There’s a digitizer in there somewhere that converts a mechanical value of force, to a digital value. Low-quality digitizers have little errors called differential nonlinearities that would make some values more likely than others. The value of 184.8 might be one of those that is on a part of the response curve where there’s a little flat spot.
Imagine putting a large bucket on the scale and slowly filling it at a steady rate with water. Instead of the numbers counting up (by 0.2 pounds) at a very regular rate, with a poor digitizer you’d see it go up irregularly, spending more time on some values and less on others, likely even skipping some numbers. The 184.8 number could be one of those that it would spend way too much time on.
I’m not a scale expert, but showing that your scale is inaccurate - and they all are, aren’t they? - and saying that manufacturers have an elaborate “accuracy cheating scheme” is not the same.
CurtC: I agree the problem could be differential non-linearity in the A/D, but I would first suspect a mechanically-induced non-linearity problem.
What others have said is correct… you basically have a load cell, instrumentation amplifier, and ADC. We have a lot of experience calibrating load cells where I work, and have run into similar problems, i.e. “sticky values.” This is usually the result of someone failing to exercise the load cell before use.
Let me explain… a load cell is (usually) a strain-gage configured as a four-arm (four-resistor) Wheatstone bridge. When subjected to a force, two of the resistors will be lengthened while the other two will be shortened. This causes an imbalance in the bridge, resulting in the bridge producing a differential output voltage. The thing to keep in mind is that each resistor is subjected to mechanical stress. The resistors are not perfectly elastic; when under tension or compression they tend to “stick” at certain lengths and “hop over” certain values. In other words, as you’re pulling on (or compressing) a resistor the length does not follow the force in a perfectly smooth fashion. To correct this (at least temporarily) we instruct our technicians to “exercise” a load cell before use. This is done by slowly subjecting the load cell to its full load rating (in both tension and compression) three times before use. Not only does it help the readings become “less sticky,” but it removes hysteresis.
You’re micromanaging your body if you are concerned with increments less than 1 lb.
As for bathroom scales, it registers what it ‘sees’. It sees more weight, or what it believes is more weight, and registers higher.
Whatever you are actuating when you step on and off doesn’t actuiate perfectly everytime. Maybe moisture in the air, or the temp in the room, or the exact position of your feet all add some degree of variance in how the mechanics execute when you stand and step off the scale.
No. In my post I stated that a load cell must undergo at least three full tension and compression cycles in order to:
a) Remove hysteresis
b) Smooth out and remove the “stickiness” in strain gage (4-arm Wheatstone bridge)
So here’s my suggestion: Take a large weight (the maximum allowed by the scale’s specs) and place it on the scale three or more times. Afterwards tare it and record your weight. I’m confident the linearity and accuracy will improve afterwards.