I'm still working on some configuration, but I've started posting so if all four of you still want to follow my blog, you might want to head over there.
Friday, October 12, 2012
I've moved to hackerbee.com!
I've gotten serious about building the bee hive datalogger and I've never been particularly fond of the name here, so I've moved to hackerbee.com
I'm still working on some configuration, but I've started posting so if all four of you still want to follow my blog, you might want to head over there.
I'm still working on some configuration, but I've started posting so if all four of you still want to follow my blog, you might want to head over there.
Wednesday, October 3, 2012
Some awesome temperature logging projects!
I found some great temperature logging projects this week -- one linked on hackaday.com, (http://www.maxjusticz.com/honey-bee-tracking-box/ ) and another linked in the comments of the first (http://mietz.imkerforum.de/the_idea.html ). This is great stuff -- the first is incredibly finished with a waterproof seal and a solar panel directly on top of the enclosure while the second has an incredible 88 sensors spread out over 11 frames of a single-box observation hive! Here's the incredible video of temperature within the hive:
This is incredible data to see as it suggests that the lower resolution 9 points between each box I'd planned is enough to catch most of what's going on in a bee hive. Sure, I might miss the occasional deviation, but if the bees reliably cluster around the brood nest as they seem to in inspections and as evidenced here, adding 10x more sensors is probably not going to give me much more information.
Heck, even 9 sensors per box is probably way too much information, but I'm hoping to track the hive cluster over winter next winter (the winter after next summer) and I'm pretty sure one sensor per box won't be enough.
Anyway, great stuff makes me want to get mine working even more! I suppose that means I hurry up and draw a final schematic so I can solder down all the components without making time-consuming mistakes!
Heck, even 9 sensors per box is probably way too much information, but I'm hoping to track the hive cluster over winter next winter (the winter after next summer) and I'm pretty sure one sensor per box won't be enough.
Anyway, great stuff makes me want to get mine working even more! I suppose that means I hurry up and draw a final schematic so I can solder down all the components without making time-consuming mistakes!
Tuesday, October 2, 2012
Meeting about bee hive monitoring
I met a guy from Portland yesterday (he goes by "Hydronics" online) to talk about data logging in bee hives. We talked about his great work building a device that counts bees going in and out of the hive.
This is a long summary of our projects and some of our ideas -- sorry if it's not interesting (this is your obligatory "wonkish" warning), but if you have interest in bees, electronics, or measuring bee hive data, I welcome any advice, criticism or comments!
The device uses a microcontroller to poll 44 little IR LED proximity sensors on 22 different gates (or tunnels) and measures whether they're going in or out of the hive by watching to see which of two sensors is triggered first through a gate. He saw some amazing things in the data -- like some spikes at 4:00 PM on a few different days that looked like orienting and some in/out spikes that could possibly correlate with a particular food or water source if they're statistically significant (i.e. not just us finding patterns in random noise).
One of the major issues I've had with his design is that it's power hungry. It draws 75 mA which is a lot of power when you're trying to run the thing off a solar panel and battery! We talked about a bunch of ways to improve the design. He's already tried turning off half the LEDs and using just one LED to drive two sensors in each gate. I think he might have to modify the board layout to put the active LED roughly equally spaced between the two sensors (one is further from the LED in his current design). This would cut the power consumption roughly in half. He could also pulse the LEDs -- turning on only the ones being measured at one time. I showed him some spec sheets of IR proximity sensors designed for use in smartphones and ebooks. They are designed for ultra-low power consumption and can run on a few dozen microamps when pulsing at 10 Hz or so. Of course, the integrated packages are $4 each, but we could cut the price significantly by engineering a similar pulsed solution that meets our needs. As always, you can spend time or money, but it's hard to get exactly what you want for cheap in a ready-made package!
Hydronics also mentioned that he thought he might be able to get the system working with ambient light -- possibly using UV or IR (or both?) so it works even when the sky's cloudy. Heck, with a cheap diffuser, you might be able to use any wavelength you wanted! This would eliminate almost all the current draw, but it might sacrifice sensitivity. The system would likely fail under certain lighting conditions like when the sun is shining directly into the hive entrance or when the sky is really cloudy. Still, paired with external ambient light sensors and using a diffused slit and two more light sensors in each black-painted gate, I think there's a decent chance you could calibrate the system to work in most lighting conditions! I'm excited to look into engineering and testing LED-free gates!
Ultimately, while 100% detection will always be a goal, the kind of data we're looking for can tolerate significant noise or downtime.
On a separate subject, I mentioned some research I've come across that tracked bees in and out of the hive with radar or with cameras. There have even been some attempts to glue radio trackers to individual bees, although this is obviously tedious! I wondered if we could put small dots of retroreflective material (3M makes some great beaded and corner-cube tapes) that would give incredible contrast in all lighting conditions for tracking with a camera. This might be further in the future (certainly after the winter when the bees rarely fly outside of the hive) but it could allow wide-angle tracking of tagged bees over a very large area with a relatively low-tech video camera! Tiny bits of tape could be pre-cut for tagging, and while it wouldn't be quite as effective, I also wonder if we could simply sprinkle glass beads onto some bees or develop an adhesive including glass beads that could provide enough retro-reflection to allow tracking. Unfortunately, I think the bead idea may require metal-coating of one side of the beads, but maybe all the beads could be half coated ahead of time and enough would be oriented the right way in the adhesive that they'd provide a good signal?
