Weight inside vs outside

12 posts / 0 new
Last post
DavidM
Weight inside vs outside

Pretty interesting experiment to see the difference between the load cells sitting inside and capturing a graph vs sitting outside ..

I took one of the new units B-06 and it sits on my desk, I did remove almost one pound just to see a response.... the second unit B-04 i moved outside into the sunlight and the weight swing was interesting to see. The mass of the second unit has not changed... the unit started out approx 49.8 and drops to 49.2 ....

would be interesting to understand what thermal change causes this deflection of the strain gauge.... Heat vs sunlight hitting the sensor legs....

I've seen folks talking about an algorithm to flatten out the thermal swing of the readings....... Anyone working on that?

Image: 
unit sitting indoor with constant temperature
Unit sitting outdoors the weight fluctuation is almost 1 pound
Paul
Paul's picture
What material is the weight?

These swings are a little bigger than I like to see (+/- .1 or .2 lbs), but not too bad.

What material are you using for the weight? Iron or steel, or something that absorbs moisture? When I do temperature testing, I don't even use a wooden board to stack the weights (barbells) on,, but put some rebar between the rails and place the weights on them.

If you have them handy, try putting supers filled with drawn empty comb (no bees) on the scale outside and see how much the weight swings over a few days time. Seriously, see what you get. This has to do with determining the signal to noise ratio. If there is +/- 2 lbs of humidity noise, then +/- .2 lbs of thermal noise is probably acceptable.

I'm not against temperature compensating the scale (there is a simple algorithm in the graphing program but it is commented out). But before we add temperature compensation, two thoughts:
1. Lets make the scale as good as we can without temp comp.
2. We need a temperature probe that is thermally coupled to the load cell or frame.

We've just added support for the DS18B20 temperature sensor. I think it would make a good probe.

I prefer to apply the temperature compensation to the output (the visualization) instead of the input (the weight readings). IE The database stores the raw weights and the temperature compensation is applied when it is graphed or warehoused. This way a bad temperature probe or bad readings don't make the weight go crazy.

DavidM
Material & Weight

The attached photo shows the unit sitting in my driveway in the sunlight. There are only hive boxes no additional weights . The material is pine boxes painted

In the attached picture I weighed each hive component listed below to get an accurate weight
Outer Cover 5 lb 8.5 oz
Inner Cover 1 lb 10.7 oz
8 Frame Deep 7 lb 9.3 oz
8 Frame Deep 7 lb 9.9 oz
8 Frame Deep 7 lb 11.7 oz
8 Frame Deep 7 lb 11.0 oz
8 Frame Deep 7 lb 15.5 oz
Bottom Board 3 lb 13 oz

Total Weight 49.6 Pounds after re-stacking the boxes and getting a quick read the scale showed 49.22 pounds so it was pretty close..

49.6-49.2= 0.4 pounds of offset from actual to scale weight.

Devoir

Image: 

DLM

DavidM
Chart

I calibrated this scale initially with 98 pounds or dumb bells from my wife's work out room. The weighed boxes are 49.6 pounds vs 49.21 pounds when i took the reading today.

Image: 

DLM

Paul
Paul's picture
24 hour cycle outside with dumb bells

Let me change my request... Could you remove the wood and put the 98 lbs of dumb bells back on the scale outside for a 24 hour period? Also, to clarify, is this just wood, no wax (foundation or drawn comb)?

If/when you remove the current weight, check the return to zero ie see how close to zero it reads after you remove the weight.

Maybe leave it unloaded for an hour or two and see if it creeps.

DavidM
Paul many thanks....

Paul many thanks....

Also, to clarify, is this just wood, no wax (foundation or drawn comb)? These are empty boxes no frames ( no frames with Wax or comb or foundation), Nothing inside these 8 frame deeps's it's only the wood.

I will remove the hive boxes and stack dumb bells on the scale ... I'll send an update in a short time with a photo for clarification..

Devoir

DLM

DavidM
93 Pound Test

Paul,

I have setup the next test as follows

1- B-04 Scale is the same unit as the prior test. No changes made to the calibration or Pi settings
2- Hive boxes shown in above post was removed.
3- Added 93.06 pounds of weights all dumb bells except the two angle brackets for support they weigh 3 lb 1 oz + 90 pounds of dumb bells
4- test stand is in same location I didn't move it or make any additional changes.
5- Scale sample read 93.21 vs actual weight 93.06 = 0.2ish pounds of difference or negligible.
6- Will run test for the next 24-72 hours....

Test started 8/5/2018 Sunday 3:15 PM Central time

Image: 

DLM

Dennis
Load Cells vs HX711

I seem to recall from a previous discussion that the HX711 is a potential source of temperature-driven variation in weight readings. Perhaps it would be worth looking at the output when the HX711 is provided with a continuous static input. If we get the same variability, then it eliminates the load cells as the culprit. In fact, an alternative experiment could be conducted very quickly indoors with DavidM's setup by simply heating and cooling the HX711 to see if it causes a variation in weight.

DavidM
Indoor Variant

Dennis. Excellent observation I will attempt to run a separate test on a duplicate unit sitting indoors. Let me see if I can do that this weekend.

Any suggestions on what method to provide heat to the board? and how hot it needs to get?

In the mean time I've got 3 full days of data with the outdoor 93 pounds...... It's pretty interesting to see the data. I plan on posting it Friday evening

DLM

DavidM
Test Run Results 96 hours or 4 days.....

Okay here we go....

Test case is the units sitting outside 93.2 pound of dumbells .......

Photo one is after the first 24 hours...the unit shows approximately a 0.2 pound drift in the first 24 hours.

Photo two is after 4 days of operation this is where it gets interesting.... not sure what to make of this..

over the four days we see 94.35 max - 93.13 min = 1.22 pound variance in the weight....

PS the wife wants her weights back.... Might be able to hold her off a few more days.

Image: 

DLM

Paul
Paul's picture
Rain and 1 lb gain on Aug 8

It looks like the weight first goes up about 1/4 lb during the rain. It looks like there may have been .44 inches of rain that day. Is it possible that almost 2 cups of water got into the frame and then evaporated? The bigger one lb gain occurred when the sun was shining so I really don't have a good answer.

Dennis
testing for HX711 weight drift

DavidM -- Sorry I took so long to get back to you. If the HX711 is sensitive to temperature, I doubt it takes a lot of heat to produce a shift in the output. When I originally noticed the drift on an outdoor empty hive, the night time temps were about 45 and the daylight temps were getting up to the mid 70's. Every day, the weight would drop by a pound or so, and bounce back up in the evening. So I'd imagine the heat from a blow dryer would be sufficient to produce a drop in the output, assuming the HX711 is indeed temperature sensitive. Perhaps cooling it a bit initially with a bit of ice to see if the weight goes up at all. Then letting it come back to ambient temperature then warming it with the hair dryer to see if the temperature drops a bit. I'm assuming this experiment would be done indoors where the temperature is fairly constant.

Log in to post comments