bathroom scale hysteresis

Sneaky scales fudge weights deliberately.

I suspect that many modern, microprocessor -controlled bathroom scales have programmed in hysteresis to increase user confidence in the scale's accuracy. The scale remembers fresh measurements, say 166.2#, and if it next measures same weight +/- a little (within a short time), say 166.8#, then the scale decides to report the *original* measurement

166.2#. Neat. Sneaky. User believes the scale is highly repeatable.

On such a scale, I weigh myself multiple times and get the same reading (to the 0.2#) each time. 166.2#, 166.2#, 166.2#, 166.2#, ... Then I weigh something different (myself holding a load), to reset the memory. Then I weigh myself again... Now, get something like 166.8#, 166.8#,

166.8#, 166.8#, ... very solid again, but 0.6# different from 1st series of measurements.

I tried 6 models (2 or 3 brands) at a retail store display and find this "feature" common.

Is it important? Maybe yes in the following scenario -- in some sports like wrestling, boxing, judo, you have multiple competitors weighing in at the same time, same scale, with possibly very similar weight. Some competitors are concerned with as little as 0.25#. In this case it seems one competitor could inherit the weight measurement of the person in front of him.

In a perfect world, competition weigh-in equipment should be certified/calibrated. But since the bathroom scale appears so repeatable, some competitions now use modern, microprocessor

-controlled bathroom scales.

Somebody's going to say... "you shouldn't do that". Right, I agree. But (a) it's happening, because (b) this hysteresis (memory)phenomenon isn't widely known, I suspect.

Anyone care to confirm? Contradict? Repeat the experiment on their own scale? Comment? In my experiment I didn't bother to determine what weight difference resets the memory -- 1#? 1.5#?

Regards, John Ruckstuhl

Reply to
john.ruckstuhl
Loading thread data ...

What models/brands? I find it hard to believe that the manufacturers would do that, because it would require additional programming in the scale, and most modern scales are quite accurate as is.

FWIW, I have a 4 year old Tanita, accurate to 0.2 lb, and I've tested it like you suggest on a number of occasions. In all such tests, (weigh in, weigh-in holding a light weight of some sort, then weigh in again) the scale returns to within 0.2 lb of the original reading, and usually to the exact original reading.

If I step off, reset, and step back on (without changing anything), I get the same reading about 65% of the time. The other 35%, it varies by 0.2 lb - I assume this is when my actual weight is somewhere between the two reported weights.

Perhaps your test was flawed somehow? Or, maybe you were testing some crappy scales?

--
GG
http://www.WeightWare.com
 Click to see the full signature
Reply to
GaryG

[snip]

The reproducibilty of an electronic bathroom scale depends on the surface underneath. If it is stone or rigidly laid hardwood you'll get good precision. If it is soft like linoleum or carpet, you won't. Any need for true precision weighing cannot be satisfied by a cheap consumer scale.

--
Uncle Al
http://www.mazepath.com/uncleal/
 Click to see the full signature
Reply to
Uncle Al

Exactly the same experience I had with a digital scale. It worked out to a 5 pound window - I dont know the time window. Showed weight to within 0.5 pound as I got on and off with increasing hand-held weights until about 5 extra pounds.

I still have it, plan to rework the electronics with a loud voice output and leave it in the bathroom to catch unsuspecting house guests....

"Ow, oooh, get the heck off me!"

--
Luhan Monat: luhanis(at)yahoo(dot)com
http://members.cox.net/berniekm
 Click to see the full signature
Reply to
Luhan Monat

Bathroom scales have historically been rather (to be polite) inaccurate. Notice in a grocery that the scales are all spring scales (ie: inexpensive) and are marked "NOT FOR TRADE" or "NOT LEGAL FOR TRADE". They all read *approximate* values, and if not calibrated, can be off by 20 percent of full scale (either way) for starters. If you want more reliable or accurate readings, use a balance type of scale: Toledo comes to mind, Cenco comes to mind; there are others.

Reply to
Robert Baer

I've noticed with balance scales that the weight shifts by several pounds depending upon where you stand on it ... I assume you're supposed to stand more or less in the center, an inch one way or another can make a difference of a pound or two (important if you're doing a sport with weight divisions).

-JS2

Reply to
JS2

"I've noticed with balance scales that the weight shifts by several pounds depending upon where you stand on it ... I assume you're supposed to stand more or less in the center, an inch one way or another can make a difference of a pound or two (important if you're doing a sport with weight divisions)."

This would be the incentive to repeat the last reading. This would make the electronic scale seem more dependable, stop customers from worrying about water weight gains and losses. It would take a very small amount of digital memory and programing. I am sure there is even memory for two or three "normal weights". There is probably a mass produced chip that includes this program and you will need to find one with a different chip to change the repeat. I bet you even paid extra for the "added accuracy"!

Reply to
Quantum Mirror

Not only have I never heard of such a thing, I've seen the workings of scales that go to great lengths to avoid just that. If your scale shows different weights when the object is on different places on the platform, then you have been robbed.

Cheers! Rich

Reply to
Rich Grise

Now this is just plain silly. Who in their right mind would go to such great lengths, just to produce a reading that isn't even what the weight on the platform is?

The phenomenon that the OP, John Ruckstuhl, mentioned: "On such a scale, I weigh myself multiple times and get the same reading (to the 0.2#) each time. 166.2#, 166.2#, 166.2#, 166.2#, ... Then I weigh something different (myself holding a load), to reset the memory. Then I weigh myself again... Now, get something like 166.8#, 166.8#, 166.8#, 166.8#, ... very solid again, but 0.6# different from 1st series of measurements."

Is from hysteresis. He steps on the transducer, it's been sitting all day stabilizing, so it reads low. He steps on the scale holding weights, It shows his 166.2 + the weight + some unknown error. He steps off the scale, and it springs back to something greater than zero; the scale waits until it's done bouncing, re-zeros itself, and now when he steps back on it's biased high. The effect would probably be even worse if he just stayed standing there and set the weight down.

The point being that the sensor takes a "set." And, probably not so much these days with little strain gauges, but with mechanical scales, there's sticktion to deal with.

Cheers! Rich

Reply to
Rich Grise

I've seen it on doctor's scales, and the typical scale (lever arm with weights) used in most gyms (don't have any kind of scale at home), and hence in wrestling and judo competitions. Maybe they're just old scales so some internal mechanism is worn out, but it seems to be very common. Not an issue if you're looking to measure your weight accurately (just stand at the center), but something that can be used to your advantage (or disadvantage) if you're making weight division and even a half pound makes or breaks you. My thought was that if the scale isn't designed properly, it might change the action point on the lever arm (even .5% change is good for a pound if you're about 200 lbs).

-JS2

Reply to
JS2

This would not explain my experience, just picking up a little extra weight each time and getting back on the scale. It just would not move until about 5 pounds total had been added. Only explanation I have is that is it 'faking' the accuracy; and doing it quite intentionally.

That's why is on my list of projects for re-engineering. Maybe Mavin's voice "You standing on my like that is just so Depressing!"

--
Luhan Monat: luhanis(at)yahoo(dot)com
http://members.cox.net/berniekm
 Click to see the full signature
Reply to
Luhan Monat

A properly designed and calibrated scale should not exhibit this behavior. Calibrations should be performed according to NIST Handbook 44, which includes a shift test, a return to zero (testing mechanical hysteresis), an increasing load test and others depending on the type of scale.

Also, ISO 17025 for calibration labs require all as found readings to be measured and reported before any adjustments to calibration are made. As left readings are also recorded. Acceptance tolerances and maintainance tolerances are specified in Handbook 44 and vary by Class specifications. Most legal-for-trade requirements defer to Handbook 44.

