Post Go back to editing

# ADA4084-2 Spice Model Weird Noise Behavior

I am trying to design a ADA4084-2 into my latest design, but I am receiving strange results when I do noise modeling using the Spice circuit.  The voltage noise at the output is extremely large, even with relatively small feedback resistor values.  It almost seems like an improper value for the current noise has been input in the spice model.

Basically, the problem can be produced by just putting a 5k feedback resistor (from output to inverting input), while holding the non-inverting input at a constant voltage (-2 V in my case).  Power supply rails are +/- 5V.

In this configuration, the resulting predicted voltage noise at the output is 600 nV/sqrt(Hz) ! (~10 Hz to 1 MHz)

If I reduce the feedback resistor to 1 Ohm, I get the much more reasonable value of 4 nV/sqrt(Hz)

I have also performed the same simulation with other parts (AD4528, AD4004) and those behave as expected for all values of the feedback resistor.

I really want to use this part, and the noise values in the datasheet are completely acceptable.  I just want to make sure that I am not missing something dramatic.

Dustin

Dustin,

A couple of things.

--  Try running on +/-15V, with Vcm =0V.

Should be about 550 fA/rt-Hz at 1 kHz.

If that doesn't work, are you running LTSpice?  LTSpice sometimesw doesn't like the way we do the noise

modeling.

Harry

• Thanks for the quick reply.

In this configuration (+/-15 V, Vcm = 0), the answer is the same.  I am using Orcad PSPice.

The output voltage noise level scales with the feedback resistor, which indicates it is current noise.  The value for the current noise would be 118 pA/sqrt(Hz) at 1 kHz, based on all of my simulations.

The voltage noise appears to be accurate.  When the feedback resistance goes to zero, output noise approaches 4 nV/sqrt(Hz), so that looks fine.  Somehow the current noise is not being simulated properly.

For an additional datapoint, I ran the exact same simulation with an ADA4004.  For that device, I get a currrent noise near 1 pA, which matches the datasheet.