Weight sensors.

[[email protected]]

Hi,
I have an application that needs to measure various things in shop
floors in plastic packets basically to try and figure out how many
packets I have based on the combined weight.

Some broad based goals I'm trying to get at :
- In production quantities I might require a few hundred of these
scales.
- The weights will be from 1 kg to around 30 kg.
- Smallest unit weight (hence resolution) would be around 100 gms.
- CHEAP!

I was initially thinking of the same kind of systems used in typical
digital bathroom scales, however no one seems to be suggesting this
approach in a weighing application. Any idea why this is the case?

Load cells seem to be the most common way of doing this, however they
seem to be EXPENSIVE... which directly conflicts with my goal of CHEAP.
i.e both the load-cell and the digital circuit that will essentially
give me an output RS232 (or something similar) . All put together I
want each scale to be in the USD $150 to $200 range.
I read somewhere that piezoelectric based load sensors are
comparatively cheaper but am not able to find anyone selling
those.(couldn't find - or don't know what I'm looking for at
www.kistler.com)

Can someone please give me some insights on whether my goals are
achievable, and what technologies/companies/products I should be
looking for?

Thanks,
A

 

[Spehro Pefhany]

Hi,
I have an application that needs to measure various things in shop
floors in plastic packets basically to try and figure out how many
packets I have based on the combined weight.

Some broad based goals I'm trying to get at :
- In production quantities I might require a few hundred of these
scales.
- The weights will be from 1 kg to around 30 kg.
- Smallest unit weight (hence resolution) would be around 100 gms.
- CHEAP!

I was initially thinking of the same kind of systems used in typical
digital bathroom scales, however no one seems to be suggesting this
approach in a weighing application. Any idea why this is the case?

Load cells seem to be the most common way of doing this, however they
seem to be EXPENSIVE... which directly conflicts with my goal of CHEAP.
i.e both the load-cell and the digital circuit that will essentially
give me an output RS232 (or something similar) . All put together I
want each scale to be in the USD $150 to $200 range.
I read somewhere that piezoelectric based load sensors are
comparatively cheaper but am not able to find anyone selling
those.(couldn't find - or don't know what I'm looking for at
www.kistler.com)

Can someone please give me some insights on whether my goals are
achievable, and what technologies/companies/products I should be
looking for?

Thanks,
A

Counting scales are really well-available technology. Just buy them
fergoodnessake.

Kisler and other similar stuff basically has an AC response. Good for
detecting the injection pressure in a plastic mold perhaps, but no
good for weighing. You want strain gauge type sensors, which are fussy
to work with.


Best regards,
Spehro Pefhany

 

[[email protected]]

Some broad based goals I'm trying to get at :
- In production quantities I might require a few hundred of these
scales.
- The weights will be from 1 kg to around 30 kg.
- Smallest unit weight (hence resolution) would be around 100 gms.
- CHEAP!

The last time I was shopping for counting scales, they seemed to start
at around US $300 or so, and I think they fit the above criteria. I
know the resolution was much better than 100 g, but I'm not sure if the
cheap ones went up to 30 kg total or not. If you can buy a hundred at
a time, you should be able to negotiate a better price with the
manufacturer or dealer. Google "counting scale" for many vendors.

Matt Roberds

 

[[email protected]]

I also have some other things I need to do, RF data transmission to a
remote location, possibly temperature sensing etc.
any idea if there are kits available for such counting scales?

Just wondering, how come load cells are so expensive??

 

[John Fields]

I also have some other things I need to do, RF data transmission to a
remote location, possibly temperature sensing etc.
any idea if there are kits available for such counting scales?

Just wondering, how come load cells are so expensive??

---
The strain gage itself has to be manufactured so that its output will
be predictable and repeatable over temperature and time with a _very_
small flexure resulting in a _very_ small resistance change as its
output. Then, it has to be precisely mounted onto the flexing surface
of the load cell so that the flexure of that surface is translated
into the desired resistance change in the strain gage per unit of
force exerted on the load cell. In order for that to happen, the load
cell itself has to be precisely machined out of material with
predictable characteristics, the strain gage properly secured to the
flexing surface, and the finished unit calibrated.

 

[martin griffith]

---
The strain gage itself has to be manufactured so that its output will
be predictable and repeatable over temperature and time with a _very_
small flexure resulting in a _very_ small resistance change as its
output. Then, it has to be precisely mounted onto the flexing surface
of the load cell so that the flexure of that surface is translated
into the desired resistance change in the strain gage per unit of
force exerted on the load cell. In order for that to happen, the load
cell itself has to be precisely machined out of material with
predictable characteristics, the strain gage properly secured to the
flexing surface, and the finished unit calibrated.

How good is Hookes Law these days?
I noticed, a few weeks ago that digikey had some panasonic very
linear, linear pots, sorry no p.no.
How about a spring with a linear pot attached?



martin

 

[Roger Hamlett]

Hi,
I have an application that needs to measure various things in shop
floors in plastic packets basically to try and figure out how many
packets I have based on the combined weight.

Some broad based goals I'm trying to get at :
- In production quantities I might require a few hundred of these
scales.
- The weights will be from 1 kg to around 30 kg.
- Smallest unit weight (hence resolution) would be around 100 gms.
- CHEAP!

I was initially thinking of the same kind of systems used in typical
digital bathroom scales, however no one seems to be suggesting this
approach in a weighing application. Any idea why this is the case?

