Amplify strain gauge output – help

Dear all,

I need to amplify the signal output of a strain gauge. I did not find anything that I can just 'plug and play' or maybe I didn't look in the appropriate place so I am trying to design one myself. Problem is I am no electrical engineer, I specialise in something else. I actually came up with a design after some research and I wanted to confirm if this would do the job.

Anyway, attached is the design I think I am supposed to use, but I couldn't verify it with a circuit simulator. So clearly there is something wrong somewhere, be it design or software.
 







Notice that there is a Wheatstone bridge where only one resistor is the strain gauge, of nominal 120 Ohms. The bridge outputs that I am trying to read are around 0.01mV and the sensitivity of my logger is 1mV, so I need to amplify the signal at least 100 times, but 1000 is well welcomed! Note that on the circuit the resistances R1 and R0 and the resistances of R2 and R3 are relevant to an amplifier of 10 gain, according to the formula Vout=(Vin1-Vin2)(R2/R1).

The circuit drawn is for a differential amplifier, I think that's what I need?

Also, would this setup work whether the bridge output is positive or negative?

Anyway, please ask if I left any info out!

Juan

Load cell amplifiers are readily available, just do a web search on the key words.
Far better to buy than build, but you need to check the specs carefully all the same.
I applaud your effort , and your basic idea is OK, but the detail is important, and there are far better ways of doing it. You don't say what resolution, full scale output, and operating temperature is expected from your system, these need to be considered in the design. Is the 10uV figure given full scale, or the resolution.
Additionally, you need three other foils in the load cell with the same electrical parameters that are passive mechanically, to ensure that your drift from the load cell is minimised. There is also a linearity issue with a single active foil.
See this article from TI, http://www.ti.com/lit/ml/slyp163/slyp163.pdf
What you have to consider is that you don't know if the output of the amplifier is due to a change in the load cell resistance, or due to drift in the amplifier, which is a critical issue with the sort of gain you require.
Resistor matching and temp coefficients are important, as are the drift characteristics of the op amp.
You have to do an error budget, working out from the specs for the resistors and amplifier to figure out what drift you can expect, and it can be far greater than the bridge output if you don't take care of it.
There are far better ways to design a load cell (or instrumentation) amplifier, I would refer you to an excellent article by Analog Devices on analog load cell, amps. http://www.cypress.com/file/51291/download
I have both bought and designed and built strain gauge amplifiers over the years, these days I would buy analogue, but I still design digital using sigma delta converters, which connect directly to the load cell.
Spark Fun have a digital system,
https://learn.sparkfun.com/tutorials/load-cell-amplifier-hx711-breakout-hookup-guide.
Cheers,
Richard

Hello Richard,

Thanks for the info, I am going through the reading material at the moment.

The main reason I want to design it myself is that I only have one gauge installed on the specimen, so I would need to complete the bridge with either 3 gauges or 3 resistors, as all the load cell amplifiers I found are for full bridges. So I thought that if I need to create a board with the additional 3 resistors, I might as well get the amp on the same board.

Given the loads I am measuring, the output in the bridge will _vary_ between 0.01mV and 0.50mV (this is zeroing an offset of course). But the resolution of the data logger (which is an EVO4 from Aim Technologies) is 1mV. So if I amplify 100 times then it would, in theory, be able to detect signal changes, as I would get a variations of 1mV-50mV.
I think what you mean is that my system needs a resolution of 0.01mV, whilst the output scale would be 1-50mV for 100* (or 10-500mV for 1000*, which is preferred).

I am not sure if working temperatures are (to my inexperienced self) a concern, since the amplifiers would be sitting on the wishbones of an open wheel race car and not directly touching the metal, so the temperature shouldn't be affected by external sources. Of course, this circuit might generate lots of concerning heat, that's something I do not know.
In any case the rear dampers sit on the gearbox-engine, and they get up to 50 celsius. So I doubt the wishbones will get up to that temperature, knowing that they are under a constant stream of fresh air.

Again, thanks for your reply.

Kind regards
Juan

Hi,
I would be curious what op-amps people are using for this type of application?

I do some current sensing on low value resistors and the signals get pretty low. I have been using the LMP2011 part with 60uv offset worst case, 1uv typical.

I use these in some power supply power controllers. Early LMP2011 parts had some "Pull Low" issues so I biased things up some. I have schamtics at www.sunduino.com for examples.

Bob K.

You should be concerned about change in temperature. I suggest you make some temperature measurement tests in situ while driving to see what it is likely to be over the time of the measurements.
Consider that standard resistors vary by up to plus or minus 200ppm/degree C, if you build the design you came up with initially, you have to use precision resistor arrays, which have closely tracking temperature coefficients, or the measurement will be lost in thermal effects.
In the case of an op amp, consider a 10uV/degree C offset coefficient.
Load cells are designed to compensate for temperature changes by matching the foil as closely as possible to the base material, steel or aluminium have different coefficients of expansion, and the foils are closely matched. That is why it is best to use dummy foils.
If you use an instrumentation amp, they have all of the resistors except the gain setting resistor built in. The gain set resistor still has to have a low temp co.
Cheers,
Richard

Hi Bob,
I would think that an auto zeroed op amp would be ideal if you need to monitor current in a linear fashion over a very wide range of current. If you just need to detect max current for shut down purposes, with a reasonable tolerance, then it is probably over kill.
Additionally, the temp co of the sense resistor will be the biggest contributor to error, I would think, ambient and self heating will both influence it greatly.
TI, Linear Tech, Maxim, and others make devices specifically for monitoring current in high side sense resistors, one from TI even has a low temp co resistor embedded in the IC, some of these may be more appropriate, depends on the error budget , board space, and the economics.
For load cell apps, there are better ways of getting stable gain than using individual op amps with external gain setting resistors, I would recommend an Instrumentation Amplifier, there are also auto zeroed types available. You have to watch the switching noise in some of them, limit the output bandwidth with filtering typically.
Cheers,
Richard