main goals:
For this project the main focus of the hardware was the ability to collect power from the surrounding environment and to use it for logic and communication. The main focus for the housing was to have a waterproof cheap housing that could protect the electronic hardware and be shaped easily.
Below is a overview of the techniques used:
First is a schematic of the inside hardware that each module contained.The different systems are numbered below.
Technical: overview
GENERAL OVERVIEW
1. Speaker:
A 16 ohm speaker was used to draw less power without lousing amplitude and a plastic membrane was used for waterproofing.
2. Sound circuit:
For the production of the tones a circuit based on a digital pot and a 555 chip is employed. This allows the tone to be set with only one call to the digital pot from the micro controller, allowing the controller to continue to function even when making a tone sequence.
3. IR sensor:
To detect movement a IR sensor is used.
4. Logic control:
For the logic control I used a ATMEGA 168 chip.
5. Power storage:
For the power storage I implemented two different storage techniques, four modules use NIMH batteries to store power and four use capacitors to store power. Each module gets power from solar panels mounted to the top of the modules.
6. RF device:
For communications I used Xbee modules running the Zigbee protocol and the older Znet from digi (I am still deciding which is best).
The circuit is divided into two parts, the sound generation circuit on the right and the logic communication on the left. Two grounds are employed in order to switch the sound circuit on only when needed. Because of this blocking diodes are needed on the SPI interface to get rid of the creation of any virtual grounds (they are pictured going into the mcp47100) and also on the ground from the solar source.
This circuit was designed with two things in mind, power conservation and a quick reliable start up after power lost. To achieve these a compartmentalizing of function was used to be able to power up and down the circuits only when needed.
Using the mcp47100 digital pot with the 555 Timer frees up the microcontroller to keep dealing with incoming RX TX data while still generating a tone. There is also a slight power savings because the chips functioning power can be drawn from the second power source instead of through the microcontroller.
Circuit boards:
I then made boards using EAGLE CAD with large ground planes to help reduce the noise of the system.
HOUSING OVERVIEW
HOUSING OVERVIEW (A big thank you to Monica Johnson who built the majority of the modules and helped research the waterproofing techniques):
For the housings 3/4" ply wood was used in combination with silicone and an enamel overcoat. I originally was going to TIG weld aluminum casings but wanted the materials to be more accessible. In putting a emphasis on using easy to obtain non toxic materials the viability of using "home made" housings will be tested and documented. I am curious to see the long term sustainability of using these techniques and hope it will be helpful to others.
The stages of construction:
First the modules were designed and sketched out:
From those sketches the creatures were outlined and cut out to form the animal's shape, using basic tools: jigsaw, table saw, router ect. (Monica's hands):
Next foam insulation was placed in. The foam had two functions the first was to insulate the electrical components and batteries from freezing the second is to add a layer of moisture protection. The foam I ended up using was a balance of cost compared to effective insulation:
Like most everything you can get a good overview at:
www.mcmaster.com
These are the tech specs of the foam that was used:
Flexible Low-Temperature Polyethylene Foam Rubber Insulation
* Temperature Range: -160° to +200° F (flexible to -100° F)
* Heat Flow Rate (K-factor): 0.25 Btu/hr. x in./sq. ft. @ 75° F
* Density: 1-2 lbs./cu. ft.
* Color: Black
Next the housings had to be sealed up in reparation for sanding. Regular silicone was used to seal the edges of the modules, marine wood glue and regular drywall screws were used to fix the seams.
The most important part of the sealing process was the outside coatings of enamel, this is what gives true protection to the housings. It used to be that any decent metal or marine enamel was made of the most toxic, hanis stuff. Now though water based products are coming on the market as viable solutions to enamel coatings. Below is what we are trying out with this project (if your in the New York area Abbot paints in Brooklyn is a pretty good resource for paints and sealants).
To seal the housings the below were used:
INSL-X AQUALOCK PLUS 100% WATER BASED PRIMER/SEALER
with 3 coats of
CORONADO RUST SCAT LATEX HIGH GLOSS ENAMEL
The last was to put all the circuitry in enclosures and plug them in. The enclosures are just regular ABS plastic that was drilled out to accommodate the connections.
I hope to document the effectiveness of the sealed wood housing over time.
