ADA4522 1/f noise?

Hello!

The datasheet of ADA4522 shows the measured voltage noise density for only  > 10 Hz frequency region.

I was measured the power spectral density (square of voltage noise density) in the 0.001-32 Hz range.

The result is shown at the attached figure.

OPAMP was configured as non-inverting with the 100 Ohm and 10 kOhm resistors (both 0.01 %).

The "+" pin was shorted to ground.

The PCB layout was as in the following article

(figure 13) to minimize the thermoelectric voltages. Extra "dummy" resistor at the "+" input was added and set to 0 Ohm or to Rg||Rf.

The PCB was placed into the Faraday box, which placed into the styrofoam box with no airflows.

The record of 55 hours length was obtained, and the periodograms width the 2000 seconds rectangular window was averaged.

The questions are:

1) Does ADI have the such extended Voltage Noise Density measurements, say - down to 0.1, 0.01 and 0.001 Hz?

2) The estimated white noise ASD  = sqrt (49) = 7 nV / sqrt (Hz) is some higher than was expected for this configuration

(see green line level at the figure). Is it correct result?

3) The expected corner frequency was about 0.005 Hz, where the white noise level intersects the noise floor of the measuring system.

However, the ~0.1 Hz corner frequency was obtained.

Is this corner frequency intrinsic to the ADA4522 OPAMP? If so, is the ADA4528 OPAMP has the similar behavior?

Or the corner frequency obtained still relays to the uncompensated thermoelectric voltages?

Thank you in advise,

Peter.

Simultaneous noise measurement of the ADA4522 and the ADA4528 opamps, mounted on the same PCB. Common bipolar power supply +- 2.5 V. The PCB mounted as described above. Gain set to 1001 (10 and 10000 Ohms SMD 0602 1% resistors).



A new picture added.
[edited by: TPeter at 3:27 PM (GMT -5) on 11 Nov 2018]
  • I don't understand.  Zero drift amplifiers are not supposed to have any 1/f noise?

    Also, a 10k resistor has a voltage noise of 12 nV/rt-Hz, so I don't see how you got 7??

    Harry

  • As the datasheet claims about the ADA4622: "They also offer low voltage noise density of 5.8 nV/√Hz (at f = 1 kHz, AV= 100) and reduced 1/f noise component." So,  "reduced" - is the keyword. But, how much reduced - still unknown for any of zero drift amplifiers. At the same time Zero-drift Op Amps page  says that "By design, the zero-drift amplifier has no 1/f noise nor does it suffer from long-term drift." As can be seeing, the information about 1/f noise in zero-drift amplifiers is some contradictory.

    To refer to output, the noise of the 10K resistor multiplied by 1, while the OPAMP noise multiplied by 101, resulting the ~586 nV/rt-Hz, so the impact of the 10 K resistor is negligible when we coming back to RTI.

  • 0
    •  Analog Employees 
    on Nov 8, 2018 2:44 AM

    Hi Peter,

    From the ADA4522 designer's own lab measurements, we have also seen this same 0.1 Hz corner frequency where noise begins to rise again. Although theoretically the gain in the autozero amp should be enough to push this corner much much lower than that, we also probably ran into some parasitic thermoelectric voltages as well in the measurement setup. This may be as low as is practically possible, since we also observed best measurement practices while collecting this data like you did.

    Since this seems to be intrinsic to measurement setups and not specific to any parts, my best guess is that 4528 would show a similar corner.

    Best regards,

    Catherine

  • Hi Catherine.

    Did you saw the following report  (page 16)?

    The 1/f noise corner for the ADA4528 does not observed down to 0.01 Hz.

    Now I obtain the same result for this part with my measurement setup, the corresponding picture with the simultaneous measurement of the 4522 and the 4528 parts will be shown here soon.

    So, the question about the intrinsic noise of the ADA4522 remains open.

    Best regards,

    Peter.

    ---------------

    The new measurement results was added on my first post, see above.