Saturday, April 9, 2011

Final Report

PDF Version

Wasted electricity has become a social issue in recent years. With growing awareness of “dirty” energy (such as coal) many green interests have expressed an interest in reducing our power usage. Hydro companies have jumped on this as a chance to charge people for every kW they consume. Which has, in turn, led to a rise of energy metres.

However, these energy metres were made with the Hydro companies in mind, not the home owner. Such readouts only show a number – a poor indicator of the side effects power saving grants. We therefore set out to make a meter that was different from the others. One that abstracted away your power consumption and instead showed users the effects of their actions. We named our proposal Flower Power.


The original form of Flower Power was a block that plugged into the wall. On top of the block was a flower, which glowed when plugged in. The block itself offered a power socket, allowing another device to be attached.

The flower was an indicator of power consumption. If the attached device exceeded a set (and as yet undetermined) power draw over a given time, then the flower would begin to “wilt” - growing more limp over time while its light slowly fades. Extended time would cause the flower to lose more of its life, until it hung completely limp with its light off. This was meant to indicate that the flower had lost its “life” expending the effort needed to keep the device working.

When the device stops drawing power, then the flower would begin to regenerate. It would slowly stand up and begin to glow again. This indicated that the flower (like nature) could recover if you give it time.

At the time we were inspired by the work of the people at OpenEnergyMonitor, and decided to re-purpose their circuits for our purposes. The power consumption would be detected by an Arduino board, which would then control the flower's tilt. We had no idea how their power detection system worked though, and began investigations in the next week. We also began researching muscle wire and motors for use in our flower tilt mechanism.


Our first low level prototype had two changes from the previous one. First, the flower would be separated from the wall block (instead being placed in a normal flower pot). A power cord would link the pot to the plug, and the pot would indicate the power consumption of the circuit. The second change was that the block was replaced with an expansion plug. This would encourage our end users to use more power, while the flower discouraged this use. It was our hope that this ambiguous context would make people even more aware of their power use.

While progress had been made with understanding the power sensor, it soon became apparent we had errors in our wiring diagrams. Work on the prototype was suspended until such time as we could finalize the internal circuitry. We also decided to use a motor string “reel” system inside the flower stem to cause the stem to tilt.

Our second low level prototype offered many changes from the first.


We reassessed our flower pot, questioning why we used a flower in the first place. In the end we decided to stick with the flower. We realized that flowers are delicate and easily damaged - it was therefore more believable that a flower would wilt from power drain than a small tree. Animals were discarded as a viable option, as such shapes would encourage play with our prototype, leading to it possibly be broken. The flower itself was chosen because it is commonly linked with nature and green ideals. We wished to make users
see how they would “kill” those ideals by wasting power.

With the assistance of Carman we realized that both the LED and the droop feature on our prototype were performing the same function. We thus sought to find a use for both (or failing that discard one in favour of the other). In the end we realized there was no indicator of your current power use. You could see that you used too much power over the last hour, but not how much you were draining the flower at that moment. We thus changed the function of the LED light to show users their current power drain, with a dimmer light indicating greater power drain. The LED's dimness would be divided into 6 tiers of power consumption - the higher the tier, the quicker the flower would wilt over time. We chose 6 discrete tiers so that users could get immediate feedback when they reduce their power use, while still preserving the use over time intent of our design.

In addition, we received all the power sensors needed for our circuit during the week, and quickly realized they would make the plug far too large. So we moved all components into the flower pot and placed the 6 socket expansion on the front. This had the added bonus of making the power drain metaphor even more direct. We also finished analyzing the circuit thoroughly during the week (save for resistances of the devices) and created our first accurate wiring diagram (see Image 3)

At last, for our medium fidelity prototype we assembled all the parts together, taking great care to make the device look good. Our reasoning for this was sound – if our prototype looked hacked together, then people would not be inclined to trust it as a power bar. We thus spent great amounts of time finding the right pot and wires, melting holes in the pot with a soldering iron, and attaching devices to the pot such that the imperfections were invisible. We also too care to solder every electrical joint in the pot and cover them with electrical tape.

Our prototype is meant to be artistic in its message and form, but also very utilitarian. It can be used as a simple power bar, or as a decorative piece, or both. It also piqued our interest as a programming challenge - how can we get the Arduino board to sense power usage? We believe we have met all three challenges admirably.

We completely redesigned the flower wilt mechanism in this prototype. In previous iterations we cited muscle wire and string pulleys as mechanisms. However these proved to be impractical or impossible, based off of the technical knowledge of the group and our shopping prowess. We have therefore changed the mechanism to a rotational one involving a servo motor (see below) We also reworked our pot - incorporating a rock bed where the stem meets it. This bed of stones disguises the opening, making the flower feel more "alive". The stones were chosen for their low frictional value, so they would not hinder the servo.

