OPA189: OPA189 noise in unity gain setting

Well the model looks right at about 5.2nV flat going down in frequency. You would have to be more specific on how your measurement system is mapping to a 15nV number. 

Hi Michael,

We amplify the Opa189 output with a low noise amplifier, feed the output to an audio analyzer to convert the analog signal to digital data and calculate the FFT. We verify our measurement system noise by shorting the input to our measurement system.

If we set the Opa189 to gain of 11, of course using the same measurement system, the input referred noise pretty much matches the specs.  

Here are some of the measurement results. Blue is for Opa189 in buffer configuration. Orange is our system noise. Yellow is for gain=11, with 10 ohms and 100 ohms resistors.

Hi George,

Please try the following configuration and it may reduced the overall noises. The configuration is based on working bandwidth of 80kHz. If the 80kHz is the -3dB point, you can further reduce the overall noises. 

Since you are able to measure the noise, it may be conclusive to check it out rather than simulating it. 

OPA189 Noise Analysis 07152024.TSC

Enclosed is the recent application note that you may be interested in. 

https://www.ti.com/lit/wp/sboa586/sboa586.pdf?ts=1721079572601&ref_url=https%253A%252F%252Fwww.google.com%252F

If you have additional questions, please let us know. 

Best,

Raymond

Hi Raymond,

Thanks for the information.

I tried to measure the input referred noise with the noise gain=11 after I saw the higher than expected noise in buffer setting.

Our application is for OPA189 in buffer setting. The low noise and low DC error of OPA189 are very good match for our application. We were expecting the noise level to be close to the 5.2nv in the broadband part.

George

Hi George, 

I am fairly sure that the OPA189's Input Voltage Noise Spectral Density vs Frequency looks like below. 

Gang(George) He said:

after I saw the higher than expected noise in buffer setting.

There is chopper frequency at approx. 125kHz range. Perhaps you are seeing the noises contributed from the broadband region marked in red. 

what is the unit of 5.2nv in the broadband part? Is this Vrms or peak to peak. OPA189's input voltage noise spectra, input referred noise, en at 1 kHz: 5.2 nV/√Hz. 

If you place a LPF after the OPA189 and get rid of noises contributed by chopper frequency, you should see the voltage noise spectra at approx. 5.2 nV/sqrt(Hz). 

If you have other questions, please let me know. 

Best,

Raymond

Hi Raymond,

"what is the unit of 5.2nv in the broadband part? Is this Vrms or peak to peak. OPA189's input voltage noise spectra, input referred noise, en at 1 kHz: 5.2 nV/√Hz. "

The broadband part power spectrum density should be about 5.2 nV/√Hz. It is a Vrms value.

"If you place a LPF after the OPA189 and get rid of noises contributed by chopper frequency, you should see the voltage noise spectra at approx. 5.2 nV/sqrt(Hz). "

Our measurement system is band limited. The signal is band limited to 80Khz and the sample rate is at 192Khz. There is no high frequency (>80Khz) content folded back. With OPA189 configured to noise gain=11, the noise measured with the same measurement system just matches the datasheet.

I am thinking that when OPA189 is configured with unity gain, the noise spectrum density is higher than 5.2 nV/√Hz.

George

Hi George,

Gang(George) He said:

Our measurement system is band limited. The signal is band limited to 80Khz and the sample rate is at 192Khz. There is no high frequency (>80Khz) content folded back.

Did you configure the buffer circuit similar to the simulation below? If the OPA189's bandwidth is limited, you should see less overall output noise, but the input referred noise spectra is slightly higher with the LPF filter, which the pole is configured at approx. 72 kHz. 

OPA189 buffer Noise Analysis 07162024.TSC

If you present the actual OPA189 buffer circuit, maybe we can back calculate what the input noise spectra should be approx.. The input noise spectra figure in nV/√Hz should be comparable regardless what gains are used in a measurement. 

If you have other questions, please let me know. 

Best,

Raymond

Hi Raymond,

The OPA189 buffer setting for my test is shown on the first post above. My measurement is from Vout to ground.

As you know, I also measured the noise gain=11, which is also shown on the first post above.

For both configuration, buffer and noise gain=11, the Tina simulation tells me that the input referred noise power spectrum densities are both at 5.2  nv/√Hz.

From real system measurement,

The results are input referred noise power spectrum density, which are shown on the third post in this series. I also put the  noise power spectrum density of my measurement system for reference.

noise gain=11, input referred noise power spectrum density is at about 5.2  nv/√Hz.

buffer configuration, input referred noise power spectrum density is much higher than 5.2  nv/√Hz

George

Hi George, 

I am starting to understand your query. I think that you are not compare apple to apple in either circuit. 

With Noise_Gain = 11 V/V, the pole is approx. 1.27MHz (GBWP)

With Noise_Gain = 1 V/V, the pole is approx. 14MHz. 

Even the input signal is limited to 80 kHz, the op-amp might still have sufficient bandwidth to amplify noise components beyond this frequency. If you want to compare apple to apple, you need to limit the op amp's BW consistently. You should use the op amp configuration that I simulated in the previous reply. 

Anyway, this could be the discrepancy why you are seeing the differences in the input voltage noise spectra density figures, which are normalized. 

I assumed that the above measurement is made in the same circuit and power supply. If not, these could be variables as well. 

If you have other questions, please let me know. 

Best,

Raymond

Hi Raymond,

For the actual noise measurement, we tried our best to avoid any variables. All the tests are on the same battery power. For noise gain=1 and 11, use the same opa189.

I just got a TI-PLABS-AMP-EVM kit and measured the noises of:

our measurement system, OPA211,OPA140, OPA277,OPA188 AND OPA171.

I used the same setup for all the Op amps by just changing the 8-pin dip adaptor.

The measurement is for buffer reconfiguration. Actually, I used the TI-PLABS-AMP-EVM circuit 5 by: shorting R27, insert JMP17 and JMP50, Input=GND. Measured noise from Output2 to GND. Again, DC power is batteries, +/- 8.4V.

Here is the results.

Except for the OPA188, the measured noise is higher than the datasheet. All others noises match the datasheet.

I use 4 nV/√Hz for my system broadband noise.

These measurement results show that our measurement process is sound.

I am waiting a SOIC OPA189 to measure the noise for the same setup.

information of TI-PLABS-AMP-EVM

https://www.ti.com/lit/ug/sbou150a/sbou150a.pdf?ts=1721673327685&ref_url=https%253A%252F%252Fwww.ti.com%252Ftool%252FTI-PLABS-AMP-EVM

I have tested a SOIC OPA189 on the TI-PLABS-AMP-EVM. The broadband part of the noise is about 7.5 nV/√Hz, which matches the datasheet.

My inquiry can now be closed.

Thank you for your help.

George