WEB GATEWAY OVERVIEW:
To connect to the interweb an XBEE coordinator was attached to a ATMEL chip and a LANTRONIX XPORT. There are a lot of gateway choices out there going from RF or 802.xx to Ethernet. Digi makes a bunch so does Lantronixs, I used the xport because it is fairly cheap and easy to code for even if it is a little buggy. Basically all the circuit does is receive a RF communication and (if it is not junk) makes a PHP GET request to the servers web page. In this way the data is recorded on the internet.
Power Storage OVERVIEW:
Two different storage techniques were used with this project. The first system stores its main power using NIMH batteries, the second uses ULTRA capacitors to store it's main power. Both systems used a bank of secondary capacitors to store energy for the sound production. Two methods of storage were used to test which method is more viable for intermittent power systems.
Power storage
For the Will o the wisp modules I broke the system up into three parts;
The first part was the power system for the logic and function controls (micro-controller frequency generation and turning modules on and off), the second was for the amplification of the sound and driving the speaker, third was the power needed for the charging of the batteries/ultra capacitors.
Battery size considerations
The battery-bank is comprised of 4 NIMH AA batteries, which are rated at 2100mAh; this was chosen to balance the solar panel, which supplies a 300ma maximum output.
So:
300 mA (solar supply)*8 hours = 2400 mA - (50mA*8 hours usage)
->Gives a possible 2000mA of storage for a day of super equatorial sunlight.
And a full battery could run the logic
Total storage /8 hours of use a day
2100mA/ (50mA*8-hours)= 5.25 days of operation off a full charge
I split these into separate systems because each had a different power need and use.
->The charging circuit needs a small current only when there is power to put into the battery.
->The speaker/amplifier will take in an unregulated source which has a much higher current draw but only needs it for intermittent short bursts.
->The logic control system needs a regulated constant small current.
The first is the battery charging system, which takes the variable solar energy
and regulates it in order to cause the appropriate chemical reactions in the batteries. This process allows the battery to develop a voltage, which can then be used as a constant source of power later.
The charging system requires 5-36volts to operate and will draw 6ma of current for the circuitry to operate so the available power will always be:
INPUT-6mA
MORE HERE ON HOW IT WORKS
For charging circuit components:
3mA for 555 chip
1 mA for comparator
2mA for transistors + voltage divider
6ma total
The battery charging circuit employs a comparator with a 555 chip to apply timed pulses to the battery according to the batteries charge voltage. This is the final circuit schematic I replaced the lm 399 with a lm311N and the 555 timer was changed over to a CMOS low power version (lm555CN) saving ~3ma! Also the timing capacitor was changed (10uf -> 4.7uf) because I am using half the voltage of the original circuit.
The second system is used for the logic and RF communication. This system needs to be tightly regulated in order to keep the RF and logic functioning properly. The system runs off the battery bank, which is then regulated by a LM low dropout 3.3-voltage regulator. The battery bank is charged to 4.8Volts and the regulator can give 3.3V @100ma with a drop out voltage of 130mV (which means it will provide a 3.3 voltage using the battery as long as the battery is in the range of 4.8v-3.43v), this gives a powerful second layer to keeping the power supply constant.
Power usage overview of components
Digital pot 500 uA
TIP 120 *whatever you put in base
TC1262-3.3vab 80 uA
TS 555 100uA
LM386 4 mA
LM339 800 uA
Atmega168 8ma
xbee 40ma
Total:
53.48mA for logic and RF communication.
The third system runs off a bank of 2.7 3F ultra capacitors that feeds into the lm386 amp driving the speaker. This system is directly charged from a separate solar panel, I kept this separate for two reasons first to cut down on the noise of pulling 200mA off the same circuit as the logic source. Second is conceptual, it is interesting to see the current changes in the available solar energy by not averaging it out as much over time. This allows the system to show which module is doing well in terms of power through the volume of the speaker as it plays.
200 ma for amp+speaker (intermittent on a separate system).
COMPONENTSThere were many considerations in picking the components that make up these modules here is a brief overview of the components in relation to their functional.
list of components here:
MICRO-CONTROLLER
The MICRO-CONTROLLER is the "brains " of the operation it is the one piece of hardware that should be operating at all times. It coordinates the status and functioning of all the other peripherals. For this project I am using the atmel's Atmeg168, I chose this microcontroller for several reasons. The chip is versatile has analogue in and out and supports SPI communication, it also has low power sleep modes. Using this chip I can also mix Arduino libraries with lower level programming functionality to achieve greater control. Another major design consideration was not implementing a bootloader on the chip it is important for two reasons. First and most important was that I found when using an uncontrolled serial in source (the XBEE) the microcontroller's bootloader got confused by the Xbees input during a startup/reset of the chip. This is very bad if your working with intermittent power in a device that is hard to get to when deployed. The second is that it frees up a tiny bit more of programming space.