The power sensors were constructed, and detailed wiring diagrams made for the whole pot.


The wall power is our 6 foot cord that goes to the wall. Upon entering the pot it branches two ways. One goes to a DC power supply which keeps our Arduino board running. The other line continues on, with the hot wire passing through a current transformer (CT) used in the current sensing circuit. The circuit then forks, with one half going to an AC step-down transformer (used to detect voltage), and the other to the sockets which the user will plug devices into.


From the 5V connector of the Arduino we have 2 resistors and a capacitor. This configuration has the net effect of adding +2.5V to the voltage produced by the CT. This prevents the power from flowing out of the Arduino board's analog in pin, causing the board to be damaged. The CT is paired with a 170 ohm "burden" resistor. The resistor prevents surges from the ct sensor (caused when devices are turned off) and helps to translate the current sensed by the CT into a voltage that the Arduino can read. The burden resistor has been specifically selected for our CT by us, so as to provide the maximum power range to the board with the minimum risk of surges.


This circuit has a similar setup on the 5V pin, and it provides the same function - it adds 2.5V to the power given by the AC transformer. The other two resistors step down the voltage from an alarming 7V (from the Arduino board's point of view) to a more respectable range of about 4-5 volts.


The flower is made of a floppy plastic tube covered in green tape. The tube has a certain sag to it - giving a natural feel. To "wilt" the flower, we rotate a disk the tube is attached to inside the pot, making the tube lean forward. The tube sags in that direction, making the feeling more natural.

Image 8: Medium fidelity prototype

Our final prototype can be seen above. By the time of deployment all power sensing was accurate to around about 2W. However, we found that the stem control mechanisms were not working properly. When the Arduino board powered on it automatically caused the flower to rotate forward 90 degrees. It would rotate back again after around a minute of waiting. In addition, all LED functionality did not work. However, all circuit tests of the pot proved it to be sound, and there were no errors in the wiring of the power expansion. We decided to deploy our pot anyway to get a sense of what effect stem angle would have on our test family. Sadly, things did not go according to plan.

For our deployment, we chose to give our prototype to a family of three: a mother, father and and son. They live in a relatively unremarkable home, comparable to the one we ran testing in. The son in the family is a fellow SIAT student who is known to John. The mother is a housewife, who is very power-conscientious. The father is a retired individual.

Due to the presence of the mother, the family is well known for its efforts at consuming power. The household will often turn off lights that aren't in use, or devices that aren't very effective. We felt that our device suited these ideals admirably, giving us a chance to see the reactions of those who would be the most interested. We assumed that it would likely be deployed in a work area, such as a computer desk - though other avenues were considered, such as a kitchen or bathroom.

Our intent was to get them to realize how much (or little) power they consume. However, when deployment began problems started to surface. The father and mother were reluctant to be on camera because they are not able to communicate well in English. We allowed for this by only recording the son in the family, and giving our device to him. We first introduced the device as a standard power bar, and told him to use it as he wished. However, when he hooked up his devices (a laptop, phone charger and gaming system) to the flower pot there was no reaction from the flower. Our user tried to plug in different applications to our device to see if there was a change of behavior, but nothing he tried seemed to work. He did note that the led blinked on occasion though – which was odd because it did not work at all before deployment began.

Such odd behavior gave him little indication of power use at all, and because of that we were unsuccessful in changing his opinions. However, the deployment did give us much useful feedback. One of the first things we have learned is that we need a broader testing phase before we deploy our flower pot. Despite extensive testing at Mark's and YiTian's houses our flower pot still failed to work in the deployment house. It is possible that there is a difference between the houses which may affect the pot's effectiveness. It is also possible that there are more errors in our code and circuitry we have not foreseen. The errors we had at the deployment house had never been seen before, and warrant further investigation.

Our interviews also gave us insight into the preconceptions the son had toward power use. He seemed to think that more technologically advanced objects (such as computers) would consume more power than less advanced objects (such as hair-dryers). In reality, hair-dryers consume MUCH more power than laptops – it is quite possible our design could cause him to re-think what devices he uses. He also received little punishment for using too much power, though he did show an awareness that saving power was a good idea. It is possible that we could make him save more power by nudging him along with our flower pot.

Lastly, we learned that Flower Power had made a good impression as a power bar. It was portable, reliable and good-looking. Our interviewee pointed out that it was well suited as a tabletop power bar, providing sockets on the table so you need not plug into the wall. He even moved it around in an attempt to get it to react. However, the interviewee did note that he would not buy it himself, stating that its simple design was not well suited to a standard household. We were unable to explore this avenue further, but it may be one worthy of investigation.

The overall deployment of Flower Power was a failure. We did not succeed in changing any perspectives on power consumption. However, this failure can be used to improve our design, and get better test results next time.