Hello guys,

I have a project that requires that I have a sensor network with 50 sensors that are to be converted to adc and preferably sent via wireless packets to a cloud. I must say that I am using a pre-built sensor for for measuring strain, I need to understand how I must proceed with purchasing the remaining units and as efficient and simple as possible. Via ADC, WiFi transmitter, cloud storage, etc. It's relatively simple I guess but could anyone kindly help me to decide the best possible approach to go about this project. It will be really helpful. Thank you

 

Great start, but you need to elaborate on many details for any kind of useful answer.

What is the part number of the strain sensor you are using?
Where are they mounted? How far apart?
How often do you need to take measurements?
How many bits of precision do you require?

Specifying clearly all the details is 80% of the way toward a solution.

 

What's the application?
What's your budget?
What's your capabilities with electronics/electrical?
Are you hoping to build a solution from scratch using discrete components or are you looking for a modular approach?

I'm going to suggest that you use a PLC to measure and record your data but that of course will require a large budget, a great deal of programming and engineering experience and a network infrastructure.

I could of course suggest the PIC/micro approach which would be far cheaper but would require an equal (if not greater) amount of programming and electronics experience.

 

Great start, but you need to elaborate on many details for any kind of useful answer.

Thank you kindly Sensacell. My answers in bold.

What is the part number of the strain sensor you are using?
Here is what i am using. It is a resistor type sensor. I want to measure the weight from the deformation of this sensor and this change is perhaps to be figured out from a wheatstone's bridge?

Where are they mounted? How far apart?
They are mounted with a seperation of about 10 by 10 cm each in a square matrix fashion.
How often do you need to take measurements?
I was hoping to take measurements at a rate of 1000Hz. i.e. 1000 times a second. Preferably each sensor would need to send its data a rate of 1/millisecond. Either sequently. I would be most greatful if there is already solution that is in the market as i would not want to reinvent the wheel.
How many bits of precision do you require?
Again, this upto 2 decimal points are sufficient enough. But my main concern is extracting a digital data derived from these 50 sensors. along with the amplifiers/ wheatstone bridges associated with it. which to be honest would like to know more about.

Kindly do help me. i would be most greatful if i am able to execute this in the fastest and already existing way as possible
Specifying clearly all the details is 80% of the way toward a solution.

 

Dear Buttonthief, my responses in bold

What's the application? I would like to measure the weight of cars directly on top of this network (I have a protection which has been already on top of the sensors(a strong polymer). Basically with about 50 sensors my application would be able to improve and get a better result.
What's your budget? As of now, this is of less concern and am able to manage it no matter the cost. Just would require a solution that is easy to implement (preferably already in the market)
What's your capabilities with electronics/electrical?
I have done some basics in electronics in bachelors, worked with arduino projects, and micro controllers.
Are you hoping to build a solution from scratch using discrete components or are you looking for a modular approach?
I would most prefer a solution that is already available. basically the only component i am not able to replace are the standard strain gauge sensors which change resistance upon compression or expansion. I need to amplify and transmit these change in resistances simultaneously from each of the sensor and send it wirelessly preferably .

I'm going to suggest that you use a PLC to measure and record your data but that of course will require a large budget, a great deal of programming and engineering experience and a network infrastructure.


I could of course suggest the PIC/micro approach which would be far cheaper but would require an equal (if not greater) amount of programming and electronics experience.

 

How much change in weight do you need to measure? 1 lb? What's the maximum and minimum car weight? The range. Do you want to measure 47 lbs?
Do you need axle weight? Or just total weight?
Why is 50 better than 1, 2 or 4 sensors?

 

How much change in weight do you need to measure? 1 lb?
Basically there is an algorithm that detects the changes. i wouldnt need to measure anything that is a pound, i would certainly like to measure heavy loads preferably from 350Kg to 15000kg range.

What's the maximum and minimum car weight?350Kg to 15000kg ( or atleast starting from 1000Kg) The range. Do you want to measure 47 lbs?
Do you need axle weight? Or just total weight? I would just require the total weight. But then again i just need to capture the changes strain that were amplified and transmit this information.
Why is 50 better than 1, 2 or 4 sensors?
The more sensors i have in a square matrix, i am able to recreate the changes strain while a car is in motion

 

Thank you kindly. So you want to measure all 50 sensor changes over a duration of time. Do you have a trigger to start and stop the process? Do the cars have similar speeds? This will help set the scanning rate.

I couldn't tell much about your sensor. From what I understand........if you can find a sensor that only reacts to the vertical force component......then the motion won't matter.

