I run a small single seater racing car which means 2 things in terms of weight. I need to keep it as close to the minimum of 420kg as I can, and I need to ensure its corner loadings are balanced.
there are of course fancy scales designed for such a purpose, but as with anything in racing, it attracts more zeros on the price than is reasonable.
The simple way is to use 4 sets of bathroom scales and this is doable with some wedges so I can roll the car on without damaging the scales. the usage issue is that I have 4 displays to read, and I have a feeling modern scales will auto-off quite quickly and might be awkward to use.
so I wondered, if I have the scales, could I interface their strain gauge to a small 328l and then serial cable or Bluetooth the data to a central display like the 'proper' ones do ? or am I better making the whole thing myself ?
does anyone have experience with strain gauges, what is the interface generally - I guess an analogue output, and if I want to weigh 180kg in 0.25kg (min) increments what hardware could I use to do this accurately given I could potentially assume the first say 80kg on anywheel is a given.
any guidance/tips much appreciated.
Strain gauges and weighing cars
Strain gauges and weighing cars
I can help to the extent that I've considered hacking a bathroom scale
for a similar project and abandoned the notion while awaiting some
stroke of insight that might ease the task.
The scales I bought for $12 each contain four three-wire strain gauges
epoxied on small beams. They are driven by a squarewave and yield small
(mV) differential signals that a mystery IC under a black blob
interprets. The blob directly drives a 32-segment LCD via conductive
elastomer connectors.
https://dl.dropboxusercontent.com/u/917 ... CN1513.JPG
https://dl.dropboxusercontent.com/u/917 ... CN1517.JPG
Although there exists a seven-pin connector on the circuit board, I
could make no sense of signals on it; it might be for factory
calibration. Lacking reasonable data on that connector, the only other
output is the LCD; decoding a multiplexed 32-segment LCD to TTL levels
is not trivial, requiring voltage translation and timed latching.
Alternatively, you might consider small-signal processing for each
resistive strain gauge; I haven't pursued that beyond some thought. I
suspect you'd need to generate a known signal, amplify the small result,
precision-rectify and scale to the ADC input range, and probably correct
for temperature. The XMega processors offer differential inputs and
gain, perhaps easing some of that task.
FWIW, the scales I looked at use a vibrating spring switch (the black
cylinder on white wires in the photo) to wake the scale. You might
defeat an otherwise-intact scale's auto timeout by shaking that cylinder
with a cellphone vibrator motor; the scales should continue to display
until you turn the vibrators off.
I'll be interested in what progress you make.
Tom
for a similar project and abandoned the notion while awaiting some
stroke of insight that might ease the task.
The scales I bought for $12 each contain four three-wire strain gauges
epoxied on small beams. They are driven by a squarewave and yield small
(mV) differential signals that a mystery IC under a black blob
interprets. The blob directly drives a 32-segment LCD via conductive
elastomer connectors.
https://dl.dropboxusercontent.com/u/917 ... CN1513.JPG
https://dl.dropboxusercontent.com/u/917 ... CN1517.JPG
Although there exists a seven-pin connector on the circuit board, I
could make no sense of signals on it; it might be for factory
calibration. Lacking reasonable data on that connector, the only other
output is the LCD; decoding a multiplexed 32-segment LCD to TTL levels
is not trivial, requiring voltage translation and timed latching.
Alternatively, you might consider small-signal processing for each
resistive strain gauge; I haven't pursued that beyond some thought. I
suspect you'd need to generate a known signal, amplify the small result,
precision-rectify and scale to the ADC input range, and probably correct
for temperature. The XMega processors offer differential inputs and
gain, perhaps easing some of that task.
FWIW, the scales I looked at use a vibrating spring switch (the black
cylinder on white wires in the photo) to wake the scale. You might
defeat an otherwise-intact scale's auto timeout by shaking that cylinder
with a cellphone vibrator motor; the scales should continue to display
until you turn the vibrators off.
I'll be interested in what progress you make.
