Which instrumentation amplifier should I use?

Hi Nathan,

I think the biggest hurdle you're going to need to overcome is the tradeoff between gain and bandwidth. Without knowing the rise time of your pulse I can't get an exact number for the necessary bandwidth, however it will need to be well above 33kHz (1/pulse width).  Taking a look at the datasheet for the INA118 you should see that the bandwidth of the amplifier will not be adequate in a gain of 1000:

Choosing a wider bandwidth instrumentation amplifier that uses a current feedback topology (INA111) improves the outlook, however the bandwidth in a gain of 1000 will still be insufficient to accurately reproduce a pulse:

What I recommend is that you reduce the gain requirements of the instrumentation amplifier and then follow it with an operational amplifier. For example, if you configured the INA111 for a gain of 100, and then followed it with an opamp configured for a gain of 10, then your overall gain would still be 1000, but you would have a much wider system bandwidth (assuming that you select an opamp with sufficient bandwidth in a gain of 10).

Other factors that you should consider are the source impedance (e.g. what is the impedance of my sensor?) and noise requirements. All instrumentation amplifiers draw some amount of input bias current which will flow through your sensor and may produce a DC offset at the output. For high-impedance sensors, FET input instrumentation amplifiers will be preferable (INA110, 111, 116), where low impedance sensors are less effected by input bias current errors.

Nathan,

Two additional parts to consider are the INA163 or INA166. These parts are intended to be used with low impedance transducers, such as microphones, and therefore sacrifice input bias current for lower noise and wider bandwidth. The INA166, has a fixed gain of 2000 and a bandwidth of 450 kHz.

John,

Thank you very much for this insight.  If I wanted to use an op-amp (for a gain of 10) instead of an instrumentation amplifier to perform the same tasks as above which one would be suitable?  The other specification is that the power supply that I'm using is a single supply not a split supply.  Do I need to use  decoupling capacitors?

Nathan;

I agree with the previous posts, breaking up the gain into two stages will allow you to achieve greater bandwidth so that the transient response will be OK for your application. An INA163 is a good choice if you have dual supplies but it is not too well suited for general use on single supplies.

If you have a +12V or +15V single supply, try an INA826 in a gain of 10V/V followed by a THS4221 in a gain of 100V/V. Be careful to observe the common mode input voltage requirements (see the data sheet).

An alternative, operating on a single +5V supply might be an INA331 with a gain of 25V/V followed by an OPA365 in a gain of 40V/V. See Figure 2 in the INA331 data sheet for information regarding its common mode voltage, etc.

High gain bandwidth circuits like this MUST be carefully laid out on a PCB using a good ground plane. Attempting to build a circuit on proto board is guaranteed to oscillate.

Yes, Nathan, you ALWAYS need decoupling (bypass) capacitors.

I have procured a dual power supply (since I have found out the hard way designing amp circuits with a single supply is significantly more difficult) and I think that I will go with the first suggestion of configuring a INA111 instrumentation amp with an op amp.  A few more questions before I begin buying/building. 

1. Is there a minimum voltage that the INA111 must receive before it begins to amplify?  The voltages I expect to be amplifying are going to range from 0 to 15 mV.

2. Would it be better to configure the INA111 with another INA111 to get my required gain instead of using the instrumentation amplifier configured with an op amp?

3. Does it matter the order i amplify the signal?  i.e. instrumentation with a gain of 100 then an op amp with a gain of 10 versus an op amp with a gain of 10 then an instrumentation amp with a gain of 100 versus two instrumentation amps with 100 and 10?

Thanks for your prompt replies,

Nathan

Nathan;

If more engineers were as wise as you are regarding dual supplies, they would encounter far fewer problems.

1. The INA111, as well as other amplifiers, amplify input signals all the way down to zero but the limit to seeing the extremely low signal levels is the noise.

2. No, an OPA111 followed by an op amp is a better choice. An IA is a differential input device and the first IA has only a single-ended output so it is simpler to use an op amp-- its less expensive and less complex as well.

3. As usual, it is a tradeoff. Higher gain in the IA gives better CMR, PSR, DC offset, and drift. but higher gain gives lower bandwidth. A gain of 100V/V is a good choice for the IA.; place the op amp in a gain necessary to achieve your desired overall gain.