I am trying to use the Spice model for the AD8627, that I found on the AD website here:
The specification on the datasheet for this op amp is 0.5 fA/√Hz at 1 kHz, but this is not what I measure from the model. I am measuring the input current nosie following the method suggested here:
The method consists of placing a current-controlled voltage source in series with the input of the op amp, to be able to convert the op amp's current noise into a voltage noise that can be measured. When doing this, I measure 7.2 fA/√Hz at 1 kHz, instead of 0.5. I have attached a screenshot, because I could not upload the asc file for some reason.
Where does this inconsitency come from?
I've confirmed that the AD8627 Pspice model shows input current noise to be 7.2 fA/RtHz vs. 0.5 fA/RtHz typical in the datasheet.
I'm not sure what the reason for the difference in the model might be.
However, any source resistance interfacing to the AD8627 has to be very large (i.e. 320 Mohm) before the modeled noise current of the device exceeds the thermal noise of the resistor. So, hopefully even though the model predicts a higher noise current than the typical datasheet, most applications won't notice this as long as the source resistance does not go this high.
BTW, you could upload zip, pdf, and image (e.g. jpg) files onto Engineer Zone if you wish to share:
AD8627 EZ Noise Current 10_7_19.zip
Excellent, thanks for confirming this, and for the advice about when it might become significant.
To be honest, I have noticed these discrepancies on a few of these Analog Devices op amp models now - is it something that I would be able to correct myself, and could you point me to where this is defined in the Spice model file?
Thanks for you time.
Sorry, but I am not an expert to say why some op amp Pspice current noise models are not right-on vs. the datasheet typical value.
Regarding correcting for input referred noise current of any model which is not accurate:
I can think of a way to increase the modeled noise current by adding external noise currents to both op amp inputs. However, as in this case, when the modeled noise current is higher than it should be, I can't think of a way to account for that in the simulation so that the noise is lower than modeled! An alternative would be to reduce the excess noise algebraically. which is a bit cumbersome.
As an example, below is a simulation which artificially increases the device's input noise current to 10 pA/RtHz using external sources G1 and G2:
Creating 10pARtHz noise current 10_7_19.zip
Thanks for this. A useful trick to know.
Finally, what is the procedure for notifying Analog in the future about the problems with models? Do you have a modelling team I could contact, and who would then usually fix the models and update the website?
The best way to report an issue with a part's model (or anything else) would be to contact the Tech Support page: