Hi all,

Is it a bad practice to amplify the input signals of an instrumentation amplifier?

My concern is this:

I need to use an instrumentation amplifier with small signals (uV). The input leads need to be at least 15cm long for the application, therefore not amplifying the signals straight from the source would introduce a lot of noise distortion.

I was about to amplify both of the inputs of the instrumentation amplifier streight from the source, but if I’m not mistaken, this is not recommenced and is considered as bad practice.

Any kind of suggestions would be highly appreciated.

 

Why not put the instrumentation amplifier at the source, where you were going to put some other amplifier?

Bob

 

I wouldn't call it "bad practice" but there are downsides to doing that, such as added noise from the amplifiers themselves and gain mismatches between the amplifiers effectively negating any benefit from the IA's common-mode rejection.

If you're having interference problems due to the distance between sensor and IA, and you can't relocate the IA to be close to the sensor, you can use filtering on the IA's inputs such as shown in Fig. 49 on p. 20 of this data sheet (for Analog Devices' AD8221 instrumentation amplifier) to reduce the interference.

 

hi,
I would agree with OBW, the tracking of two OPA's would be problematic.

E

 

Thanks for the prompt feedback.
@BobTPH because inputs A and B are not physically located close to each other.

 

@OBW0549 and @ericgibbs

I am opting in usig the INA331. Will using the LPF shown in the Analog Devices' datasheet effect the performance of this IA? Can this type of filter be used with any IA?

Thanks in advance.

 

I am opting in usig the INA331. Will using the LPF shown in the Analog Devices' datasheet effect the performance of this IA?

No.

Can this type of filter be used with any IA?

Yes.

 

That filter circuit will work with any IA but for best results use the best components:
I recommend 0.1% tolerance resistors and 1% tolerance for the smaller capacitors. The tolerance of the large capacitor is unimportant. If you use ceramic capacitors they should be C0G types to minimize issues with dielectric absorption. Polypropylene is a reasonable choice for larger values. PP caps will have to be through-hole since PP melts at reflow temp. Avoid polyester (mylar). There is one type of SM film cap that has good dielectric absorption properties - I can't remember which at the moment - might be polymethylpentene They seem to only be available in really sloppy tolerance, as I recall. The "best" through-hole film caps, polystyrene, are no longer made.
Any of the free simulation tools will do quite well for the filter network if you want to play with the effects of component tolerances on common mode rejection (you don't need the IA for the simulation, just a meter where the IA would go.

Because your inputs signals are physically separated you likely will not have equal common-mode noise for the two. This could be quite a problem and undo much of the good that an IA can do - it depends on frequencies of noise & signals. If your source impedance is high, you might improve performance by using remote unity gain buffers. DC offset is a potential problem you'll need to consider. If your bandwidth is not too high there are some very nice autozeroing op amps available at reasonable cost. You might actually do better with some remote gain, but once again use very good components - like Vishay bulk foil ultra-high precision very low temperature coefficient resistors that are several dollars each. I also recommend you investigate using shielded cables with driven shields for the inputs. It's a bit complicated but can be advantageous in certain circumstances. You may have quite a difficult task ahead.

 

Forgot to mention: Don't forget that there absolutely must be a path for the input bias current of the IA, even if it is mere picoamps, to flow, normally to amp circuit "common". You cannot put a fulling-floating signal across the IA inputs and have it work properly.

 

Forgot to mention: Don't forget that there absolutely must be a path for the input bias current of the IA, even if it is mere picoamps, to flow, normally to amp circuit "common". You cannot put a fulling-floating signal across the IA inputs and have it work properly.

ebp is certainly correct. Equal value resistors to the specified common point of the inst amp are required. 100K is a good value. Floating inputs will drift to the point of saturation of one or the other channels, giving incorrect outputs, and if you are lucky, not burning up anything.