Reply to
ck

I've answered this in a different branch - the sensor has a certain amount of hysteresis, but also excellent "repeatability" as long as you haven't moved the sensor's threshold in between, which is what John says is happening. The sensor takes a "set", or something very much like that. But I guess I'd have trouble blaming sticktion when they're using strain gauges - but anything, to measure how hard it is being stressed or strained, has to have some kind of change take place, or there wouldn't be any signal. Generally, that's movement of some kind. So, if it's something that flexes, it simply doesn't unflex back to _precisely_ the same position, for any number of reasons. At least not right away. Probably after awhile, yes. So when he first gets on, the sensor has been relaxing all night, so has no "set." He steps on the scale first time, and puts whatever "set" that 166.2 puts on it. Then, when he steps off and on, The force on it is the same, so it doesn't add any "set", and since the thing has good repeatabilty, it reads the same. Then, he steps on it with weights on, and it reads, say, 191.2. (It would be very interesting to see this number, if the weights are calibrated). This increases the "set" to S(191.2). Now, he puts 166.2 on it, but it doesn't have enough force to even _reach_ the higher "set," so effectively zero has backed away by 0.6, and it reads 166.8.

Nothin' to it! ;-)

Cheers! Rich

Reply to
Rich Grise

Sorry, something is getting missed here. I kept picking up an extra pound of stuff each time and then getting back on the scale. So it had some chance to re-zero. But all it did was report exactly the same weight each time. When the accumulated extra weight was about 5 pounds, then, and only then, did it jump to a new weight. Then again, it claimed that to be the real weight as a dropped off a pound at a time getting on and off the scale.

The software in the scale was obviously fudging the numbers to produce a false sense of accuracy.

--
Luhan Monat: luhanis(at)yahoo(dot)com
http://members.cox.net/berniekm
 Click to see the full signature
Reply to
Luhan Monat

On Sat, 28 May 2005 15:17:48 -0700, Luhan Monat wroth:

I'd be willing to bet that there is a sticker on the scale or a note in the owners manual with a phrase like "... not legal for trade".

You bought a freakin' bathroom scale fer cripe's sake, not a NIST certified research instrument. Get over it.

Jim

Reply to
James Meyer

For about $180 you can get a freon charging scale with 400 lb range and accurate and reproducible to a fraction of an ounce. I guarantee there is no funny stuff like you are seeing in the firmware :-). I forget the brand we got last year at work but it has those specs.

-- Regards, Carl Ijames carl.ijames at verizon.net

Reply to
Carl Ijames

I think it is quite common. Mine does the same thing. What I do is move about to get the maximum, then the minimum, then split the difference (typically about 5 kilos!).

--

John Devereux
Reply to
John Devereux

I am 'over it', hun. This discussion is about whether the error was just 'cheap design' or 'intentionally fraudulent'. It was only of interest to me at the time because such technology (micro controllers) had not been used before to fake extra accuracy.

I simply recommended to my friends that they buy analog scales: inaccurate maybe, but at least they could show moderate weight gain or loss reliably.

--
Luhan Monat: luhanis(at)yahoo(dot)com
http://members.cox.net/berniekm
 Click to see the full signature
Reply to
Luhan Monat

I think you just got a bad scale. My 4-year old Tanita does not have that problem at all. I can step on and off, and always get the same reading (or, within the 0.2 lb resolution). If I step off, drink an 8 oz glass of water, and step back on, it shows that I've gained 1/2 lb. If I weigh myself before and after a crap, it shows (appropriately) 1.5-2 lbs lost.

No weigh (get it?) would I go back to those crappy old analog dial scales.

GG

Reply to
GaryG

Then your scale is lying to you. It actually has a resolution of 5 lbs, but since the designers thought that wouldn't sell well, they just make up some random interval, or maybe interpolate based on the time of day or barometric pressure or something.

Yes. Obviously. :-)

Did you get it for $5.00 at an auction? ;-)

Cheers! Rich

Reply to
Rich Grise

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.