Experience with different load cells

Hardwareentwicklung / hardware development Scale design
1

Hi

I’m developing a beehive scale.
I’d like to share with you my experience until now.

I tried to use bathroom load cells with a load cell amp Chinese bought on Amazon with a bad result.

After that, I bought a SparkFun load cell amp and, the result was not good again.

I read many posts in this community and it’s clear that you have a good experience with scales and load cells :D

Now, I’m facing with load cell bar to buy for my project and I’d like your suggestion about it.

Based on what I read I found interesting Omega E120A (thanks to Clemens).

Could you share your opinion?

2

Hi @marzianoumano welcome at the hiveeyes community! You may have read the thread Scales with bathroom load sensors so your experience is one more unsuccessful trial in a longer list. I still wonder why this type of scale is still sold e.g. from beehivemonitoring.com but they point out

Accuracy of weight: +/- 0,25kg

I fear it is much more and not a variance only, but also sudden shifts as I have seen on some charts on scales of this type.

3

HI @clemens

I agree with you.

For that reason, I tried to post here my experience to understand what you suggest after all tests you have done.

What do you think about omega e120a? Bosche?

I’d like to buy a good load cell without problems with temperature and other stuff…

Thanks in advance

4

For a workshop I had initially planed to order the 100 kg version of the Mavin NA1 via Maritex / Poland, see also Welche Wägezelle für low-cost-Waagen? - #10 by clemens. I had ordered some test pieces before, I did not test the load cells extensive but they seemd to be ok.

Maritex could not deliver 40 pc. timely, so I searched for alternatives and found the one from Omega. In the end I had not time to check the cells against temperature stability before and I ordered the needed amount for the OpenHive ESP32 / TTGO T-Call SIM800C workshop.

We built around a dozen scales and I asked the beekeeper to put the scales outside with some constant weight. So you can see the scales (arround half is online) by choosing a scale in the drop down and comparing weight and temperature. Grafana dashboard – documenta: Stockübersicht & Bienenwetter

Beginning of the measuring periode you may find hives with a constant weight. In the meantime some beekeeper had put the scale under the hive and the temperature sensor in the hive so we can get no comparison data from this hive.

I konw that Hive 19 had a constant weight from 2021-10-11 11:00 to 2021-10-30 11:30. Changes: 2021-10-17, 11:10 new BME sensor, from 2021-10-31 on scale under a hive. So it may be worth to have a look at the weight and temperature correlation in this dashboard.

Looks like a variability of +/- 400 g before the weight peak on 17th (sensor replacement) and +/- 200 g after. For my feeling a bit too much.

5

Thanks, @clemens

I read your late post very carefully. The data reported shows any critical issue for our field application and for that reason I’ll exclude buying Omega.

Always here in the community, I found an interesting topic about BOSCHE H40A

In this second case, the tests tell us that you cannot trust BOSCHE H40A (other my option).

What could I buy to have a good result?
How do commercial scales resolve this kind of issue?

6

I would not say closing that the Omega E120A is not usable. We have used the load cell in a primary not intended setting. So it is a single point load cell that “expect” a single load on top, fixed only on one point and apart from that “floating”. We use this single point cell in a two point setting and so forces from other directions than top to bottom can occur. So my conclusion on this point of time would be: Checking the load cell in a single point setting and see if the temperature effect is still there and maybe then trying to minimize this side forces, e.g. by a thick rubber layer and check temperature effects then in a two point setting again.

I did a lot of tests and collected much data with the Bosche H30 for this setting: Open Hive "Scale Bar" for Single Side Weighting and I evaluated the load cell – or at least the pieces I got as reliable and less temperature related by using this cell in a single point setting. But using this load cell in a double setting, in this case it was a scale on the side of a bee hive and on the other side a piece of wood, leads to unexpected steps in the weight graph. It was not the picture of a 1:1 temperature related effect more a sudden shift in the graph. So It may be related to suddenly moving material triggered by heat influence or you have two independent effects, side forces and temperature.

This may be a nonsatisfying answer for you but it is a more “scientific”, we have to do more research or “it belongs” … :-)

Even if we judge the Bosche H30 as a good load cell 50 EUR plus tax is far too much for a low budget DIY project, especially in a setting with two load cells. So Bosche is imo no option for a really cheap scale. But perhaps you may tell us more about your plans and how related a cheap price is to that.

