LTSpice current noise calculation

[Guille]

Hi everyone,

I'm trying to calculate the input current noise for an opamp but the result doesn't agree with the datasheet: First I set a big resistor (10M) at the opamps input and run a noise simulation like so:







 Suppousedly most of the noise at the input should be due to the noise current running through the big resistor (6.9fA/Hz½ @ 1KHz according to the datasheet) and noise voltage contribution should be negligible (3.2nV/Hz½ @1KHz). The math checks out, you should get a total input noise at 1KHz of:

V_RTI = √( (6.9E-15 * 10E^6)² + (3.2E-9)² ) = 69nV

And finaly since those 3.2nV are negligible we could get the noise current by dividing our total input noise by our source resistor

i_noise=69nV / 10Meg = 6.9fA

The problem is running it on LTSPice gives a total input noise of 137μV instead, which accounts for a current of 13.7pA, more than 4 orders of magnitude above the datasheet!

The opamp model specifies in it's .cir file that current and voltage noise are appropiately simulated. I don't understand where the disparity stems from but I can't proceed optimizing the systems noise floor if the simulation is wrong in a way I don't udnerstand :/


Here's a link to the sim files since the forum won't let me attach them: https://drive.google.com/drive/folders/1zqDaC4hu1jqUYtOfWeSqcT3Ud-xvuTNG?usp=sharing

[Guille]

UPDATE: I simulated the opamp in open loop configuration and got a good reading, 12.6fA which is much closer to spec. Now the question is what's thowing off the measurement in my original design?

[magic]

This test circuit makes no sense, the opamp is wired with positive feedback so in the real world it would simply hit the positive or negative rail, depending on the polarity of its input offset voltage.

[iMo]

With magic's fix..

[magic]

Where did I say that?

This circuit doesn't make much sense either, because current flows mostly through the 2k6 feedback resistor. So you get mainly voltage noise.
The simplest sensible circuit is to ground IN+ through your "1Meg noiseless" and short IN- to OUT.

[iMo]

.. with 1Meg noiseless..

[Guille]

Wow, the fact that I checked the inputs and still managed to f### up says I should get my sorry ass back to uni. Many thanks guys! The mistake is clear now, however that last measurement (492nV dropping on a 1M noiseless resistor) implies the current out of the opamp is 492E-9 / 1E6 = 492E-15 or 492fA, there is still something off I think.

[iMo]

This fits the DS pretty well, imho..
It is the sim of the input noise spectral density vs source resistance (x axis, wired as in the DS) at 1kHz.