The circuit for the Atmel is the most inclusive it is tied into almost all the other peripherals in the device as you can see from the diagram it is hooked into the IR sensor (to monitor presence detection), the XBee for the RF communication, it control’s the digital pot using the SPI protocol, controls the on off function of speaker with the TIP120 and monitors the capacitors voltage through the voltage divider. The batteries power it through the 3.3-volt regulator.
dataSheet http://www.datasheetcatalog.org/datasheet/atmel/2545S.pdf
ATMEL LINK http://www.atmel.com/dyn/products/Product_card.asp?part_id=3303
For the communications I am using the XBEE series2 Znet modules I might switch over to Zigbee ZB. This provides me with true low power mesh networking for the system. The network system is made up of three module types; end devices, routers and coordinators. These modules work together to keep accurate routing tables of all devices on the network so the message routing protocol is taken care of for you.
The systems routing tables are dynamic in real time, being able to re-route messages if a module goes down. This ability to "heal" the mesh dynamically is huge in terms of making dispersed systemic functioning; it allows all module great freedom of operation while still being able to coordinate their movements with the larger system. For instance the modules can pop in and out of sleep mode whenever their power levels dictate it without worrying about crashing the whole system.
Otherwise if a link had to be established all the time the modules power would be redirected into staying awake and routing messages instead of monitoring its own environment and function. The only module that has to be on full time is the coordinator, which sets the original network settings. This should be set with a redundant power system or hooked to mains.
The actual circuit of the Xbee device is relatively simply it is powered off the batteries through the 3.3 volt regulator and communicates with the microcontroller by the RX and TX pins.
zigbee link
DIGI http://www.digi.com/products/wireless/zigbee-mesh/
overveiw of xbee: www.johnhenryshammer.com/WOW2/pagesHowTo/networkOverview.php
Infrared detector
The decision to use an IR detector was made because of the location of the network. It is going to operate outside in upstate New York where there are a lot of bats so the original plan of using sonar ping sensors was unfeasible. I didn't want to mess around with building my own module so I got a little digital IR sensor, complete with fresnal lens for about 7 dollars. It is a little sensitive (has a range of about 15') but can be restrained by reducing the output surface of the lens.
The circuit is very basic you feed power and ground to the module (it says it operates 5-9V but I found it works fine on 3.3V) it then has a third line that goes high when the module is triggered, this is connected to a I/O pin of the Atmel chip.
dataSheet: http://www.futurlec.com/PIR_Module_B.shtml
I am using a small circuit built around a digital pot and a 555 chip to generate the "call" or output tones. The method
involves using the 555 chip to generate a square wave pulse which is sent into an amplifier. To change the frequency (and therefore the tone) of the 555 chip the a change to the digital pot controls the discharge rate of the 555's cycle. The digital pot utilizes the SPI interface (http://en.wikipedia.org/wiki/Serial_Peripheral_Interface_Bus) to control its internal resistance.
This allows the microcontroller to send it one message and have the digital pot worry about the upkeep of the signal, which frees up the controller to perform other tasks. Both the CMOS based 555 chip and the digital pot are very low powered devices so there is not an appreciable amount of overhead as added power . The tones that are generated from this method are very "digital" the output is a perfect square wave with no overtones so it has been a real challenge to try and get interesting sounds form this setup.
The resistance from the digitalpot is fed into the 555 chip, which changes the frequency of the output square wave, in effect changing the tone generated. This signal is too small to produce any meaningful drive of the speaker so it is fed into an lm386 amplifier, which utilizes power stored in the super capacitors. As the power is used the voltage in the capacitors drops, this drop is read through the voltage divider into the microcontroller. When the monitored voltage drops enough the microcontroller de-asserts the active pin to the TIP120 breaking the connection of amplifier to speaker and shutting of the "call".
digitalPot dataSheet
http://www.chipcatalog.com/Datasheet/B0E11FEFF34648DA1777E9A536316283.htm
google 555 timer