Bob's Sensors and Sensor Network Blog

Energy Scavenging LTC3108 OSHW boards are here

Received the first set of LTC3108 boards this weekend from pcb.laen.org  (boards are high quality, recommend highly).  Assembled them with a few challenges because I'm a novice at using DFN packages, but got the process figured out.  The photos show the results.  Initial testing shows that they perform according Linear Technologies specs (no surprise here).  In this photo input of 19.3 mV and about 2 ma is generating 3.28 V and charging up the 220 uF storage cap nicely.

The trick in this is to balance the load power/duty cycle requirements with the storage capacitor size and available input power.  More about that in the next post.  Meanwhile, here's a closer image of the board (click on the photo to enlarge).

The details (schematic, Eagle files, BOM, etc.) will be available soon under OSHW license (soon as I get github figured out!).  The board design follows closely the published LTC3108 reference designs and is fixed at 3.3V output.  There's extra pads provided to increase the storage capacitor size as required for a particular application.

LTC3108 breakout board coming soon.

I am working to put together a breakout board for the LTC3108 so you can experiment with energy harvesting yourself.  Should have prototypes together in a week or so.  Will make it available via Open Source Hardware as well as plan to make arrangements for kits.  Thinking of including options for Peltier thermoelectric generators and smallish solar cells.  Designing board so storage capacitor can be added to as needed for the application.

 

Energy scavenging with the LTC3108

Jeri Elsworth ( @jeriellsworth ) mentioned her work with energy scavenging today and that inspired me to describe a little of what I have been thinking about related to the subject.  I have been experimenting with the Linear Technology  LTC3108 in anticipation of using it to scavenge energy for wireless sensors.  The function of this chip is to convert very low level voltage sources (20mV - 400 mV) into standard 3.3V and 5V outputs.  For example, in the picture below is shown a peltier thermoelectric generator generating 33 mV (the yellow HP voltmeter on left) and the LTC3108 converting that into a usable 5V (the orange Extech voltmeter on right).

 

The peltier thermo-electric generator converts heat flow through the device into electric power (and vice versa).  In this case, a 10 ohm wirewound resistor is being powered by the lab bench supply to generate about 2.5 watts of thermal energy to heat the underside of the peltier device.  The heat flows from the warm underside to the cooler upper surface (see below) generating the 33 mV output to the LT3108. 

Of course, voltage conversion is one thing but available power is the other important consideration.  In general the available power is low too (micro-watts to milli-watts) and the conversion is less than 100% efficient so the available power at 5 V will be even less.  However, at least in wireless networks, the sensor nodes are only active a small percentage of the time.  The rest of the time, the LTC3108 can be storing energy, e.g. in a supercapacitor.   The sensor node can then pull as much as needed from storage to power the  node intermittently.  I am working to understand the balance between the ability to store energy and supply it to the sensor node (straight forward problem really but fun). 

I am planning to build a prototype board and wonder if there's enough interest in the community to build multiple of these and also supply a couple of energy sources, say a peltier device and small solar cell, in a kit for others to experiment with this technology.  Feel free to contact me on twitter @wa7iut or email if you are interested.

 

Football Sensor project moving ahead

You can find out more about the football sensor project at Battle Sport Science and the Impact Indicator on their site.

Turning Wireless into Wine

I've been working with Ron Bitner of Bitner Vineyards to install a wireless sensor network.  The purpose is to help in water management and also to help predict the onset of various pests and disease.  For example, Powdery Mildew is a fungus that is bad for grape quality.  It turns out that if you can monitor the micro climate of the grape canopy, you can predict with accuracy when the powdery mildew will occur.  The action you take then is to spray a fungicide on the grapes.  The advantage of this is that you only spray when it's really needed.  Otherwise, you have to spray prophylatically during the part of the season (wet) when powdery mildew occurs.  With the ability to focus spraying when it's only needed comes the twin advantages of reduced costs and improved sustainability (reduced chemical exposure).

