Just a quick update: the streaming weight and temperature data is back online. Apparently, I just needed to reset the ConnectPort X2 I’m using as a portal between XBee radios and the internet. It’s annoying that it dropped out in the first place, and I’ll see if this method of data collection is really robust enough. For future systems, I will likely need to use an SD card as hard backup, something I skipped for the prototype mainly to avoid the hassle of dealing with another library and more hardware (not really a good excuse).
All along, I’ve assumed that my postal scale would be fully temperature compensated — I assumed it would be at least partly done in software, but it seems that the standards for postal scales are significantly lower than I had assumed. They’re probably published somewhere, but a measurement is probably better than any spec.
I was confused about what I was seeing in the raw data, and during 4 days where it was too cold for the bees to go outside (the first 5 days in the graph) I used the data during 20 degree C daily temperature swings to run a regression analysis to statistically estimate the temperature sensitivity of the load cells.
Here’s the raw data:
And here I adjusted for temperature. You can clearly see the 5 days where the bees just consumed honey and then the two days where they brought in nectar and pollen, increasing the hive weight during the day (and continuing consumption at night). According to the regression analysis, the new 3lb package of bees consume around half a pound of honey a day and the postal scale signal varies by 0.0331 pounds per degree C.
Note that a regression analysis is a horrible way to do accurate compensation. For one thing, I’m assuming that the bees really were inside for the whole 5 days and that their consumption of honey was constant even when temperature varied (almost certainly false). I’m sure the bees consume faster in the cold than at warmer temperatures while they’re keeping their brood warm.
Tomorrow, I’ll see if I can figure out why the feed has dropped offline and I’ll take the hive off the scale and put on a fixed weight. The fixed weight will allow me to determine an accurate temperature compensation without convoluting factors.
This first picture shows shows the inside of the postal scale (Adams CPWPlus200). I designed one of my first circuit boards (with lots of mistakes which is why it’s a total mess to look at) to hold the power regulators that convert from 12V from a marine battery to 3.3V for the Arduino Fio and temp sensors as well as 9V for the scale. You can see I’ve run a U.FL connector from the XBee radio (bottom right) to an external duck antenna on the right to get the signal out of the metal box.
This next picture shows the scale on a hive bottom board. I had one available and they’re about the right size and shape so I just used two bottom boards with the scale sandwiched between them. Here you can see the scale readout. I ran a wire from the Arduino to the power on switch so I could turn the scale on and off remotely.
This next picture shows all the wires running from the bee hive to a second box housing the battery, a charge controller (for when I get a solar panel mounted out there) and the scale’s readout. The bees are still a bit confused that I raised their entrance a couple inches by adding an extra bottom board and postal scale!
Finally, here’s a wide view of the 3 hives at the apiary (and the battery “hive” box).
I’ve got lots more to add (including some really great processed weight data) but I’ve been “forced” to take advantage of the good weather to build a garden before I miss planting season! I’ll get some pretty graphs up tomorrow, and hopefully on Saturday, I can go out to the apiary and figure out why I’m no longer getting live data from the hives (since 8:40 this morning).
I’ve got first data from the bee hive streaming to Thingspeak.com here: https://thingspeak.com/channels/3004
My initial installation of the ConnectPort X2 isn’t ideal so the data is dropping out periodically (I’ll be adding a repeater to get through 4 walls) but it’s amazing that the system is working!
I’ll also post more details and pictures when I have time, but I wanted to throw this out there to share my celebration.
On Saturday, I opened up the hives for a one-week inspection to make sure the queens are there and laying eggs properly. Both hives have around 4 frames of bees that were active and calm in the 70 degree weather. I found each hive had 2-3 frames with a great pattern of eggs and brood. This picture is from after I smoked the bees a bit, so it looks a bit sparser than before they got smoked.
Once the population gets much higher, it gets really difficult to find the queen, so I took a quick video to show the queen and some tiny day-old larvae! I really need an assistant or a tripod if I want to get into taking videos regularly, but YouTube has a stabilization feature that I think really helped make the video easier to watch. It causes the camera to move around unnaturally, and it looks really weird when I get close up to the cells with brood, but the alternative is looking at a super shaky video when I’ve got a frame in one hand and my phone in the other.
It’s not critical to actually find the queen in each hive (I didn’t see her in the other hive) as long as you can find eggs or brood, but it’s always exciting to find that one egg-laying queen bee!
In other news, I was struggling to rewire the bee counter test board, and after destroying two of the sensors, I gave up and ordered a revised version with (hopefully) the correct pinout now. It was sent to the fab today, so it should arrive in a week or so.
I’ve also got plans to install the hive scale and temperature sensors next weekend. The system is ready except for mounting of the solar panel. The battery can last more than a week without recharging, so I’ll be able to test it even if it’s not permanently installed.
On Sunday, April 21, I got my two 2lb. packages. My wife was kind enough to allow me to keep them in the kitchen even though they buzzed louder and louder as they warmed up in the house. I was waiting until the evening to install the bees as they’re less likely to try to find a new home if it’s starting to get dark when they’re installed.
My son thought the bees were pretty neat once he understood that they couldn’t get out of the box to sting him.
