View Full Version : Looking for simple inexpensive way to weigh

Don M
12-23-2011, 04:00 PM
Clever title 'eh?

Anyway was thinking of an idea of a way to weigh items placed inside a plastic bin like these:


Maybe some sort of film (resistive?) maybe like an inexpensive resistive touch screen overlay?

For instance if you were placing screws or nuts or bolts in the bins and wanted to be able to measure the weight to weigh count. Wouldn't have to be absolutely accurate. The problem may be on how small a resolution of weight change might be available from such a film.

Any other ideas of some sort of substrate or film or just another way to do this? Has to be simple and inexpensive due to volume involved. Would like it to be inside bin but another consideration could be something the footprint size of the bin underneath where the bin sits.

Duane Degn
12-23-2011, 04:25 PM
If want to resort to using a balance, I've use one of these (http://www.balances-online.com/digiweigh-dwp-98-counting-scale-3-to-30-kg.html) (the 3Kg version) for several years. I've been very pleased with it.

It outputs through a serial line (after I added the appropriate chip to the empty socket inside).

12-23-2011, 04:29 PM
Bathroom scales use something called a strain gauge and NerdKits has an article on interfacing one to a microcontroller.


12-23-2011, 04:35 PM
erco weighing in... :)

Not easily interfacable, but cheap & accurate:

$2.89 shipped! http://www.ebay.com/itm/40kg-x-20g-Electronic-Portable-Digital-Hanging-Weight-Pocket-Balance-Scale-/260907824221?pt=LH_DefaultDomain_0&hash=item3cbf51745d

$4.96 shipped http://www.ebay.com/itm/Digital-Pocket-Mini-Balance-Scale-500-x-0-1g-500g-Weight-Weigh-Silver-New-/280794896932?pt=LH_DefaultDomain_0&hash=item4160ae1a24

Don M
12-23-2011, 04:43 PM
I don't have any experience with a resistive touch overlay for a LCD screen but do they have a linear output with respect to touch pressure?

Don M
12-23-2011, 04:51 PM
Or something similar to this stuff: http://www.parallax.com/Store/Sensors/PressureFlexRPM/tabid/177/CategoryID/52/List/0/SortField/0/Level/a/ProductID/384/Default.aspx

This could be put inside the bin in the center and then a "floor" for the bin put on top that has a dimple at the bottom center to come into contact with the film.

12-23-2011, 06:59 PM
If you have the time to build the mechanical side, then use a spring and lever to drive change in a variable resistor, and pick up the result with sigma delta. Non-linearity could be corrected in software assuming the spring is consistent.

Alternatively use the capacitance of two metal plates with a moderately flexible dialetric. You could try a double sided fiberglass PCB, with a fixed resistor and as above read using sigma delta. The microprocessor must be close to the capacitor (metal plates).

Though these home cooked solutions are not likely to be accurate nor repeatable over the long term due to the fatigue in materials. An off the shelf solution would be better, but probably not as much fun.

12-23-2011, 07:02 PM
Don't think that sensor wil do what you want.
The sensor included in the kit has a resolution of ≈ 1.6lbs
Also touch screens just respond to touch, not variable pressure.

12-23-2011, 07:12 PM
I am going to go to Hell for suggesting this (I actually work for a company that specializes in industrial scales) but you could try the resistive foam that is used to make most keyboards nowadays. That might give you a REALLY EXTREMELY APPROXIMATE estimate of whether the bin is nearly empty, half, or full. If you can get any better than REALLY EXTREMELY APPROXIMATE though you're poised to make a fortune in the scale biz.

There are basically three ways to do this right:

1. Float the bin on a set of flexures or levers that will keep it parallel and transmit the force to a single sensor, the cheapest of which would probably be a couple of plates held apart by a spring whose capacitance changes as weight pushes them together.

2. Buy a self-contained engineered strain gage load cell which is designed to bolt to a firm mount on one end and a platform on the other, which basically contains the flexures and such mentioned in #1 internally. Prices start around USD$200.

3. Put four separate load sensing elements at the corners and add them up either by analog or digital methods. You might find that the conductive foam I mention works better if you use four small bits on the corners wired in parallel.

