This thread is just for monitoring other projects that uses the cheap 3 wire load sensors – and fail – in the most cases.
We are talking about this around $ 9 cheap 50 kg load cells that are used in bathroom scales, e.g. the model sold by Sparfun:
The first trial to use this kind of sensor for bee scales is direcly from the founder of Sparkfun Nathan Seidle:
The Internet of Bees: Adding Sensors to Monitor Hive Health
It is interesting that he did not use the sparkfun load sensors but a hackt bathroom scale!
The results are not realy good, a long calibraion time, over days (!), and not really stable output leads to the conclusion that this model is not usable for a pofessional or even hobby grade bee scale.
Maybe one exception -that works- using that kind of cells is be the BS4.
I did not test this device myself but got a discussion with his designer and he told me that as far as your expectations are to follow the overall weight of the hive, the performances were fine…
Hi Lorenzo, I have seen this “balance de ruche (balance apicole) Beescale4 (BS4)” also some month ago as you shared the link on The Beescale4 (BS4) by Jérôme Alphonse and I wonder what kind of sensors this are. i do not thinkt this are bathroom load sensors but a kind of weight or force sensitive mat. But I don’t know. So when you have contact to the designer and they are willing to share some details it would be nice they can give some information about the sensor.
I had an interesting conversation with @iconize from the Beep project. They are also keen to develop a cheap as possible scale design.
So the guys did not rely on the bad information they found about long term weight on ultra low cost cells - as me ;-) - but did proof this information and tested different quality grade cells. Pim is confident that a new scale is doable with this kind of sensors.
I’d like to hear more about constrains. I think software based temperature compensation will be mandatory.
The design is super strong and flat. The only downside is that the film coated plywood is nicely waterproof but also expensive.
@clemens The two wooden plates for a 50x50 cm (Segeberger), or 48x38 cm (Spaarkast) scale are as expensive as €13,- including the cutting. We are also looking into injection moulding a series of 1000 pieces with a very nice design (to be continued), if these proto’s are doing their job well.
this sounds very promising. We will be very happy if you could share yet another prototype version with us - maybe the 11th one? We would really like to bring it online for the upcoming period as yet another source for data acquisition and prototype testing at the same location as our Labhive One, where we conducted our first tests. I believe @einsiedlerkrebs will also look forward to it.
Please don’t hesitate to let us know if you need some financial compensation for doing so. While we haven’t received any funding as a group yet, we will certainly find a way to support you.
@iconize That’s great to hear. I paid for the wooden plates on my first prototype with “Siebdruckplatten” much more in the local DIY store. I think I bought 21 mm one. You milled some holes in the material for the microcontroller and the cable. What thickness did you use?
[I corrected a typo: we used 21 mm plywood, not 12 mm]
@freedom2000 did a setup with this bathroom sensors also, see H frame or two bar frames? and was not really happy with fluctuation of the measured weight:
I’ve observed and posted some strange drift behavior concerning low cost half bridge load cells (used in bath room scales) in another tread and in the discussion that arose, Clemens pointed out to (t)his tread and I read that quite some work has been done and summarized here concerning the instability of low cost load cells. The general conclusion is that the drift make them unsuitable for long term logging at high resolution requirements.
Aware of load cells drift but unaware of the significance and all your warnings on this forum, I’ve purchased a set of 8# with 2x hx711 boards for about 15 Euro, and of course ended up with also quite disappointed results. To compensate for my bad purchase experience I decided to do some reading and thinking about all kind of possible causes for this drift and compensation discussions at this and other forums. After a while I was wondering if many of the drift causes are independent of the load, except for the creep due to deformation of the load cell under continues high loads. For all these load independent drifts they will be there at the same level in the the loaded situation as in the unloaded situation, which is actually the offset-correction and a mandatory procedure for load cells to make accurate measurements. And so I wonder if you can reduce the drift impact by increasing the offset-determination frequency.
I’ve spend a setup which can check the offset-drift ‘on the flight’ and some time to test this hypothesis and a bit unaware of this tread I’ve summarized the results is in another tread concerning temperature compensation, but in hindsight would better fit into this tread.
Given the long inactivity of this tread, I think many of you have chosen to move on to more ‘stable’ but also expensive single point load cells, but I think for newcomers to this issue it may be of interest to consider this option given the good results that were obtained, so I decided to add a link to my post here in this tread.
If you need more explanation or discussion about this approach, you can reach out here or at the other post.
Regards, Geert