Btw. it was the H30, see Strain gauge load cell tests in a DIY climate chamber - #10 by clemens, that was tested in the climate chamber not the H40 and Martin / @zmaier found good pieces of the H30 model and not so temperatur stable exemplars. So even different cells from the same model may vary.

You are asking about the commercial scales. Years ago you could not find any scale with tow load cells, as far as I remember Wolff Waagen was the first commercial scale with two load cells. All other used single point settings. But in the meantime you have our colleague from https://easyhive.org/ or https://www.beehivemonitoring.com. The last one has scales with a quite broad variance for the bathroom cell we should not discuss ;-) but also the XS with +/- 10 g! But I would not count on it. Perhaps you get weight scores in 10 g steps but I fear temperatur effect is not extraordinary better than we have seen. I think the luck of the commercial scale is that nobody put a static load on it over a long time with changing temperature. There will be always a bee hive / rain / snow / a beekeeper and so the temperatur effect is not obvious.

The cheap load cells we are using are produced for a grocer’s shop setup. You unload a scale tare it, put apples on it, measure the weight and so on. There is no load over time and without a tare procedure intended. So I fear we have to search for load cells that fit our needs by chance.

7

Ok, it’s clear what you say.

At this moment I filtered my list with these load cells:

  • Omega E120A: following your suggestions it could be a good experiment. (I read the specification and it works with 9v but ESP32 can pilot until 5v. Could it be a problem?)
  • BOSCHE H30: If it worked well I would buy it but as you reported it’s not sure. So it’s not at the top of my list.
  • BM11: I request a quote about it. Reading the datasheet could be a good option.
  • Mavin NA151: it could be a good experiment.

Could you suggest anything else?

Honestly, I’m trying to do a reliable solution with a limit of 100€ (micro, load cell, sim, etc…) it could fit my needs.

I exactly agree with you.

Reading many topics, articles and talking with different persons (I was at an exhibition in Italy a month ago to scout it) I’ve arrived at the same conclusion.

These days I thought to contact someone that sells a commercial scale to test one with a fixed weight. Then I could buy it if it will report the correct value for at least a week (but as you said I’m not so confident that commercial-scale works well).

In that case, I’ll go to x-rays the scale bought to understand all the secrets.

8

I fear that the Zemic BM11 with “stainless steel, hermetically sealed” will blow the budget of 100 EUR by far. I would expect the Zemic alone will cost 100 EUR or more.

Have a closer look at the NA151 you have linked. On this load cell you have only one big screw thread for a single screw feet per load cell. So you have to use 4 load cells for this setup, see Comparison of Different Load Cells for a 4-Point Scale and Füße für Waage mit 4 Wägezellen - #15 by clemens. For a setting with two load cells you have to use cells with 2 threads each so you can fix a L- or U-profile on the top and bottom side. 200 kg is also a bit high. It is better to have a oversized scale than a too week one, but with a 100 kg load cell x2 you have still 200 kg it can measure and in the most cases you have 150 % “limit load” and 300 % “breaking load” so you are still save with 100 kg Emax!

That’s nice because this is still my goal for a DIY scale. :-) For my last trial I could not make it, see OpenHive ESP32 / TTGO T-Call SIM800C the budget for this was 170 EUR.

A lot of belongs to your scale design. especially the mechanical part will eat up some money. It is also difficult to switch from one to an other load cell if the form factor is changing. So I would first decide / fix a scale design (1 load cell, 2 load cells or more), then fix the form factor and then choose a load cell. My current favourite design is the 2 bars variant as descibed on Waage OpenHive 2bars [für T-Call] I found for this load cell form factor also many cells, see Waage-Konstruktion mit 2 Wägebalken und 130 x 30 x 22 mm Wägezellen but I did not keep up with a systematic test for this load cells. So in case you are also interested in this it would be a welcome support and motivation to go on in this direction. On (or under :-) my desk are the 130x30x22 cells from Omega, Marvin (as I remember both 100 kg) and a pair of CZL601 but only 50 kg so it could be a gola to do some systematic tests till Christmas.

Btw will it be an open source project you are planig or a commercial one?

