op amp feedback resistors value

[testian]

Hi,

what rules are there for selecting the right value for the feedback resistors (not the ratio)?

Lower value -> more stable, higher power consumption ?
Higher value -> more noise, lower power consumption?
Usually between 1k and 10k ?

[capt bullshot]

By chance, I came just across this one:

http://www.analog.com/media/en/analog-dialogue/raqs/raq-issue-135.pdf

[b_force]

lower value also means less noise of course
It sometimes also depends on the other components being used in the circuit, capacitors for example.

[LvW]

Hi,
what rules are there for selecting the right value for the feedback resistors (not the ratio)?

It is common practice to treat not only the opamps open-loop gain Aol as ideal (infinite) but also the input impedance (infinite) aswell as the outpiut impedance (zero). As a consequence, all gain calculations (magnitude and phase) contain errors which result from these simplifications.
However, these errors can be kept within acceptable limits if the simplifications are valid (allowed).
Regarding the utilized feedback resistors this means:
* The parallel combination of the feedback resistors must be SMALL (rule of thumb: at least factor 100) if compared with the known input impedance at the inverting opamp input node ; and
* The feedback resistor between the opamp output and the inv. input must be LARGE (rule of thumb: at least factor 100) if compared with the known output  impedance of the opamp (specified without feedback) .
* Hence, if possible we select resistors in the range 1kOhm... 100kOhm.

Hence, sometimes a trade-off is required, which might be influenced also by a requirement for a minimum input resistance of the whole circuit (in case of an inverting amplifier)

[Zero999]

I think more information is required to give a good answer here. What sort of op-amp is it? How is it being used?

It isn't just the input impedance which determines the range of the input resistors but the bias currents. Generally speaking, the higher the bias current, the lower the input and feedback resistors need to be. If the op-amp doesn't use bias current compensation and doesn't have rail-to-rail inputs, then matching the input resistance of both inputs can minimise the offset error, due to the bias current.

More often than not,  it's what's connected to the amplifier's input which determines the input resistor values and type of op-amp used. For example, a preamplifier for a low impedance dynamic microphone (under 600 Ohm) is better with a BJT op-amp such as the NE5532 and low input resistors and feedback resistors, <10k and <100k, respectively. A preamplifier for a high impedance piezoelectric microphone will be better served with a FET input op-amp such as the TL072 and high value input and feedback resistors, >1M and >10M respectively.

[basinstreetdesign]

The value of the feedback resistor depends on several things.  It must satisfy all of those things simultaneously.
It must not be any larger than the SMALLEST upper-bound value from several issues and not be any smaller than the LARGEST lower-bound value from several other issues.

For upper bound values:

Most general purpose op-amps will work fine with Rf anywhere up to several megohms and down to zero. 
However, Rf should not be so high that the amplifiers' input bias current drawn through it causes excessive offset voltage which cannot be equalized with another resistor on the other input or by some other way.  As a general case, anywhere from 1K to 1 Meg will work fine for amps such as LM741's, TL072/82/ series, etc.

The speed of the op-amp can also be a factor. Usually the makers of video-speed op-amps or with GBW greater than 50 MHz will recommend Rf about 620 Ohms to 1K max.  If that is your case, assume that the stray capacitance around the inverting input is about 1-2 pf.  Dividing 2*pi*GBW by 2pf will give you an upper bound to Rf.  For example a chip with a GBW of 100 MHz will not be happy with Rf much greater than about 800 Ohms or else the pole created by that 2pf cap and the feedback resistor will cause instability near the GBW figure, 100 MHz.  It can be equalized with a small cap across the resistance from the inverting input to AC ground so that that time constant equals the first one.

The feedback impedance should not be so high at any frequency that the op-amp doesn't get enough feedback signal to stay stable.  Dropping a small cap across Rf can keep the amp stable at high frequencies at the expense of losing some bandwidth.  This may be perfectly fine if you are not concerned with high frequencies such as with a sensor amp.

On the other hand:
The datasheet or app-note may say that Rf cannot be lower than a certain value or else the amp will get unstable and oscillate.
Perhaps the chip must never have a gain less than a certain minimum for stability.  This will also be given in the datasheet.

You shouldn't use the Rf as a load resistor if it means that the amp has to supply more current than it can easily handle without heating both the amp and Rf.  If you need a lot of current stick a current boosting transistor on its output.

This certainly is not all of the issues which may bear on the subject.
Hope this helps

[danadak]

Feedback Rs (value and technology/type) affect -

1) Noise
2) AC response
3) Loading, both input and output
4) Stability, phase margin
5) Drift
6) Offset
7) Crosstalk
Power consumption

Just a few issues.


Regards, Dana.