Low power project ideas

[kodon]

Hi,

I ordered the MSP430 and some solar panels to create some kind of simple project. The problem is that I can't think of anything useful and yet relatively simple project... So I'm asking you for suggestions for a project. BTW I happened to like very much the CCS compiler that I used to make simple led blinkers. You can really see what is happening inside the µC.

My aim is to learn something about lowpower stuff I guess. TI's controllers seemed very efficient in that perspective. Without any powersaving it kept running for 3 weeks blinking the onboard red and green LED's. That was with two AA sized Eneloop batteries.

When I have an idea for a project, it usually has some sort of wireless communication link and I think I don't have the guts yet to accomplish a project with that complexity. So if you have an idea worth doing, bring it on!

Regards: Kodon

[T4P]

MSP430's according to them are ultra low power devices ... 
Their EZ-430Chronos Watch is a prime example of what a watch can be ... 
Microcontrolled and very low power .



But you can use a MSP430 to ,
how about a MSP430 temperature and humidity logger ?

[westfw]

Here are some things that I've had on my "list":

1) Consider the cheap solar-powered garden light.  It has a small solar cell that charges a small battery during the day, and a photocell that detects "night" and turns on an LED.

Now, the ones I have are pretty stupid.  They need that photocell even though both the solar cell and the LED could be used as light sensing devices.  They tend to go on at dusk, well before they are needed, and as a result use up a significant part of their charge before it even gets dark.  These are things that could be fixed by adding a low power microcontroller and some pretty simple code.  For the more advanced experiment, I'd think a microcontroller could implement a more efficient switching supply for the LED, too...

2) Consider  the "BEAM Robotics" "Solar Engine."  For these, a solar cell charges a large capacitor until some threshold voltage is reached, and then permits the cap to discharge though a motor or other circuitry.  This is usually done using some clever analog circuitry.  That circuitry frequently involves a "voltage detector" like the panasonic 1381x.  But that IC has a fixed voltage threshold and is about as expensive as the low-end MSP430s.  Could an MSP430 be a replacement (using a combination of the internal brownout circuitry, and active monitoring?)

3) Can you do "small wind power" energy harvesting devices using (for example) a HDD spindle motor and a child's pinwheel?  Perhaps getting enough power to periodically report wind speed?

4) how about a device that I can put in a prospective solar-power location with a prospective solar cell, and will record just how much power I get over a given day?

[kodon]

Nice thoughts

Yes, I have the Chronos watch also.

I almost ordered a bunch of supercaps for trying to make a device that does not need any external battery But then again, I have a heap on unused Eneloops lying aroud so why not use them...

The solar cells are spec'd 3.6V and 210mA @ max so I could charge the batteries with some schotcky diodes.

The garden light could use the µC's internal Real Time Clock and shut down at midnight. I have a street lamp outside my window and it drives me nuts when I leave my window blinds open.

Which brings me to an idea of automatized shade blinder Solar cell detects if the sun is rising and opens the blinds, and then again at midnight it closes them allowing me to sleep. Waking up is a nicer experience (according to Philps marketing department) when you get a real daylight. Formerly I was planning to make a wakeuplight but that one nevere made it... Anyway that would require a small DC motor to turn on and off the blinds. If the solarcell charges the batteries enough, it could maybe actuate itself twice a day and then just collect more energy.

While testing the solarcells, I noticed that sunlight is much more efficient powersource than artificial lighting (lightbulb, LED etc). So it could be possible to detect just sunlight apart from the street lamp. And by the way the lamp shuts down at 2 o'clock at night time.

but anyways, great ideas... keep 'em coming!

PS: this lowpower thingy was actually pretty interesting, 6 or 7 months back I made a temperature sensing wireless node with Arduino and HopeRF radios. It sends the temperature of my sauna so I can check it while being in computer The batteries ran out just few days ago. It was an interesting project also.

[Rerouter]

hmm, makes me wonder if there is something even lower power.. something i am designing has to have a total system consumption under 1mW

[jpelczar]

Maybe you could make a clone of Dave's microwatch or scientific calculator ? The other idea might be to create RF remote controller working at 433MHz ?

[amyk]

hmm, makes me wonder if there is something even lower power.. something i am designing has to have a total system consumption under 1mW

You'll probably need to use a 4-bit MCU at those power levels. Something like the EM6607 only draws 3uA at 1.5V, or 4.5uW.

[BBQdChips]

How about doing an RF Thermometer.

As an added feature later (in software),  add memory for historical data, say, by the hour.  Add a visible light sensor for a makeshift cloud cover sensor.  Basically, a weather station. Might be fun.

[kodon]

Maybe you could make a clone of Dave's microwatch or scientific calculator ? The other idea might be to create RF remote controller working at 433MHz ?

I think the project idea should be something used outdoors, and the RF part sounds too difficult task for now

Anyway, I just managed to make a Real Time Clock out of the MSP430. It was a struggle with the Code Composer Studio v5 For some reason I couldn't make a new project working, so I had to make it ower an example. D'uh But here is something (attached images) what I did today. I used just some hot glue to keep things together. The RTC code draws ~60µA according to my 5$ multimeter wich is not very low power yet (without any LED's).

