Hi,
I'm trying to make an amplifying circuit and I'm sort of without a clue on what OpAmp to use.
I'm amplifying the result from one half-bridge from a Wheatstone bridge. The input is 5V so the output should be 2.5V + the signal I want to amplify.
In a first step I want to cancel the 2.5V by putting the 2.5V+Signal into the non-inverting input of the OpAmp and I put 5V in the inverting end with a large resistance between the 5V and the Opamp as well as a large resistance between the OpAmp inverting Input and the Output. This should let just the signal pass. (basically an inverting amplifier with the ground at 2.5V+Signal)

Then I want to amplify that signal with a standard non-inverting amplifier. The signal can be up to 50mV.

I'm finding a lot of OpAmps that say max Input Offset 6mV, 10mV, etc. Can I not use those for the 50mV signal with a non-inverting Amplifier? Not to mention the 2.5V vs 5V. Is there an OpAmp that can do this?

I want to amplify that signal 200 times to get to 10V. I take it any OpAmp can handle that... what about 10'000 times? I have another signal that is as low as 0.375 uV at the sensitivity I want so i need to amplify that 10'000 times to get it to 3.75mV which is what I can reliably read.

So basically I need 3 OpAmps. One that can take 2.5V input difference without amplification, one that can take 50mV difference and amplify 200 times, and one that can take 0.375 uV and amplify 10'000 times.

Any tips?

 

Your first stage should be an instrumentation amplifier(IA) read more here http://www.intersil.com/data/an/an1298.pdf
After that more amplifier and filter stages using standard opamps should be implemented. One more thing. Do NOT try to make your own IA. Use a commercial one. This because the resistors used in the IA. Must have a very good match. Even 1% resistors will cause unwanted errors

 

Look at the AD623 instrumentation amp from Analog Devices. Very easy to use. But if you have only a single voltage (from a half bridge) you'll need a 2.5V level to match it against. Most likely you'll need precision resistors to generate this.

 

Your first stage should be an instrumentation amplifier(IA) read more here http://www.intersil.com/data/an/an1298.pdf
After that more amplifier and filter stages using standard opamps should be implemented. One more thing. Do NOT try to make your own IA. Use a commercial one. This because the resistors used in the IA. Must have a very good match. Even 1% resistors will cause unwanted errors

True, thanks. After the IA I should have just the signal. Any suggestions for OpAmps after? Or is 10k amplification no problem with all OpAmps?
I'm a student and haven't often had to purchase OpAmps so far. Usually they're just handed to you for an exercise... So I'm kind of lost when a search on Farnell.com turns up 5000 choices

Look at the AD623 instrumentation amp from Analog Devices. Very easy to use. But if you have only a single voltage (from a half bridge) you'll need a 2.5V level to match it against. Most likely you'll need precision resistors to generate this.

Thanks. I was going to use a potentiometer between two resistors because I'll have to compensate for a half-bridge offset anyway and not just the 2.5V. The resistors in the wheatstone bridge are anything but precision(between 3050 Ohm and 3450 Ohm. There are multiple bridges I need to amplify). There's also an offset from some other source I can't find because the resistor differences only account for about 60% of it. So I'm planning to tune the voltage so I get exactly 0V after the IA after I finished assembling the whole circuit.
Do you know anything I need to look out for when shopping for potentiometers? Any sign to recognize more stable ones? I need to put it on a circuit board, making it probably a lot more difficult to find an appropriate one.

The data sheet I found on the AD623 says it can only handle input voltage offsets of 500 uV.
Page 3 under http://www.farnell.com/datasheets/82242.pdf
This prevents it from being used with my 50mV signals. Or am I misinterpreting terms?

 

In this case "offset voltage" is a maximum error specification. It means that if you connected the two inputs to the same voltage, the output would act as though the difference between the inputs was actually the stated amount. So 500uV offset at maximum gain (1000) would give you an output error of 0.5V.

I hope these non-precision resistors are stable. If they vary randomly (thermal effects) you'll go crazy trying to get an accurate reading.

An instrumentation amplifier has a terminal for input of an offset voltage so you can steer the output to whatever quiescent level you want. On the AD623, it's pin 5.

 

I was hoping it was something like that! Thanks for clearing that up.
Is this Offset usually relatively fixed, due to production? (changes with temperature of course) or is it random between 0 and 500uV every time you turn it on?

I'll completely lose my sanity by the end of the project then... But since the whole thing should be packaged in the end and they're attached to a large piece of steel acting as a heat sink, I hope it'll be relatively stable in a controlled environment.

 

Using a standard opamps after the IA is the correct thing to do. You could create a single gain stage with gain equal to 10000. But that will not be very smart. Due to something named Gain-bandwidth product http://en.wikipedia.org/wiki/Gain-bandwidth_product Use several stages. Remember most IAs also have some way of selecting the gain.
You will probably see that your offset voltage vary from day to day. So you should implement some zeroing arrangement as part of the instrument. The simplest way to do this, is as software function. Of course this will require AD conversion. Some opamp that I think will work well in your project may be OPA 277/2277/4277. And also LT1014. Remember to order some spare as you will fry some parts.

 

Ok, thanks for the link, I'll watch out for that.
I thought of using the IA for the 200 gain then I only have to amplify the exits I need at 10k by 50. That also makes the resistor errors a bit easier to handle.
AD Converter is intended so that shouldn't be a problem.

I hope I don't fry any...

 

There are some good opamp primers on this site ...look at VolIII / opamps.
(At the top)
They cover all the basic configurations and explain how and why things work.
Al

 

I suggest that before building anything, you do an error budget. Set up a spreadsheet and calculate successive cells based on the stages in your circuit, with the possibility of introducing errors up to the worst cases you expect to see. Some of these will be static and can be nulled out, but others are unavoidable noise. Will your signal to noise ratio be enough to get useful data out of this design?

When you talk about resistances with uncontrolled values, and a high gain, I don't get a good feeling. But if you can get numbers that show a usable result, then you can feel some confidence.