Hello, i am new to hardware circuit design. This can be a very simple question for many. I have a strain gauge setup whose output has to be measured in voltage and to be sent to a microcontroller unit (outside machine) out of coolant environment inside the machine. Hence i am looking for a way to convert the measured signals to a form which can be transfered through serial communication, so that voltage drop due to resistance in lengthy cables can be reduced. Can you pls suggest a set of hardware required for this conversion and pros and cons if possible

 

I had a somewhat similar setup. I had a thermocouple, it measures themperature, but the output is a few millivolts. What I did is I found a small board that was designed to work with thermocouple, this board had ic that digitized the thermocouple signal and then sent it to uC using SPI.

So the question is. Is there a product that already exists that works with your strain gauge?

If the output of strain gauge is a few volts, you can just get one of those little uC boards (Arduino Nano, Picaxe, or similar, something really small) and use onboard ADC to digitize the gauge output and use onboard uart to connect to another uC that will act as a main control.

 

How long is a lengthy cable?

John

 

Check out Picmicro weight scale MCP3421.
Max.

 

I think that in this age of electronic marvels, the way to do this is first an instrumentation amplifier, then a microcontroller which will digitize the amplifier's output, and create a serial stream to be fed to the external equipment. "Voltage drop due to resistance in lengthy cables" really isn't an issue, as the current needed to carry information is very small. Other transmission-line effects could occur though; do you think you'll need differential data transmission?

 

Adding to a question about the line length.
What is the precision you want to have?
What data rate is needed?
Will it be a unique device or you want to have a series?
The specifications define design solutions.

 

How long is a lengthy cable?

John

It would be more than 3m in length. The main reason is the strain gauges will be near coolant environment and so i wanted to keep my µC away from it. So initially A-D conversion and then sending to µC as serial data

 

I think that in this age of electronic marvels, the way to do this is first an instrumentation amplifier, then a microcontroller which will digitize the amplifier's output, and create a serial stream to be fed to the external equipment. "Voltage drop due to resistance in lengthy cables" really isn't an issue, as the current needed to carry information is very small. Other transmission-line effects could occur though; do you think you'll need differential data transmission?

i have not decided about the data transmission yet. I do have two output from strain gauges and hence need 2-channel ADC before converting to serial data

 

The main reason is the strain gauges will be near coolant environment and so i wanted to keep my µC away from it. So initially A-D conversion and then sending to µC as serial data

This coolant environment, what is the expected temperature? If it is too cold for a µC it might be too cold for other electronic devices as well.

 

Adding to a question about the line length.
What is the precision you want to have?
What data rate is needed?
Will it be a unique device or you want to have a series?
The specifications define design solutions.

I am planning to use a 12-bit ADC, and would like to have 1mV resolution.and data rate is more than 100kS/s. This will be connected to a microcontroller and microcontroller will be connected to PC

 

Do you have the datasheet for the unit strain gauge sensor you are using?

 

I am planning to use a 12-bit ADC, and would like to have 1mV resolution.and data rate is more than 100kS/s.

Usual load cell has about 2 mV/V and excitation is from 3 to 12 V. Hence the output signal is 6 to 20 mV. What does your 1 mV resolution mean? Or do you have some specific load cell?

Data rate 100 ks/s - it is period of 10 us. Do you mean the clock of serial communication? Or do you want to have a new sample every 10 us?
If the latter is right, than you have to pass samples to a MCU with a rather fast line, minimal clock being somewhere in MHz region. I think SPI has some advantage. So you can take a SAR ADC with SPI.

 

Usual load cell has about 2 mV/V and excitation is from 3 to 12 V. Hence the output signal is 6 to 20 mV. What does your 1 mV resolution mean? Or do you have some specific load cell?

Data rate 100 ks/s - it is period of 10 us. Do you mean the clock of serial communication? Or do you want to have a new sample every 10 us?
If the latter is right, than you have to pass samples to a MCU with a rather fast line, minimal clock being somewhere in MHz region. I think SPI has some advantage. So you can take a SAR ADC with SPI.

If the sensor output is in mV range, than like John P said, OP will need instrumentation amplifier, then ADC. ADC with some sort of serial communication feature would be perfect.

 

OP will need instrumentation amplifier, then ADC. ADC with some sort of serial communication feature would be perfect.

Why do we have to play guess game? Let the topic setter clear the issue.

BTW, 3 m is so short distance that all the electronics may be moved outside the site. Just a load cell, cable - and after that we may use whatever we find to measure the signal.

 

http://www.microchip.com/wwwproducts/en/MCP3421
The app note in #4 also has code.
http://ww1.microchip.com/downloads/en/DeviceDoc/51918A.pdf
Max.

 

#15.Those have 18bits resolution!

 

#15.Those have 18bits resolution!

Programmable!
Max.

 

Usual load cell has about 2 mV/V and excitation is from 3 to 12 V. Hence the output signal is 6 to 20 mV. What does your 1 mV resolution mean? Or do you have some specific load cell?

Data rate 100 ks/s - it is period of 10 us. Do you mean the clock of serial communication? Or do you want to have a new sample every 10 us?
If the latter is right, than you have to pass samples to a MCU with a rather fast line, minimal clock being somewhere in MHz region. I think SPI has some advantage. So you can take a SAR ADC with SPI.

I do not have any commercial load cells. it is a film made of gold and chromium deposited on metal surface and acts like strain gauge. this is measured with wheatstone bridge (preferably quarter bridge). I have only the initial resistance value of the film and after further discussion with superiors, the sampling rate is not a constraint and can be very low also.

hence i am thinking of connecting two strain gauges to microcontroller after passing them through filters and instrumentation amplifiers and establish continuous data transfer between microcontroller and PC using serial port (either USB or SPI) Is my approach is right ?

 

Why do we have to play guess game? Let the topic setter clear the issue.

BTW, 3 m is so short distance that all the electronics may be moved outside the site. Just a load cell, cable - and after that we may use whatever we find to measure the signal.

If i understand you correctly, all the circuits except the strain gauge, bridge circuit and amplifier needs to be near the site and others can be moved outside ?

 

More of it. An amplifier mat be moved to a distance too. In old times there were 20 m cables from the last electronic component to a load cell (strain gauge). Now it is possible to move strain gauge ADC closer - but your 3 m is not a distance at all. Only 4 wires (2 to feed a bridge, 2 - signal), just make 2 twisted pairs from a usual isolated wire. I do it every day