12-23-2011, 07:25 PM
The only way that I know of to get the accuracy that you need to be able to "count parts by weight" is to use a load cell combined with a nice instrument amp and d/a to convert the output to a digital value. Low cost is a relative term. Of course the heavier each item weighs the easier and cheaper it'll be. :-)

Phil Pilgrim (PhiPi)
12-23-2011, 07:49 PM
Get a bunch of cheap servos, and put a cam on each one. Mount them under the bins, such that when each rotates, the cam lifts the bin a little. Wire all the red and black leads in parallel with a current sense resistor in series with the black lead to ground. From the current sense resistor, connect an op-amp and ADC to provide a current measurement to the micro.

Now, activate each servo in turn and measure its current profile as it rotates. The heavier the load, the more current it will take to lift the bin -- or the sooner it will stall. What the cam does is to guarantee some movement before the servo stalls. So you can measure how far the servo got before hitting its stall current. You will also be measuring the quiescent current of all the other servos, but that can be subtracted from the active reading.


Duane Degn
12-23-2011, 08:03 PM
Don't think that sensor wil do what you want.
Also touch screens just respond to touch, not variable pressure.
The touchscreens that come with Rayman's 3.5" and 4.3" LCDs have a X, Y and Z component. The Z is apparently how hard the screen is being pressed. I'd be very surprised if you could use this information to weigh things though.

Beau Schwabe
12-23-2011, 10:30 PM
If the platform is sitting on a 'spring' and the 'spring' is part of an LC oscillator circuit, you could simply measure the frequency change in the LC circuit as the spring compresses.

12-23-2011, 11:04 PM
Some interesting ideas for solutions I just wonder if they would give enough resolution to allow for counting parts by weight.

As a modification on the spring idea would be to us a hollow core inductor and have the weight of the parts move a ferrite core in and out of the coil. Not unlike the commercial coils for "home made" crystal radios that I used to get from Radio Shack and another long closed electronics parts store when I was a kid!

Then I discovered I could wind my own coil from wire taken from the "guns" of thrown out TVs and paper towel rolls. :-)

12-23-2011, 11:31 PM
4x5n, every idea like that has already been tried, tested, tested more, tested even more, and either is in current use or doesn't work. The problem with using an inductor core is that it's not linear. Your big challenge is to get around the problem of keeping the weigh platform parallel, so that force at any point on it results in the same change in whatever sensor you have taking the load. The very cheapest solutions to that which work at all tend to be mechanically complex, and using multiple sensors instead tends to be even more expensive than the mechanical lever arrangements.

Scales are one of humanity's oldest technologies. The ancient Egyptians literally worshipped them, personified as the goddes Ma'at, because of the hidden truths they revealed. As I like to tell my customers, humans have been using scales for 5,000 years -- and trying to figure out how to cheat them for 4,999 years.

Jorge P
12-24-2011, 12:16 AM
You can make a variable resister/pressure sensor from anti static foam that IC's come in :smile: http://forums.parallax.com/showthread.php?131784-What-would-you-use-for-a-resistor&p=1002959&viewfull=1#post1002959

12-24-2011, 12:56 AM
I have to agree with localroger. Almost any technique you can think of has already been tried and tested, and the useful ones are in service now. There are only two techniques I am aware of that would give you the required accuracy. They are a load cell or magnetic levitation. Both of those require regular calibration to guarantee accuracy.

I set up something similar for a company that sold a lot of small consumables and parts for scientific instruments. A laboratory scale with serial output was connected to a PC running a program that performed the calculations. The functions performed were:

Store a part number and maximum quantity for an empty bin.

Store an array of points (weight/quantity) for a calibration curve for each part. The first weight was that of the empty bin, then increasing quantities of parts were added and recorded.
Calculate calibration curve factors and store them.

Place the bin for the displayed part number on the scale. Weight of bin is subtracted and quantity calculated.

Parts Picking
Enter/select bin/part number and load parts onto scale until desired quantity is displayed.

A second lower capacity scale was added later to increase accuracy for the very light parts like ceramic washers and spacers.

12-24-2011, 01:36 AM
I have a reloading (ammo) scale that I hacked a few years ago. (Its a Scout II). I tapped the serial info going to the LCD display on the scale. It was 4800 baud 8N1.I used my Oscope to find the serial stream, It was 5volts...Perfect for the stamp.An easy hack....
The Scout II only reads to about 1 1/2oz, So I think you'll need something a little bigger.
You might be able to hack this little scale,And make it talk to a Micro....Its only $20 bucks.
I've seen larger scales in the same price range that might be easer to hack.(Larger parts/components/pin spacing)

12-24-2011, 01:21 PM
For a home built solution try a balance with a sensor at the pivot point, pot, encoder, switch etc. the balancing weight could be fixed and set at various distances or variable.