9

I give you a quick response.

My project is not commercial and if it will work as open source it’s OK.

My scope is build a beehive scale to be used by my wife and her brother in their daily job (beekeepers)

So I’m very glad to contribute on your project.

This evening/night I will read very carefully what you wrote and I’ll give you feedback as soon as possible.

10

As promised yesterday, I have deeply read what you wrote.

The Zemic’s sales contacted me and we planned tomorrow a brief call (when they contact by phone the price is high :smiley: )

I’m a practioner and you are the expert. :grinning:
You have more experience than me about it. I take into account your suggestions.

I admit to having not considered the mechanical part in my budget. I have always thought the most difficult part of the project was the electronic side, whereas, once a time I would have finished load cell, micro, etc… then I will go ahead with the mechanical part. Reading your words this approach it’s not the correct strategy and I’m available to change it.

My goal is to find a stable solution to use next season and I’m available to invest time and not much money to reach it.

Let me know how you want to proceed and If I could help you in some way.

11

A systematic check of different load cells regarding temperature would mean to build scales with one “scale bar” (not two), put a constant load on it, add a temperature sensor and set it up outside exposed to the current day and night temperatures. Log all data and compare weight and temperature over time.

12

Here is a comparison of three “cheap” load cell types (unfortunately in german), but maybe this helps you:

2 Likes
13

Thank you so much @mrtn. :+1:
Google Translate allows us to read a post written in German too. :smiley:

Unfortunately, CZL 601 is available only in China or India and the shipping time is at least 1 month. Before buying it I explain my technical doubt about CZL 601. :thinking:

Based on your experience, the fact CZL 601 works at 9 volts whereas micro (such as ESP32) works at 3.3 volts could represent an issue. What is your opinion about it?

In this case, I would put a load cell amplifier (I have a Sparkfun load cell amp) between micro and load cell.

14

Btw. there is a globe symbol under each German post you can click on:

2021-12-03 16_06_25-Greenshot
2021-12-03 16_06_25-Greenshot733×294 15.5 KB

… then some magic is happening …

2021-12-03 16_06_40-Greenshot
2021-12-03 16_06_40-Greenshot1005×453 46.8 KB

1 Like
15

3.3 V vs. 5 v or x V is not a problem for the load cell. The load cell is more or less a resistor and will work with 3.3 V also.

But some breakout boards for load cells are working on the edge of the specification or below in case there are designed for 5 V systems, see Stromversorgung HX711 - #16 by Andreas

You will always have to use a load cell amp, but this is not for the excitation voltage but amplifies the signal from the load cell to the microcontroller. The load cell signal is in the mV range, so a normal Arduino / ESP can not handle such a small signal with the built-in ADC. You need an external PGA and the load cell ICs also have an ADC with higher resolution.

2 Likes
16

Hello everyone :slight_smile:

I’m an employee of Maritex. And I’m the PM responsible for the load cells. To follow the topics on this forum, I registered to answer some questions and comments you add here.

Unfortunately, as you know, there are very big problems with the availability of electronic components all over the world. This problem also applies to load cells. MAVIN - of which we are a distributor, since this year produces only ordering products, has very small quantities available immediately from the warehouse.

That’s why we have problems with some models (mentioned by @clemens - model NA1-100kg). Load cells that are not very popular, MAVIN produces only under order and requires a large amount to start production. Therefore, we are having availability issues, for which I apologize. Please contact me on PM, if you are interested in any model, I will try to help.

We have a large beekeeping market in Poland, so bee scales are a popular projects. My experience and purchases of our clients show that they use different models: NA2-100kg, NA3-150kgml, NA4-200kg. Also NA151-60kg- 4 pcs. in one scale. NA151 have recently been very popular in our clients’ projects. If you have any questions, please let me know and I’ll try to help.

As compensation for accessibility problems, I can offer a -10% discount code for our online store. If anyone would like such a code, please contact me on PM - I’m not sure if it complies with the regulations, so I don’t want to paste it here. If admin agrees - I will add the code in this message :slight_smile: )

We also offer this manufacturer - SEMTECH.

there will be no problem with the availability of CZL601. I will add it to the store soon.

If you have any questions or comments, I’m at your disposal.