This Topic is about how we acutally integrate a load cell into the hive designs (Colorado Top Bar and the Warré Design).

As exact working load cells are quite expensive the approach is to use only one sensor per Hive - taking the whole weight of it.

This is the load cell we're up to right now:

And here are some general ideas about the possibilities of how to install the load cell:

@jonathan @AaronM Any idea where to go from here?

Awesome @Ron.

This might be a silly suggestion, but imagine if we had a strip of weight sensors along the inside pocket where the top bars sit. This way you could measure the weight of each top bar individually, and create accurate readings of how much weight is honey, and how much is brood etc.

Just a thought!

My own preference would be for something that can't fall in a any way (wind/animals etc) so I'd go for placing the cell below so it takes, say, 80% of the weight. This would be very stable and involve very little hardware. Software can compensate but absolute accuracy is not in my opinion an issue anyway, it's relative weight change that matters.

I agree with the above - here in Australia we tend to have really big nectar flows from the Eucalyptus trees that then suddenly stop, so tracking relative weight gain is what we need.

If you are doing proper inspections it should be very easy to relate readings to hive conditions, so absolute accuracy is only really important for a consistent global data set.

Basically I need a warning of sudden weight gains, as a normal colony can fill an 8-frame Langstroth deep super in around a week. Makes a hell of a mess of your inside cover when they fill every scrap of space with honey laden burr comb.

I'm looking at the Flexiforce PS-03 for another project, but they might be good for total hive weight. The #100 sensor can be set up so that it measures up to #1000 by changing the power feed and resistors used to read its resistance.

Here's their FAQ:
http://www.imagesco.com/sensors/flexiforce-faq.html#14

SparkFun distributes these if you have trouble finding a cheap/local supplier.

Flexiforce PS-03

Jet - not sure the Flexiforce is good enough for hive weight, following is from their spec:

Repeatability: < ±2.5% of full scale (Conditioned sensor, 80% of full force applied)
Hysteresis: < ±4.5% of full scale (Conditioned sensor, 80% of full force applied)
Drift: < 3% / logarithmic time (Constant load - 25 lb.)

Are there requirements for weight measurement? Most of the locals I know here don't use scales, they just calculate weight by counting loaded frames or the experienced "heft" technique. (Few of them have more than 4 hives, most have 2.)

When my PS-03 arrives this week I'll put it through it's paces.

Got it a couple of days ago, just now experimenting with 2M ohm resistor. It has a nice curve from 1 to 10K then a jump up to 90K (me!) and 113K (me + test weights).

I'm wondering if the stats are prior to calibration. That is, if installed a dozen of these I'd have to come up with the right amount of resistance load to adjust each individual sensor. Alternatively, one could have four sensors (one on each axis) and take the average value as a reading.

Phidgets.com sells load cells and a usb interface board for them. They specialize in sensors and interface systems for hackers and their stuff is very easy to use. They also have a good library of sample code, should be straightforward to adapt them to an RPi.

Sounds good as their 50kg load cell is about 7$ each:
http://www.phidgets.com/products.php?category=34&product_id=3135_0

but the bridge is just too expensive to be price competetiv. We could use 4x an ADC (like the Hx711) which is about 5$ each. But running into the proble of lacking pin connections than...

My other project needs > 100K support, and a loaded Langstroth can break 130K, so I'm starting at the heavy end. The Phidgets sensors in that range are $50 + another $100 for the USB bridge + whatever gets the data into my garage. For $150 I can build a wireless Arduino with a Flexiforce, but I need to test how accurate it is during cold weather. We've got plenty of -10C days to work with this week in PGH.

I've used instrument amplifiers to act as the bridge and an Arduino to interface them , very low cost. I'll try to find the actual part numbers but I believe they are made by Dallas Semi. Attached is an illustration of how they were hooked up(for illustration purposes only!). I used this setup to measure force on the rigging of a sailboat.

The instrument amplifier I used in the above project was a Texas Instruments INA 125P, spec sheet can be found here: www.ti.com/lit/ds/symlink/ina125.pdf. The adjustable gain in a handy feature and the cost to rig up the whole system is very affordable.

Thanks, I'll check out that 125P.

The picture in the first posting comes from a prototype Marcus did for the frame hardware and me for load cell set up and Arduino sketch. See http://wbk.in-berlin.de/wp/blog/2014/04/durchbruch-bienenwaage/ (sorry only in German).

I would not count on OpAmps alone. They can lift up signals but after doing this the next barrier is the 10 bit only resolution of Arduino's ADC. You will only get a resolution of 1024th of the maximal load. I tried out an external 16 bit ADC (ADS1115) but was not really happy with the result. Use instead an ADS1231 or a HX711.