Edit: This thing looks actually pretty promising. I tested the RTC code without the batteries, and it barely manages to keep alive and blink the red led at my desk. This is just by using 3pc's of 40 Watt light bulbs at my roof and no extra caps! They're spot lights pointing away so I think this could stay alive for a long period of time.

I'm not using any charging logic for the batteries right now. Just a schotky diode to prevent the batteries discharging to the solarpanel.

[Zero999]

1) Consider the cheap solar-powered garden light.  It has a small solar cell that charges a small battery during the day, and a photocell that detects "night" and turns on an LED.

Now, the ones I have are pretty stupid.  They need that photocell even though both the solar cell and the LED could be used as light sensing devices.  They tend to go on at dusk, well before they are needed, and as a result use up a significant part of their charge before it even gets dark. 

The ones I have don't have a photocell, just a solar panel which doubles as a light sensor. I've not had a problem with them turning on too early. My main criticism is the shoddy construction which allows water ingress, causing them to fail within a year.

These are things that could be fixed by adding a low power microcontroller and some pretty simple code.  For the more advanced experiment, I'd think a microcontroller could implement a more efficient switching supply for the LED, too...

Sounds like fun but I could do the whole thing with just two transistors, a couple of diodes, an inductor and a few resistors and capacitors. Maybe it would be a good educational project for microcontrollers but there are far more cost effective designs. If you do this I think it would be cool to add other features such as a motion sensor so it only works when someone's there. Another idea could be a solar powered, motion activated, ultrasonic cat repeller.

While testing the solarcells, I noticed that sunlight is much more efficient powersource than artificial lighting (lightbulb, LED etc). So it could be possible to detect just sunlight apart from the street lamp.

That won't be necessary if the solar cell is also used as a light sensor.

[kodon]

While testing the solarcells, I noticed that sunlight is much more efficient powersource than artificial lighting (lightbulb, LED etc). So it could be possible to detect just sunlight apart from the street lamp.

That won't be necessary if the solar cell is also used as a light sensor.

Yes that is what I was trying to say. To use the solarcell as sunlight detection sensor aswell.

[ivan747]

hmm, makes me wonder if there is something even lower power.. something i am designing has to have a total system consumption under 1mW

You'll probably need to use a 4-bit MCU at those power levels. Something like the EM6607 only draws 3uA at 1.5V, or 4.5uW.

So that's why they exist...

[Andlier]

How about doing an RF Thermometer.

As an added feature later (in software),  add memory for historical data, say, by the hour.  Add a visible light sensor for a makeshift cloud cover sensor.  Basically, a weather station. Might be fun.

I actually did something similar, just without the radio module. I added a humidity-sensor and pressure-sensor as well though.

The board is based on the efm32 starter-kit, I just made the expansion-board with the sensors. It has pressure, humidity, light and temperature sensing capabilities. The whole setup consumes only 4uA with RTC and LCD-display with a small animation running all the time. Sensor-readings are done every 10 minutes and logged to memory.

The display alone consumes over half of the power, with display off it draws around 1.5-1.6uA, that's the average including measurements. I've estimated a 5-6 years lifetime on one coincell, it's been sitting on the shelf displaying and logging the weather for some 8 months already, still 2.97V on the battery.

The LCD shows air humidity at the left, dew-point at the right (Celcius) and temperature top right (also C).

 
 

[kodon]

Now that's pretty impressive power consumption, or lack of it

I've been thinking about that weatherstation idea. That would require a SD memorycard or similar and some sensing devices. Wind speed could be measured with 3 or 4 rotating "spoons" and optical gate which sends a pulse to the µC when sensor shaft turns.

Anyhow, I need to practise programming quite a bit so don't hold your breath I've been testing that solarcell posted above now for 3 days now. That is with two eneloop batteries and it surely seems to charge itself during the daytime (surprise or not, but it's been really lousy weather out here).

But I need to focus on the programming side of things to make progress.

Edit: inspired by Andlier I made minor tweaking and now the RTC is running with just 7µA (measured with my el cheapo DMM, I seriously need to bring more accurate one to my "office")  and I'm pretty happy with that. I put the red LED blinking to chew some extra energy so I can monitor the battery status and see how well the solarcell is doing

[Andlier]

Just for future reference; I did a full write up of my weather station project with code and pcb-files here: http://forum.energymicro.com/topic/300-efm32-microamp-weather-station/#entry830

[ptricks]

Not sure of the manufacturer but I have some solar powered outdoor lights that contain a single 4 pin device connected to the battery, the LED, and the solar panel. It uses an increase in power from the solar cell to determine day/night and control the charging of the battery. These lights were really cheap too, about $10 for 4, the part is like a to-92 but has no markings.

[NiHaoMike]

I have seen some like those as well. The problem I have with most solar lights is that they turn on too early, wasting power. The discrete ones can often be tweaked by changing some resistor values, but the 4 pin chip type is not tweakable at all. I'm surprised they went through the effort of packaging it, I would expect it to be just a blob on the board as would be typical for cheap stuff.

[SeanB]

Blob ob board for the big runs by the one OEM, the packaged IC is for the back room manufacturers who do not run a million of these through a month. They use a much cheaper thinner board and there normally only is the chip and a random value inductor on it.