Load Cell Output Signal Amplifier Design

Thread Starter

Czexican1329

Joined Apr 23, 2018
57
As for the bridge excitation, I'm open to any suggestions. I have a very stable 24V source and unstable 12V battery as voltage sources within my PLC. These are my only supplies that I have at my disposal within my PLC.
 

ericgibbs

Joined Jan 29, 2010
9,582
hi C,
Are you able to get the 10V reference suggested by Ron, if not can you at least get a 7812 regulator.
Also how many LM358's do you have available.?
E
 

Thread Starter

Czexican1329

Joined Apr 23, 2018
57
hi C,
Are you able to get the 10V reference suggested by Ron, if not can you at least get a 7812 regulator.
Also how many LM358's do you have available.?
E
Would making a voltage divider to step down the 24V to 10V be okay for the 10V reference? I believe I could get a 7812 regulator if not. And I have 10 LM358's on hand.
 

ericgibbs

Joined Jan 29, 2010
9,582
hi,
A quick question, is the 24V power supply used to power any other units on the project and is the 0v Common of the power supply Earth free, ie: not connected to Ground.?
 

ericgibbs

Joined Jan 29, 2010
9,582
hi,
I was going to suggest a split +/-12V supply, but as the PSU is used elsewhere that is not possible.

It will require getting a dual +/-12v PSU for the project, if you require a negative output.
Look at this draft idea using the 7812, [ which is no longer possible]
E
 

Attachments

danadak

Joined Mar 10, 2018
3,918
Looking at CM error using 1% Rs, only one R swept results in as much as 12 mV
offset. I think had I done all 4 R, by superposition that would result in 48 mV of error.
I then swept the + supply, got another 4 mV of error. These errors pretty much gone
if you use a factory IA (which are laser trimmed).


upload_2019-7-18_19-55-0.png


Regards, Dana.
 

TeeKay6

Joined Apr 20, 2019
572
Looking at CM error using 1% Rs, only one R swept results in as much as 12 mV
offset. I think had I done all 4 R, by superposition that would result in 48 mV of error.
I then swept the + supply, got another 4 mV of error. These errors pretty much gone
if you use a factory IA (which are laser trimmed).


View attachment 181984


Regards, Dana.
@danadak
Excellent! The input offset of the LM358 also affects performance, as does variation in the power supply voltages...and of course the tempco's of the resistors. So many sources of possible error!
 

Thread Starter

Czexican1329

Joined Apr 23, 2018
57
I talked to my supervisor, and he said that the 24V supply and 12V battery are all we can use due to the limitations of the rig I want to put this PLC on. There's no way around that. So I asked if instead of wiring the white wire W of the load cell to the negative analog input, I could wire ground to the negative analog input. My supervisor said that as long as my circuit works and provides the appropriate amplified output signal and I rewire the negative analog input to ground, he would be okay with that. So, here is my IA circuit design, with a load causing a 1.1mV difference between G and W and max load causing a 39.75mV difference between G and W. Since I would be measuring the LM358 output with respect to ground, I should read my desired amplified output voltages. Would this circuit be a possible solution or is there something else I'm still missing?
upload_2019-7-19_9-32-9.png
 

Attachments

ericgibbs

Joined Jan 29, 2010
9,582
hi,
I will run a simulation.
The low Vout of 33mV will not be possible due to using a single supply.
I would expect a lowest value of around 100mV to 200mV.
E
 

Reloadron

Joined Jan 15, 2015
5,148
Okay. I only have the one 24V supply sadly.
There are literally dozens of 10 volt reference chips out there like this one for just such purposes. A simple Google of 10 Volt Reference chips will give you plenty of results for chips which are designed for exactly your application. The stability of your excitation voltage is a key to both accuracy of the bridge and eliminating drift.

Next you want to use a good and highly stable IA (Instrumentation Amplifier, an IC designed around your intended application. I suggest you start by using a Google of instrumentation amplifiers. You can find good amplifiers designed for exactly what you wish to do with programmable gain. You can set the gain for whatever you want very simply and with a good IA low noise and distortion which is what you want along with little to no drift. They require a minimum of external parts.