Tom
Tom
After my original post I did some more digging and discovered the proper term seems to be load cell as opposed to strain gauge.
ebay as per usual turns up some interesting items - cheap load cells and a cheap adc & amp module
http://www.ebay.co.uk/itm/Electronic-Sc ... 3f384b3bde
http://www.ebay.co.uk/itm/1x-Force-Modu ... 27e0d58ef8
though the information supplied about the data output format is a bit vague to say the least.
however that still leaves the mechanical part. you can see in the video below how this can be solved, so buying cheap scales is probably the cheapest way to get the mechanical part sorted too but I can image all sorts of proprietary internals these days.
this was quite informative as a starting point
http://www.nerdkits.com/videos/weighscale/
tempted to buy a module and cheap load cell for a play but will do some more digging first.
ebay as per usual turns up some interesting items - cheap load cells and a cheap adc & amp module
http://www.ebay.co.uk/itm/Electronic-Sc ... 3f384b3bde
http://www.ebay.co.uk/itm/1x-Force-Modu ... 27e0d58ef8
though the information supplied about the data output format is a bit vague to say the least.
however that still leaves the mechanical part. you can see in the video below how this can be solved, so buying cheap scales is probably the cheapest way to get the mechanical part sorted too but I can image all sorts of proprietary internals these days.
this was quite informative as a starting point
http://www.nerdkits.com/videos/weighscale/
tempted to buy a module and cheap load cell for a play but will do some more digging first.
found this which helps a lot http://www.dfrobot.com/image/data/SEN01 ... nglish.pdf
Re: Strain gauges and weighing cars
Well, at least with the scales I tried, continually vibrating the shake switch does not keep the scale display active; it blanks and resets a few seconds after the value has stabilized. The only solution to that appears to be forcing a small continuous weight change.GTBecker wrote:... the scales should continue to display until you turn the vibrators off.
Have you made any progress with the load cells?
Tom
I've ordered a variety of things to play with.
One 4 wire gauge, just 1kg to play. I expect this work fine but hard to make a platform for a car and ensure the load is transferred to a single gauge effectively.
http://www.ebay.co.uk/itm/181284138618? ... EBIDX%3AIT
then 4 x 50kg 3 wire sensors to be used as corner gauges on a platform. should be easy enough but will require more interface boards than I intended but at like £2.50 each that's not a big issue. being 3 wire they may suffer from temperature drift and require calibration but we'll see.
as some of the stuff is coming from china it will be a week or so more.
I intend to make one scale using the 4 sensors and see how accurate it is and stable.
One 4 wire gauge, just 1kg to play. I expect this work fine but hard to make a platform for a car and ensure the load is transferred to a single gauge effectively.
http://www.ebay.co.uk/itm/181284138618? ... EBIDX%3AIT
then 4 x 50kg 3 wire sensors to be used as corner gauges on a platform. should be easy enough but will require more interface boards than I intended but at like £2.50 each that's not a big issue. being 3 wire they may suffer from temperature drift and require calibration but we'll see.
as some of the stuff is coming from china it will be a week or so more.
I intend to make one scale using the 4 sensors and see how accurate it is and stable.
By way of an update. I have spent some time on this and are close to abandoning it. There are a number of issues, not insurmountable but probably not economic from time and cost perspective to resolve.
With some rogue reading rejection coded in I can get pretty accurate stable readings. On change of weight it can take a couple of seconds to yield a stable reading.
The HX711 interface seems sensitive to timing, and when I tried to make the read routine a task (so that eventually I could run 4 on 1 processor) the stability goes off. I couldn't see that I would get 4 tasks reading reliably. so I could use 4 processors and amalgamate the readings but this is getting complex and expensive.
I could run 1 processor, and have it poll each of the 4 scales in turn. this would mean 10 or so seconds to get 4 stable readings. ok but not ideal. I might pursue this.
I have ordered a set of £20 200kg bathroom scales. They should be ok but might suffer from auto-turn off which makes using 4 on a race car anything from hassle to unworkable. they are due to arrive today so i'll see.
I can buy 200kg postal scales with lcd display on spiral cable for £40 but I have heard the cheaper end of the market is not that reliable.
With some rogue reading rejection coded in I can get pretty accurate stable readings. On change of weight it can take a couple of seconds to yield a stable reading.
The HX711 interface seems sensitive to timing, and when I tried to make the read routine a task (so that eventually I could run 4 on 1 processor) the stability goes off. I couldn't see that I would get 4 tasks reading reliably. so I could use 4 processors and amalgamate the readings but this is getting complex and expensive.
I could run 1 processor, and have it poll each of the 4 scales in turn. this would mean 10 or so seconds to get 4 stable readings. ok but not ideal. I might pursue this.
I have ordered a set of £20 200kg bathroom scales. They should be ok but might suffer from auto-turn off which makes using 4 on a race car anything from hassle to unworkable. they are due to arrive today so i'll see.