I thought a crystal was best for this. But I'm not up on the new stuff.

I haven't worked with processors for years, but one could multiplex the array, measure the sensors in sequence, on store the results.

There are some real experts on here. There are probably much better methods.

If you need it fast you will probably need a professional systems integrator.

Interesting project........I didn't realize they weighed moving airliners too now.

 

I think the easiest way (for me) would be to make a module for each sensor. The module would consist of analog for 1 sensor and a small MPU with integrated A/D and some form of communications, possibly RS-485.

Then, you place each module near its corresponding sensor.

Do this 50 times. Wire the comms together to a master controller.

Then, the master controller would handle capturing the data from the 50 modules and transferring it to the cloud.

The nice thing about a setup like this is it would be expandable to any number of sensors.

 

You could do other things too. When you read a sensor, you could compare it with a minimum reference.
If the sensor is not active......that is....doesn't have at least a certain weight......one can go to the next sensor.

Only the sensors that have a certain weight are measured and recorded.

 

Thank you kindly. So you want to measure all 50 sensor changes over a duration of time. Yes that is right. Do you have a trigger to start and stop the process? I have no trigger at the moment, and i was hoping that i would be able to use a trigger on this process. Do the cars have similar speeds? Ideally it should be at medium speeds at about 40-60km/hr. This will help set the scanning rate.

I couldn't tell much about your sensor.
From what I understand........if you can find a sensor that only reacts to the vertical force component
Right now basically any kind of reaction is sufficient. Basically i am trying to look for an acquisition system that is capable of amplifying 50 sensors simultaneously. Since each of the sensor would require to have its own wheatstone bridge i guess, its own amplifier probably, as well an adc system. after that i would want to use wifi to output it to some server or cloud, etc.
......then the motion won't matter.

I thought a crystal was best for this. But I'm not up on the new stuff.
Crystal does seem like a better idea, since i wouldnt need to use a wheatstone bridge with each of the 50 sensors. or if there exists such a device then i wouldnt mind using it.
I haven't worked with processors for years, but one could multiplex the array, measure the sensors in sequence, on store the results.
Could you kindly elaborate more kind sir

There are some real experts on here. There are probably much better methods.

If you need it fast you will probably need a professional systems integrator.

Interesting project........I didn't realize they weighed moving airliners too now
Not for airliners, its for cars and trucks.
.

 

Answers in bold

I think the easiest way (for me) would be to make a module for each sensor. The module would consist of analog for 1 sensor and a small MPU with integrated A/D and some form of communications, possibly RS-485.
Each module would have its own wheatstone bridge, amplifier, microcontroller for ADC, and a wifi? x50 and also for the serial transmission, can i send all at once or should there be 50 wifi connections running simultaneously? would that be a burden for the system?

Then, you place each module near its corresponding sensor.

Do this 50 times. Wire the comms together to a master controller.

Then, the master controller would handle capturing the data from the 50 modules and transferring it to the cloud.

I guess this should probably make sense, 50 comms, working to capture 50 signals at 1000Hz each. But would it be too much information to be sent along one channel? via wifi or etc?

The nice thing about a setup like this is it would be expandable to any number of sensors.

Answers In Bold

 

By writing your answers within the QUOTE box, you make it very difficult to reply to you. For each block of text you wish to reply to, please highlight the text and click "reply" in the menu that pops up.

Each module would have its own wheatstone bridge, amplifier, microcontroller for ADC, and a wifi?

Yes on bridge, amp and MCU. No on Wifi.

x50 and also for the serial transmission, can i send all at once or should there be 50 wifi connections running simultaneously? would that be a burden for the system?

No wifi on the modules. You would need to select a (hardwired) topology that supports the data rate and addressibility of the individual modules. The modules can be manually enumerated (i.e. via dipswitches or jumpers) or automatically (assuming the topology supports it).

You could use a single RS-485 (which is inherently multi-drop -- assuming it can handle the data rates), a daisy-chained SPI, or a custom 1 wire interface (that's probably what I would do, but it's more complicated). You could even make a "self-healing" cross-point network where each module routes automatically to the master MCU/CPU.

The master would be in charge of enumerating the various modules, syncing them, capturing and packaging the data, and retransmitting the data via [whatever] to the cloud.

I guess this should probably make sense, 50 comms, working to capture 50 signals at 1000Hz each. But would it be too much information to be sent along one channel? via wifi or etc?

What you want to do is well within the realm of possibility. What you may be capable of may be another story.