Hello.
If you have an opamp/jfet which produce 1nV/hz and a resistor that produce 3nv/hz of noise before. Do you have 4nv of noise or 3nv of noise ?
Thank you.
If you have an opamp/jfet which produce 1nV/hz and a resistor that produce 3nv/hz of noise before. Do you have 4nv of noise or 3nv of noise ?
Thank you.
Thank you, because I read if you have big resistors no need to invest in good opamp.
Everything is relative I guess, it makes no difference 6nv or 4nv in relation to 60nv of resistor noise, so thank you !
Everything is relative I guess, it makes no difference 6nv or 4nv in relation to 60nv of resistor noise, so thank you !
Definitely wrong. Uncorrelated noise does not add this way. In this case
En = sqrt(1²+3²)
Noise power adds when the noise sources are independent/uncorrelated. Voltages only add if the noise ultimately comes from the same source ( for instance the voltage gain of an amplifier scales the input noise voltage ).
You add the powers of the input resistance Johnson noise, the opamp input voltage noise and the input current noise x input impedance.
equivalent input total voltage noise spectral density = sqrt (4kTRs + Vn^2 + |(InZs)^2|)
where Vn and In are opamp input noise density parameters, Rs and Zs are input resistance and impedance.
Typically In = sqrt(2qIb), where q = charge on electron, Ib = input bias current. Opamp's with input bias current cancellation schemes are more complicated though, often with much worse current noise in real circuits.
People love to ignore the current noise contribution, yet it can be the dominant source of noise in real circuits.
[ And of course this ignores 1/f noise and resistor excess noise ]
[ oh, and checkout my graph of computed noise for various opamps plotted against source resistance (assumes no input reactance) https://www.diyaudio.com/forums/analogue-source/21809-replacing-dead-phono-opamp-8.html#post6068897 - the point to note is that no device is ideal for a large range of input resistances, you have to tailor the device to the source ]
You don't mention the opamp's current noise - the input noise current is converted into a voltage by the source impedance, and that also contributes another uncorrelated noise signal.
You add the powers of the input resistance Johnson noise, the opamp input voltage noise and the input current noise x input impedance.
equivalent input total voltage noise spectral density = sqrt (4kTRs + Vn^2 + |(InZs)^2|)
where Vn and In are opamp input noise density parameters, Rs and Zs are input resistance and impedance.
Typically In = sqrt(2qIb), where q = charge on electron, Ib = input bias current. Opamp's with input bias current cancellation schemes are more complicated though, often with much worse current noise in real circuits.
People love to ignore the current noise contribution, yet it can be the dominant source of noise in real circuits.
[ And of course this ignores 1/f noise and resistor excess noise ]
[ oh, and checkout my graph of computed noise for various opamps plotted against source resistance (assumes no input reactance) https://www.diyaudio.com/forums/analogue-source/21809-replacing-dead-phono-opamp-8.html#post6068897 - the point to note is that no device is ideal for a large range of input resistances, you have to tailor the device to the source ]
Last edited:
There are a couple of issues with the question here:
- First, units. Spectral voltage noise is measured in nV/SQRT(Hz) not nV/Hz.
- Second, no circuit configuration is provided.
Assuming the resistor is a source resistor or a feedback resistor, then in general, the total noise is the RSS of the op-amp's voltage noise, the resistor voltage noise and the op-amp current noise x resistor value.
Now let's refine. Since the resistor is 3nV/rtHz, then the value is 562.5 ohms. Also, since the op-amp is JFET, then current noise is very low (~fa). Assuming 10fA/rtHz, the op-amp current noise is 5.6pV/rtHz, which is insignificant compared to resistor and op-amp voltage noise.
Therefore, the total noise ought to be:
Vnoise = SQRT(3nV^2 + 1nV^2) = 3.2nV/rtHz... which is very close to 3nV/rtHz.
You end up having to RMS sum a whole bunch of terms rather quickly in a practical circuit:
* Noise from external sources
* Resistor voltage noise
* Current noise x resistance
...all for both inverting and noninverting inputs. Resistance is the Thevenin equivalent seen at each input.
In practice it is generally more convenient to do this sort of thing in software. You can find various noise calculators for opamp circuits on the web, including mine (I've also got another for RMS adding random combinations of noise voltages both linear and dBV and dBu, voltage densities, and resistors, precisely for studies such as these).
* Noise from external sources
* Resistor voltage noise
* Current noise x resistance
...all for both inverting and noninverting inputs. Resistance is the Thevenin equivalent seen at each input.
In practice it is generally more convenient to do this sort of thing in software. You can find various noise calculators for opamp circuits on the web, including mine (I've also got another for RMS adding random combinations of noise voltages both linear and dBV and dBu, voltage densities, and resistors, precisely for studies such as these).
Hi Stephan, I am all for tools, but I think the problem here is more basic. I feel Federico here does not grasp the basic concepts of noise. I would suggest first:
- Read up on noise basics... Wiki is a start, but at least the English article, is meh.
After that, use a tool like yours.
Best, Sandro
- Read up on noise basics... Wiki is a start, but at least the English article, is meh.
After that, use a tool like yours.
Best, Sandro
A tangent and quite a challenge too, see if you can design circuitry with as few conventional resistors as possible.