Finally, we talked about my hive logging project. My specs are always changing, but currently I am planning to put an array of 9 temperature sensors in a thin frame on top of each box for 3D mapping of temperature in the hive as well as monitoring ambient temperature and my circuit temperature (for debugging and tracking). I'll probably also add a humidity sensor, although I've never seen anybody come up with particularly useful correlations between humidity and bee behavior. I'll be installing a microcontroller inside an industrial postal scale and streaming the data back to the internet through an XBee radio and a ConnectPort X2. We've both also been looking at logging to an SD card with a real time clock for timestamps. If my radio logging isn't reliable enough, I might fall back on an SD card, but it's a bit of a hassle to get into the hive to retrieve the data!
Hopefully I'll be able to build my own bee counter for next year's season as well. Who knows how effective it'll be at first, and it's much lower priority than getting the weight and temperatures, but if we can build a good remote solution, it could be an incredible addition to a bee hive monitoring package!
This is a long summary of our projects and some of our ideas -- sorry if it's not interesting (this is your obligatory "wonkish" warning), but if you have interest in bees, electronics, or measuring bee hive data, I welcome any advice, criticism or comments!
The device uses a microcontroller to poll 44 little IR LED proximity sensors on 22 different gates (or tunnels) and measures whether they're going in or out of the hive by watching to see which of two sensors is triggered first through a gate. He saw some amazing things in the data -- like some spikes at 4:00 PM on a few different days that looked like orienting and some in/out spikes that could possibly correlate with a particular food or water source if they're statistically significant (i.e. not just us finding patterns in random noise).
One of the major issues I've had with his design is that it's power hungry. It draws 75 mA which is a lot of power when you're trying to run the thing off a solar panel and battery! We talked about a bunch of ways to improve the design. He's already tried turning off half the LEDs and using just one LED to drive two sensors in each gate. I think he might have to modify the board layout to put the active LED roughly equally spaced between the two sensors (one is further from the LED in his current design). This would cut the power consumption roughly in half. He could also pulse the LEDs -- turning on only the ones being measured at one time. I showed him some spec sheets of IR proximity sensors designed for use in smartphones and ebooks. They are designed for ultra-low power consumption and can run on a few dozen microamps when pulsing at 10 Hz or so. Of course, the integrated packages are $4 each, but we could cut the price significantly by engineering a similar pulsed solution that meets our needs. As always, you can spend time or money, but it's hard to get exactly what you want for cheap in a ready-made package!
Hydronics also mentioned that he thought he might be able to get the system working with ambient light -- possibly using UV or IR (or both?) so it works even when the sky's cloudy. Heck, with a cheap diffuser, you might be able to use any wavelength you wanted! This would eliminate almost all the current draw, but it might sacrifice sensitivity. The system would likely fail under certain lighting conditions like when the sun is shining directly into the hive entrance or when the sky is really cloudy. Still, paired with external ambient light sensors and using a diffused slit and two more light sensors in each black-painted gate, I think there's a decent chance you could calibrate the system to work in most lighting conditions! I'm excited to look into engineering and testing LED-free gates!
Ultimately, while 100% detection will always be a goal, the kind of data we're looking for can tolerate significant noise or downtime.
On a separate subject, I mentioned some research I've come across that tracked bees in and out of the hive with radar or with cameras. There have even been some attempts to glue radio trackers to individual bees, although this is obviously tedious! I wondered if we could put small dots of retroreflective material (3M makes some great beaded and corner-cube tapes) that would give incredible contrast in all lighting conditions for tracking with a camera. This might be further in the future (certainly after the winter when the bees rarely fly outside of the hive) but it could allow wide-angle tracking of tagged bees over a very large area with a relatively low-tech video camera! Tiny bits of tape could be pre-cut for tagging, and while it wouldn't be quite as effective, I also wonder if we could simply sprinkle glass beads onto some bees or develop an adhesive including glass beads that could provide enough retro-reflection to allow tracking. Unfortunately, I think the bead idea may require metal-coating of one side of the beads, but maybe all the beads could be half coated ahead of time and enough would be oriented the right way in the adhesive that they'd provide a good signal?
Finally, we talked about my hive logging project. My specs are always changing, but currently I am planning to put an array of 9 temperature sensors in a thin frame on top of each box for 3D mapping of temperature in the hive as well as monitoring ambient temperature and my circuit temperature (for debugging and tracking). I'll probably also add a humidity sensor, although I've never seen anybody come up with particularly useful correlations between humidity and bee behavior. I'll be installing a microcontroller inside an industrial postal scale and streaming the data back to the internet through an XBee radio and a ConnectPort X2. We've both also been looking at logging to an SD card with a real time clock for timestamps. If my radio logging isn't reliable enough, I might fall back on an SD card, but it's a bit of a hassle to get into the hive to retrieve the data!
Hopefully I'll be able to build my own bee counter for next year's season as well. Who knows how effective it'll be at first, and it's much lower priority than getting the weight and temperatures, but if we can build a good remote solution, it could be an incredible addition to a bee hive monitoring package!
Subscribe to:
Posts (Atom)