We decided to go with the eKo system from Memsic (which Memsic aquired from Crossbow).  Shown here (above) is one of the nodes in the wireless mesh network.  This node is a complete weather station from Davis Instruments.  The weather station is powered by the Memsic eKo wireless sensor node (the yellow weather proof box below).   This node has a approx 2"x2" solar cell/battery and battery management system that powers both the weather station and the wireless node.  Up to four sensor systems (which can be either single sensors like soil moisture or multiple sensors like the Davis weather station) can be attached to each node.

This is a closer up view that shows the weather proof enclosure of the eKo wireless node.  The top half of the enclosure is clear plastic that enables the solar cell to be powered.  Note that the upper half encloses the 2.4 GHz antenna.   It's a very neat design.  We may move the eKo node up higher depending on how the grape canopy affects the signal strength when it is in full leaf. 

All the data is fed through a gateway node to a computer for display of the data.  (more on that later)

The mesh layout is on 8 acres and looks like this:

 

One of the nice things about this project had been the gourmet lunch prepared by Mary, Ron's wife, together with a glass of Bitner Vineyards 2007 Merlot!  Yum!

 

Workbench LED Lighting System

I found some LED lights at Home Depot recently.  They're Phillips AmbientLED MR16 Warm White.  They output 130 Lumens each and have a color temperature of 3000 K (warm white).

I got an old bench light designed for fluorescent bulbs from the Reuseum.  I pulled out the bulbs and starter circuit.  Then I  got out my Dremel tool and went after the bezel that came with the light.   To hold the LED lights in place, I made some 0.050" Al plates with a center hole sized so air can flow around the LED light.

Here's the back side.   I soldered 18 gage wire directly to the pins.  The MTBF of these bulbs is estimated to be 15 years so it seemed easier to solder than to find an alternate way to connect power.

The four lights were wired in parallel for 12 V at 1.2 amps

Lights installed

 

Lights up the bench, note 1.17 A at 12.1 V on the power supply powering the lamp.  (I'll probably use a wall wart supply when this is finally completed.

I prefer the warm with color to the blue white light higher color temp lighting like typical fluorescent lighting and no flicker and little heat to deal with.  The total output is 520 lumens, which is about the same as a 40 W incandescent bulb, using only 12 W.  

 

Workbench

The guys over at The Amp Hour wanted see our workbenches so here's mine.   It's been my bench since I first started at HP in 1979.  I use it now mostly for embedded system design and testing.   The computer on the bench runs various IDE's for development.   I also use it for rf projects.  Sometimes I do embedded rf projects since I have been focusing on wireless sensor networks applications for agriculture.    Pasha (the dog) is my assistant on various projects and also my athletic trainer in the foothills of Boise.   Note: click on photo to enlarge view.

The audio signal generator between the books and scope is a HP-300C.  It is a version of the original HP design developed by Bill Hewlett as part of his Masters work at Stanford.   It is autographed by both Bill Hewlett and Dave Packard, thus the place of honor on my bench.

Here's a closeup. 

Here's another view

 

 

 

Finding the "sweet spot" in node placement

How can one find the “sweet spot” in node placement for the best signal (and presumably the best reliability) when setting up a Wireless Sensor Network?    

This paper (here) shows that there is typically not such a thing as optimum placement in a 900MHz/2.4GHz setup because the environment is dynamic.   That is, the RF environment changes through the day as different pieces of equipment such as other wireless devices or microwave ovens or motors are operated.   Also, the RF wave interference patterns (destructive or constructive interference) I mentioned change rapidly as people and equipment move around the office or factory floor.    A Wireless Sensor Network (WSN) based on the WirelessHART standard (pioneered by DustNetworks) overcomes these problems through a combination of frequency hopping (to avoid interference) and path diversity (ability to change the mesh pathways on the fly).   

How WirelessHART Provides Reliability (Four 9's) Similar to WiredHART

 

"Molasses, snails and glaciers: none are slower than an organization developing a new wireless standard. "

"Molasses, snails and glaciers: none are slower than an organization developing a new wireless standard. "   New round of low energy devices on the way real soon now.

http://www.engadget.com/2010/04/21/bluetooth-4-0-with-low-energy-almost-finally-ready-to-roll/