LMH6881: Input referred noise condition

Hello Taketo-san,

I understand your question better now.  I would like to add clarity onto why there is a single input-referred voltage noise specification, even though there are two input modes.

Please refer to Section 7.2 diagram, which I will copy below at the end of my message.  In this image, there are four resistors present on the input network for the LMH6881.  This symbolizes the input impedance setup, which is described in the datasheet as 100 Ohms differential or 50 Ohms each for the single-ended inputs.  

The noise calculation does not take the single-ended or differential nature of the input into account.  The input impedance is either 100 Ohms differential (with 50 Ohm resistors floating) or 50 Ohms + 50 Ohms for each single-ended input (with 100 Ohms differential floating).  Since the resulting impedance is the same, the voltage noise specification will apply to either input setup.

Thank you for your great question!

Best,

Alec

Hello Taketo-san,

I have observed a small error with the calculation you provided:

When considering 26dB of gain represented in V/V, the conversion is 10^(26dB / 20) which is approximately 20V/V, or 19.95 V/V more precisely.  If we continue onward with the remainder of the calculation, the 2.3nV/sqrt(Hz) value multiplied by the gain (19.95) yields the expected output voltage noise of ~46nV/sqrt(Hz).  This slight difference is due to rounding error in my calculation; however there can be confident in the noise analysis of the LMH6881 you have performed so far.

Please wait for an email response from me regarding your additional question over email.

Best,

Alec 

For the forum: Please see a snippet of the resolution conversation below:

"""

Hello Sato-san,

1. This is a differential noise specification. That is correct, it would be noise seen by Vop. 
   
   
2. Typical application for characterization follows Figure 37. The loading is correct, but the input sources should be set to either 50ohms internal or via representation by series 50 ohm resistors. The Tina-TI simulation below shows output voltage noise at 47nV/√Hz at maximum gain below 1MHz.     
   
   
     
   
   

         LMH6881 Model Notes from netlist are below:

This could be a LTSpice issue if the setup matches the Tina-TI circuit shown above. We have run into issues with customer's obtaining different results in LTSpice compared to Tina or PSpice with our other models. 

LMH6881 Noise.tsc

Thank you,
Sima

"""

I will update this thread with any additional information relevant to analyzing LMH6881 noise.

Best,

Alec

Here is the final piece of information from the discussion, illustrating how to reconfigure SPICE model to model output voltage spectral density appropriately:

 The Noise simulation in the reference circuit is not the correct output for output voltage noise spectral density. The two images below show the circuit used (removed DC blocking input caps), and the output voltage noise trend which is the same as shown in Tina-TI. 

   To setup the simulation profile, make a new simulation profile with the below settings under AC Sweep/Noise:

   Then press the run button. The simulation will have no trends on the plot. Right click anywhere on the plot and select add new trace. Below are the equations to set for either noise profile. The one on the right is the non-integrated sqrt(Hz) noise value shown in the datasheet.

This work-through is done in PSPICE.

Best,

Alec