the motor
• Wire for connecting components
• A glue gun & plenty of glue

When the flower is exposed to light the solar panel is charging the storage capacitor. As the light level drops off, the drop in charging current triggers the circuit and the energy stored in the capacitor is dumped into the pager motor, spinning the flower. In practice this means that when the sun is out the flower is charging, then as it is covered by clouds the charging current drops and the flower spins very rapidly.

Below is the schematic for the flower. It is based on the ‘Type 3 Solar Engine‘ design by Wilf Rigter. I didn’t have the same transistor to hand, but as with most things in the BEAM world, parts can be substituted for ones of a similar type with little effect on the final performance.

The components were free-formed using the ‘rat’s nest’ construction method. This doesn’t result in the cleanest finish (not in my hands anyway) but it does allow for a small form

A glue gun was used to secure the components and the bottom of an old PP3 type 9V battery was used as a support to keep the flower standing up

The flower was made out of an old beer can. Layers of petals were cut out in descending size and then stacked and hot-glued together.

This version was more of a quick prototype that a work of art, as you can tell.

I found that the wheel of a small toy car was a perfect fit onto the axle of the pager motor. This provided a nice flat surface to glue the flower too.

Here’s a video of the flower in action