• State
  • Replies 2 replies
  • Subscribers 48 subscribers
  • Views 39 views
  • Users 0 members are here
Support feedback
Options
Options
Related

OPA2325: Low frequency noise - what alternatives should I consider?

Simon Merrett
Prodigy 40 points
Part Number: OPA2325
Other Parts Discussed in Thread: TLV888

I have an OPA2325 in non-inverting high gain mode taking sensor output and feeding into an ADC. I am specifically interested in 1-50 Hz signal integrity and see some noise on my input, understandably, that I would like to see if it can be improved with amplifier selection or other measures.

image.png

AC signal in is biased to VDD/3 to line up with single-ended ADC voltage reference range. C503 will be a charge cap for the option of a sample-and-hold on the amplifier input, although its value and those of C504 and R507 can absolutely change.

Here is an example FFT with 1 kHz sine reference signal as the main peak (sorry for inverted Y axis polarity) and noticeable noise in the 4-200 Hz region, especially below 50 Hz.

image.png

Is there any better-suited opamp, opamp scheme (I am open to cascading but not sure that will help) or circuit layout that will significantly improve my 1-10 Hz and 10-100 Hz noise in the FFT? Remember that I still need pretty much full scale gain at 4+Hz for the signal, so I can't just change the high pass filter on the gain by reducing C505.

  • 0
    TI__Guru**** 166566 points

    Simon,

    Try TLV888 for low 1/F low frequency noise. It is also high gain bandwidth and low cost. 

    The high value resistors are also noisy, but that is wide band noise.

  • 0 in reply to Ron Michallick
    Prodigy 40 points

    Hi Ron, thank you for your recommendation!

    I agree the high value resistors are non-ideal. Part of the challenge as I understand it is the need for C505 to allow the biasing of the output to midrange of the ADC AREF. But the high pass created by C505 and R508 mean that I'm being squeezed to use high value resistors to allow my single-digit Hz signals to come through unattenuated. I based it on https://www.ti.com/lit/an/sboa224a/sboa224a.pdf? . 

    Would I be better off using a cascaded pair of amplifiers to allow a smaller gain on the first one, enabling lower value resistors? Then use a second amplifier on the output which wouldn't require a capacitor between its equivalent of R508 and ground to do the heavy lifting in gain? I realise that could be a trap to create more noise on the output but I don't have an intuitive feel for a good topology here and I'm sure you do!