I can buy 200kg postal scales with lcd display on spiral cable for £40 but I have heard the cheaper end of the market is not that reliable.
Strain gauges and weighing cars
> ... auto-turn off...
That's my difficulty, too, with using a 4-cell bathroom scale.
Tom
That's my difficulty, too, with using a 4-cell bathroom scale.
Tom
Tom
yeah auto-turn off on these after 15 secs.
no mechanical scales at the low price point go heavy enough.
a mid-way solution might be 4 platforms, each with their own 328, output the result serially, and then write some vb to open 4 comm ports and display the results.
as its for occasional use, this ties up least hardware and keeps the cost down.
interestingly the platform of the scale I bought uses 4 x 3 wire load cells on a glass base, but, and given each corner would cost me £10 in sensors and hx711, its not looking too bad an idea t buy and use the platforms !
no mechanical scales at the low price point go heavy enough.
a mid-way solution might be 4 platforms, each with their own 328, output the result serially, and then write some vb to open 4 comm ports and display the results.
as its for occasional use, this ties up least hardware and keeps the cost down.
interestingly the platform of the scale I bought uses 4 x 3 wire load cells on a glass base, but, and given each corner would cost me £10 in sensors and hx711, its not looking too bad an idea t buy and use the platforms !
I'm working on a professional project using load cells for impact magnitude detection.
Pretty easy - just an op-amp (external or if the MCU has it, on-chip), then into an ADC. My round 1 was on AVR mega328, then mega1284, now moving to ST32F415 to get faster ADCs and CPU speed for math.
Lots of load cells in this project. Mix of 20lb and 200lb.
Pretty easy - just an op-amp (external or if the MCU has it, on-chip), then into an ADC. My round 1 was on AVR mega328, then mega1284, now moving to ST32F415 to get faster ADCs and CPU speed for math.
Lots of load cells in this project. Mix of 20lb and 200lb.
Strain gauges and weighing cars
> ... using load cells for impact magnitude detection. [] Pretty easy...
Link?
Tom
Link?
Tom
Tom
Hi,
I found this site with a system like your are looking.
http://www.forsentek.com/appdetail_57.html
I found this site with a system like your are looking.
http://www.forsentek.com/appdetail_57.html
Hi,
Another way you can do it is using 4 s-load cell like like this one Link:
http://www.ebay.com/itm/181333008712?_t ... EBIDX%3AIT
and 4 amplifiers like this link:
http://www.ebay.com/itm/161264280835?_t ... EBIDX%3AIT
and then use the zbasic micro to read the amplifiers / display each sensor weight.
These load cell can be use in compression/tension and the price are right for your application.
Another way you can do it is using 4 s-load cell like like this one Link:
http://www.ebay.com/itm/181333008712?_t ... EBIDX%3AIT
and 4 amplifiers like this link:
http://www.ebay.com/itm/161264280835?_t ... EBIDX%3AIT
and then use the zbasic micro to read the amplifiers / display each sensor weight.
These load cell can be use in compression/tension and the price are right for your application.
I've been using these http://www.ebay.com/itm/251616232403?_t ... EBIDX%3AIT
because they are cheap and low profile. if I am going to roll a car on and off that counts.
I have 4 connected to an hx711 very much along the lines described here
http://www.scottcutler.net/catpoop/catpoop.html
As that author found, the readings from the hx711 do contain some noise but I have filtered this out and have a prototype working ok
I now intend to make 1 full size wheel scale and test this. each corner will have a 4 load sensors, an hx711 and a 328l and this will output the weight serially. I'll then use VB .net on a laptop with 4 comm ports to display corner weights, total weight etc. this keeps the cost down and doesn't tie up too much hardware for something I won't use that often.
just waiting for deliveries at this stage.
because they are cheap and low profile. if I am going to roll a car on and off that counts.
I have 4 connected to an hx711 very much along the lines described here
http://www.scottcutler.net/catpoop/catpoop.html
As that author found, the readings from the hx711 do contain some noise but I have filtered this out and have a prototype working ok
I now intend to make 1 full size wheel scale and test this. each corner will have a 4 load sensors, an hx711 and a 328l and this will output the weight serially. I'll then use VB .net on a laptop with 4 comm ports to display corner weights, total weight etc. this keeps the cost down and doesn't tie up too much hardware for something I won't use that often.
just waiting for deliveries at this stage.