You have a 5K Lb load cell with a full scale out of 3.0 mV/V so using a nice stable 10 Volt excitation reference with a 5K Lb load applied you get 30 mV out. You are using the load cell in a compression mode only so any good IA can easily be employed. Using a resistive divider as the excitation is a poor choice which will create problems with the design in the end result.

Every PLC I have worked with allowed programming so any analog input could easily be programmed to the engineering units of the users choice. I have no idea what PLC you are feeding or why you want a specific voltage in for a given weight (force) applied to your load cell(s) but with a simple programmable IA you can have whatever you wish. Additionally vendors like Omega who you bought the load cell(s) from even offer amplifiers designed for use with their load cells or any other load cell.

Look at the accuracy of your load cell or better said the overall uncertainty. I am sure you purchased it with the uncertainty in mind. Next look at how you amplify the load cell output and consider that uncertainty. Finally look at the combined algebraic uncertainty of your entire system. How good is it and how good does it need to be?

Ron
 

ericgibbs

Joined Jan 29, 2010
9,582
hi C,
This simulation is what LTS shows using your values.

As others have said, building a IA using LM358 OPA's and using general purpose resistors will give unpredictable results and performance.
I would recommend that you purchase a single supply IA, also a stable bridge excitation voltage source.

It really depends upon the accuracy and stability of your project, could you say what is the purpose of carrying out this measurement.?

E

BTW: Note the non linear response of Vout at low bridge signal levels, never goes to 0V.
[single OPA supply problem]
 

Attachments

Thread Starter

Czexican1329

Joined Apr 23, 2018
57
There are literally dozens of 10 volt reference chips out there like this one for just such purposes. A simple Google of 10 Volt Reference chips will give you plenty of results for chips which are designed for exactly your application. The stability of your excitation voltage is a key to both accuracy of the bridge and eliminating drift.

Next you want to use a good and highly stable IA (Instrumentation Amplifier, an IC designed around your intended application. I suggest you start by using a Google of instrumentation amplifiers. You can find good amplifiers designed for exactly what you wish to do with programmable gain. You can set the gain for whatever you want very simply and with a good IA low noise and distortion which is what you want along with little to no drift. They require a minimum of external parts.

You have a 5K Lb load cell with a full scale out of 3.0 mV/V so using a nice stable 10 Volt excitation reference with a 5K Lb load applied you get 30 mV out. You are using the load cell in a compression mode only so any good IA can easily be employed. Using a resistive divider as the excitation is a poor choice which will create problems with the design in the end result.

Every PLC I have worked with allowed programming so any analog input could easily be programmed to the engineering units of the users choice. I have no idea what PLC you are feeding or why you want a specific voltage in for a given weight (force) applied to your load cell(s) but with a simple programmable IA you can have whatever you wish. Additionally vendors like Omega who you bought the load cell(s) from even offer amplifiers designed for use with their load cells or any other load cell.

Look at the accuracy of your load cell or better said the overall uncertainty. I am sure you purchased it with the uncertainty in mind. Next look at how you amplify the load cell output and consider that uncertainty. Finally look at the combined algebraic uncertainty of your entire system. How good is it and how good does it need to be?

Ron
I had totally forgotten about voltage reference chips. I took your advice about the 10V reference chip for excitation of the load cell. I'm also looking at using the INA116 instrumentation amplifier (http://www.ti.com/lit/ds/symlink/ina116.pdf). The gain would be controlled by the value of an external resistor Rg. I calculated that my new maximum output signal from G of the load cell would be 30 mV. With this, my new desired gain value would be 41.666; using this value, I determined Rg should be a 1.3 kohm resistor, which means my gain would be reduced to 38.46 (which is still an acceptable gain for my purposes).
I also looked at using either the AD635JNZ or AD635KNZ IA instead of the INA116.
What are your thoughts on my latest idea?
upload_2019-7-22_12-12-47.png
 
Top