• State TI Thinks Resolved
  • Replies 6 replies
  • Answers 1 answer
  • Subscribers 31 subscribers
  • Views 230 views
  • Users 0 members are here
Support feedback
Options
Options
Related

OPA1611: Input Noise Level

mcav
Prodigy 20 points
Part Number: OPA1611
Other Parts Discussed in Thread: TINA-TI

Hello, I want to build a preamplifier with an input impedance of 5 kohm. The signal I need to measure is a 200 nVrms signal at 30 kHz. First, I wanted to take this signal with a low-pass filter with fc = 40 kHz. However, since my bandwidth is very wide (I took my noise bandwidth as 64 kHz), even if I use an op-amp with the lowest input voltage noise density (e.g., OPA1611), only the op-amp noise alone becomes 278 nVrms due to en*sqrt(64000). And because of the Gaussian distribution, the peak value of this noise becomes 278n * 6.6 = 1836 nVpp, which means that the real signal we need to measure will disappear inside the noise. I think the only area I can adjust to reduce the op-amp’s input noise is the feedback path. Since I don’t have much knowledge about bandpass preamplifiers, I couldn’t make progress. I need your support on this. Is there anything wrong in my way of thinking? What are your suggestions?

  • 0
    TI__Mastermind 26715 points

    Hello, 

    Yes, you also need to account for the input current noise combined with source impedance. The OPA1611 is a bipolar input and while it has fantastic noise performance, it can be degraded when interfaced with high source impedance. High current noise combined with high source impedance translates into additional voltage noise. For this reason you should consider either replacing the front end with a discrete JFET device for a composite amp solution or using a JFET input op amp. 

    As an example, you can see in my table below that there is a tipping point where a JFET front end will outperform a bipolar front end when the source impedance becomes large.

     

    I have written several Application notes on replacing the front on of a bipolar amplifier. When taking this approach you gain the benefit of the high input impedance of the JFET devices combined with the ultra low voltage noise performance of the bipolar op amp. 

    Useful Application Notes:

    I hope this information is helpful. 

    Best Regards, 
    Chris Featherstone

  • 0 in reply to Chris Featherstone
    Prodigy 20 points

    Thank you for your valuable help. I am reviewing the documents, but something keeps bothering me. No matter how low the input-level noise is, as long as I cannot reduce the bandwidth, because of en × √(BW) I will never be able to see the 200 nV signal at this frequency (30 kHz). I need to be able to reduce the bandwidth somehow. What kind of circuit approach should I take? How can I limit my bandwidth?

  • 0 in reply to mcav
    TI__Mastermind 26715 points

    Hi Muhammed, 

    Do you have a schematic I can take a look at in order to determine what component to add for bandwidth limiting? In general you would normally add a capacitor in the feedback beta network to achieve this. As an example using my circuit the feedback resistor is circled in red below. Placing a capacitor in parallel with it limits the bandwidth of the whole circuit. It would be much easier to simulate your circuit in Tina Ti to perform bandwidth and noise analysis. 

    Best Regards, 
    Chris Featherstone

  • 0 in reply to Chris Featherstone
    Prodigy 20 points

    Hello Chris, please give me a few days. I will share the schematic with you.

  • 0 in reply to mcav
    Prodigy 20 points

    Before sharing the complete schematic, I will first ask you questions regarding the preamplifier section, as I am experiencing some issues with it.  

    1- Below are two schematics I created using different approaches that operate with the same gain. When I perform noise analysis on both, the input noise levels come out almost identical. Considering the complexity of the JFETs, the single-op-amp solution seems more appealing. Is there anything I am overlooking in this part? Do you think my reasoning is incorrect. What can I do to reduce the bandwidth in these applications?

    2- My question is about TINA-TI: Is it possible to have it calculate the RMS value of the input noise graph within a specific bandwidth?

    3-By paralleling the op-amps, I can achieve an input noise level of 1.26 nV/√Hz. What could be the advantages and disadvantages of this approach? You may consider this question together with the first one. With this question, I'm basically trying to understand when I should prefer a discrete JFET-based design

  • 0 in reply to mcav
    TI__Genius 14185 points

    Hello,

    A discrete JFET approach would be better for very high input impedance - things like microphone capsules or Ph probes usually need very high input imedance.

    If your only goal is the lowest possible noise, and the source that the signal is coming from is low-impedance (i.e. a transformer in a dynamic coil) then it is generally easier to parallel multiple amplifiers.

    Note you will need a resistor at the output if each amplifier so that any differences in offset voltage between the amplifiers does not result in one amplifier sinking a lot of current from one of the other amplifiers.  Here's an e2e that describes the parallel amplifier behavior with an example:

    e2e.ti.com/.../opa627-paralleling-2-op-amps-to-reduce-noise-file

    We can address any further questions after the Thanksgiving holiday.

    Regards,
    Mike