Load cells seem to be the most common way of doing this, however they
seem to be EXPENSIVE... which directly conflicts with my goal of CHEAP.
i.e both the load-cell and the digital circuit that will essentially
give me an output RS232 (or something similar) . All put together I
want each scale to be in the USD $150 to $200 range.
I read somewhere that piezoelectric based load sensors are
comparatively cheaper but am not able to find anyone selling
those.(couldn't find - or don't know what I'm looking for at
www.kistler.com)

Can someone please give me some insights on whether my goals are
achievable, and what technologies/companies/products I should be
looking for?

Thanks,
A

This is a problem that has been raised here many times before. Load cells,
are designed to be accurate over a wide temperature range, and to handle
loads that may contain 'off centre' components. As such the metalwork has
to be made of materials with well known metalurgy, quite complex accurate
machining, then a pair of well compensated strain gauges. Hence their
cost...
You can make quite good weight measure systems a lot cheaper, using other
approaches, and some of these are similar to the systems in bathroom
scales. Simple beams, with a pair of cheap strain gauges, or a spring, and
then measure the movement using a pot, or a quadrature sensor (cheap
version like those used in a mouse, rather than a high accuracy system).
Going more upmarket, capacitive deadweight sensors, may also be practical.
Provided the temperature range is a lot more restricted, and you can
compensate for non-linearity in software, many of these approaches may
well be acceptable. I'd try looking at a pair of cheap strain gauges, on a
simple bar, or measuring spring extension, with a mouse type system, as
probably being the most reliable, and likely to be reasonably priced
approaches.
If you try some tests on the bathroom scales, you will find that the
results are quite remarkale for their poor behaviour, especially at the
bottom of the weighing scale...
One approach that could possibly be good, is a piezo pressure sensor
(these are units that measure gas pressures), attached to a 'balloon'
under a platform. These sensors are cheap, and if you use the
'differential' units, will measure how much the pressure in the balloon is
above atmospheric, giving a direct readout of the load on the balloon.
The same type of piezo sensors are available as force sensors, but are
generally more expensive than the form used for pressure reading. Search
on 'piezoresistive' sensors to find these.
You might also want to look at QTC's. These are rather nice, and only
involve reading resistance, which is relatively simple:
http://www.peratech.co.uk/sensors.htm

Best Wishes

 

[John Fields]

How good is Hookes Law these days?

 

[Paul Hovnanian P.E.]

How about something like a balance beam scale, but replace the sliding
weights with a solenoid driven by a PWM current source. Attach a
position sensor (optical?) to the end of the beam to drive a control
loop controlling the solenoid current. The solenoid force (and weight)
is proportional to its current.

 

[Ben Bradley]

How good is Hookes Law these days?
I noticed, a few weeks ago that digikey had some panasonic very
linear, linear pots, sorry no p.no.
How about a spring with a linear pot attached?

That would work if you're able to satisfactorily reduce friction
and other sources of nonlinearity (I think that's a pretty big if to
get good accuracy), but I recall that such scales use something with
much more accuracy, a "force feedback" device: The load cell or
whatever type of sensor's output is measured with nothing on the
scale. When something is added, a coil is energized to push a magnet
back up the scale, so as to bring the measured weight back to the
nothing-on-the-scale weight. The current into the coil is
proportional, with high linearity*, to the weight of what was put on
the scale. The weight sensor device doesn't have to be very linear, it
only has to reliably give the same output when the force is reduced to
the nothing-on-the-scale case (thus have very low hysteresis).


* Magnets's fields do vary slightly with temperature, but that could
be compensted for with a temperature sensor, or by replacing the
magnet with an electromagnet driven by a precisely regulated current.

 

[Joerg]

Hello John,

After the recent eminent domain decision it wouldn't surprise me at all
if it were repealed ;-)

Regards, Joerg

 

[Jim Thompson]

Hello John,


After the recent eminent domain decision it wouldn't surprise me at all
if it were repealed ;-)

Regards, Joerg

Yep, the jerks think political "science" IS science :-(

...Jim Thompson

 

[James Waldby]

I have an application that needs to measure various things in shop
floors in plastic packets basically to try and figure out how many
packets I have based on the combined weight. ....
- The weights will be from 1 kg to around 30 kg. ....
I want each scale to be in the USD $150 to $200 range. ....
Can someone please give me some insights on whether my goals are
achievable

....

ebay has digital scales that meet your weighing specs; eg,
http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=5595228178
is 220Lb/100 Kg, 0.05 Kg or 2 ounce res., $80 delivered.
Also has a unit weight feature.
http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=5593676716
is 52 pounds in 0.2 ounce increments. $36 delivered.
http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=5595839066
is 75.0 pounds/0.5 ounce res, 34 kg/10 gram res.

These don't list any computer readout like you mentioned
in later post and I don't know how typical that is.\

 

[Joerg]

Hello James,

These don't list any computer readout like you mentioned
in later post and I don't know how typical that is.\

Very typical. Any decent grocer's scale can do that. The ones I have
seen did it via RS232 right into a cash register controller but that was
at least ten years ago. The accuracy requirements were very stringent
and IIRC it was tested and calibrated once a year.

Regards, Joerg

 

[[email protected]]

I also have some other things I need to do, RF data transmission to a
remote location, possibly temperature sensing etc.

Uh oh... requirements creep!

any idea if there are kits available for such counting scales?

Not sure. You might be able to buy an inexpensive non-counting scale
with an RS-232 output and implement the counting part in your own
microcontroller. This microcontroller could also be in charge of the
RF data module, A/D for temperature sensing, and so on. It might be
somewhat of a pain to have to make up your own user interface for
setting piece weight and displaying the count. Or, you might want
just a count display, with the piece weight setting done by a
non-user-accessible control or a separate device to prevent tampering.

Just wondering, how come load cells are so expensive??

I didn't know, but several other posters have answered that question.

Matt Roberds