I installed the bees into two medium boxes each. I also fed them a 1:1 sugar syrup to try to stimulate brood rearing. They have plenty of capped honey from last year so I didn’t bother to cook up 2:1 sugar syrup (making it easier for the bees to get the sugar).
I forgot to add my pollen patty, but they have a few frames of pollen in each hive so it’s probably unnecessary. I’ll add a pollen patty when I check for brood in a week to make sure the queen is laying.
I also managed to mark my queens for the first time this year! I released the queen into the queen marking tube (seen below) and gently pushed her up against the screen with the foam plunger until she stopped moving around. Then I realized that I’d forgotten to open the paint marker! I had to work for a while to get the shrink-wrapping off the pen at which point I realized the paint wouldn’t flow even after shaking the pen vigorously and trying to follow the little pictographic instructions on the side. Of course, the entire time, I was trying to keep the queen warm in the marking tube without accidentally dropping her!
Finally, I sucked on the end of the pen through my veil, and while I got a mouthful of paint (which I spit promptly back onto the veil), the paint was flowing and I dabbed a conservative dot on the queen’s thorax. She was still moving around vigorously, and the temperature was above 40F, so I don’t think my learning experience caused any harm.
Here’s a picture of the first marked queen with the rest of her hive in the background. I’m not quite comfortable yet beekeeping with bare hands as I tend to flinch every time a bee lands on my fingers, so I found the tight gloves to be really helpful. They might stop a light sting, but it was too chilly for the bees to be trying to sting me anyway so I’m not sure any bees actually tried.
I gave each hive two frames of warm (room temperature) capped honey that I uncapped for them, and a gallon of 1:1 sugar water to help stimulate egg laying activity. Just as I was finishing up, it started raining pretty strongly, so I didn’t get a chance to take more pictures, but the hives now have two boxes of drawn comb (with about 8 frames of honey each) and an empty box for feeding on top.
I’m getting close to finishing my hive scale. I could throw it outside now, but I’ve got four weeks until the Maker Faire deadline, and I’m still working on adding transmission to stathat as well as thingspeak. I’d also like to track down why my sensors are reporting 85C sometimes (I suspect they’re resetting and need slightly more power). I have strong plans to install the system two weekends from now at the Acreage.
The latest bonding bill in the MN House includes support for honey bee research at the U of M. We badly need this research to help slow the decline of honey bees, so if you live in MN, consider contacting your state legislators supporting funding for the MN Bee Lab in this bonding bill.
This week, I cleaned out the two dead hives at the Lair and got them ready for the package bees that are scheduled to arrive April 20. The weaker (right) hive hadn’t touched the bee candy, but the other hive had started on it and was only about an inch away from the life-saving sugar when they died, apparently of starvation one very cold night when they couldn’t move enough to get at the sugar!
Here you can see the cluster dead just an inch away from the sugar (on top of this frame). The other hive died both an inch below my added sugar AND an inch above honey on the same frame.
There was also a little mouse that moved in under the hive. I tried to get a video of the mouse running out when I cleaned out its extensive nest, but either it had left long before I got there or it hid down in a corner somewhere. It did some minor damage to the hive, but it only touched 3 frames. I’m glad I closed up the hive before he had a chance to do more damage!
I wonder if part of the problem is that I’ve left the hives too big over the winter. Air space shouldn’t have a large effect given that the bees only keep their small cluster warm, but it’s possible that it gets more drafty or slightly colder with a larger hive. I’ve cut the hives down to three boxes with about one full of honey each. I’ll be back on bee installation day (still tentatively April 20).
There’s a lot of sensationalist reporting (or making stuff up) out there on why bee hives are dying off. A couple days ago, NBC put out a really good summary of some of the factors that have contributed to hive death recently.
To expand on the article, the neurotoxic neonicotinoids that cover planted corn and canola seeds (among others) has been shown in a couple papers early this year to block bee’s memory and ability to learn, for example learning where nectar-producing flowers can be found or how to get back to their hives. This is suspiciously similar to colony collapse disorder where bees just disappear over the course of a few days and never come back.
At the same time, I have seen really minimal evidence that neonicotinoids are actually present in hives that suffer from colony collapse disorder. Samples from afflicted hives over recent years haven’t yielded very strong patterns as one might expect if the problem was bees collecting contaminated pollen from corn or flowers near treated crops.
With the dozen or so diseases and complex interactions between the dozens of insecticides and other chemicals we routinely spray on crops, I think we really need much better funded research to facilitate the time-consuming collection of data showing when bees are being exposed to what chemicals and which combinations correlate most strongly with colony death.
Since Bayer’s last line of defense against the demonstration of really bad sub-lethal effects of neonicotinoids on bees is to simply claim that nobody’s shown that bee hives ARE being contaminated (so lab demonstration of really bad sub-lethal effects might not extend to actual hives), they’re certainly not going to spend millions of dollars looking to see if one of their most profitable products is ending up in nearby bee hives! It’s really up to publicly funded researchers to make the case, and as budgets get slashed, it could be another decade before we really know what’s going on.
I’ve sent the next bee counter test board from OSHPark. The previous version didn’t work as soon as I powered it up, and as I discussed the details with Clemens, we came up with a much simpler version that simply turns on two sensors, reads both, and turns them off again directly from the microcontroller rather than pulsing the sensors and latching signals.