When are op amp buffers required?

[Kerry]

In reading about the Dimension Engineering accelerometer break-out boards, I noticed that they sport a "buffered" signal - so I started reading up on op amp buffers.  I think I understand the general concepts, why they are used (high input impedance, low output impedance), but I still don't understand why you need something with high input impedance and low output impedance.  I would understand if we were talking about driving a motor, etc., but it is unclear why a sensor would require this.

Does anyone know what kinds of problems can occur if you were to use a sensor without a buffer, and how you know if you'll need one?

Thanks!

-Kerry

[Soeren]

Hi,

[...] why you need something with high input impedance and low output impedance.  I would understand if we were talking about driving a motor, etc., but it is unclear why a sensor would require this.

Same problem really.
You need the high input impedance to avoid loading down your sensor (if the sensor doesn't have a low impedance output by itself) and thereby getting false results.
You need the low output impedance to drive the next stage without being loaded down.

Does anyone know what kinds of problems can occur if you were to use a sensor without a buffer, and how you know if you'll need one?

If you need one, without it false results will be the problem, as mentioned above.
To see if you need it, you'll need to know the output impedance (Z) of your sensor and the input impedance of the input of the next stage - find it in the datasheets.

For low precision work, Z_out of the sensor should be at least 10 times lower than Z_in of the next stage to connect them up without a buffer. For precision work, a factor of at least 100 (or even 1000) should be used.

[Kerry]

Soren, thanks so much for your explanation.  All of this is starting to come together (slowly...).  I appreciate your help very much.

-Kerry