• State Resolved
  • Locked Locked
  • Replies 2 replies
  • Subscribers 48 subscribers
  • Views 1199 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

High output noise noise at the output of THS4271DGN non inverting buffer when feedback resistor=0E

Anoop Joshi
Expert 1125 points
Other Parts Discussed in Thread: THS4271

Hi,

I'm using a circuit which accepts a differential input through simple non inverting buffers and applies it to a Differential to Single ended converter(D2S) stage to obtain a single ended output. The schematic with the non inverting buffer section zoomed in is attached below.

The issue observed is that, the idle channel output, ie the output of the circuit with the inputs shorted to ground is observed to be very noisy. When checked with a digitizer with an inherent integrated idle channel noise of around -69dB, the integrated noise at the D2S circuit output comes to around -40dB. I'm sorry I couldn't attach an FFT plot to this mail, but it is seen that the noise floor itself goes real high. I could isolate the source of the noise and find that it was coming right at the output of the non inverting buffer. This is when the feedback resistor for the non inverting buffers were set as 0 ohms. When I checked with 250 ohms feedback resistors, it was observed that the output got totally clean, and I could measure some -68dB idle channel noise at the D2S output, very near to the digitizers inherent idle channel noise. I cannot really understand why the 0 ohms resistors caused the noise floor to worsen so much. I read in the datasheet that for unity gain configurations there are chances this op amp can show instability if the feedback resistance is 0 ohms, but that is supposed to happen at pretty high frequencies, ie around100MHz I believe. I'm doubting if it is due to some extra parasitic capacitance coming in the feedback path.. Can you please provide some insight into why 0 ohm resistance causes the noise floor to worsen for the non inverting buffer so much compared to say a 250 ohms resistance.A snap shot of the layout of the non inverting buffer section is given below: The hatched lines indicate cut out areas in the power and ground planes.

  • 0
    TI__Genius 13935 points

    The THS4271 is a very high bandwidth amplifier.  At unity gain the amplifier has the most bandwidth and may also have some gain peaking.  When you use a 0 ohm resistor in the feedback path there are a couple of parasitic capacitances that are directly connected to the amplifier output.  By using a resistor those capacitances are isolated from the amplifier output.   

    One parasitic capacitance is the input capacitance of the inverting input.  The other parasitics are due to the board traces.  When you have a unity gain configuration there is no need for Rg, but the pads are still connected and form a parasitic capacitance.

    With noise it is not necessary for an actual instability to be present to make a measurable increase in noise. 

    Due to the bandwidth of this amplifier it is only necessary to create gain peaking at high frequency to make an increase in noise.  For example, if the 0 Ohm resistor in the feedback path causes 6dB of peaking over a 200 MHz band that will show up as a large increase in the composite noise.   

    Does the FFT of the noise show a bump over a certain frequency band?  Can you run a frequency response and see if the gain has peaked? 

    [EDIT]

    Also, looking at the board layout.  It appears that the ground plane has not been removed from under Rf and Rg (R32 and R38?).   This would result in more parasitic capacitance from that portion of the board layout.  Using a resistor for Rf would effectively mask most of that capacitance from the amplifier output. 

  • 0 in reply to Loren Siebert 1
    Expert 1125 points

    Hi Loren,

    Thanks for the detailed reply. Yes, even though it was seen that the general noise floor itself had gone up when 0 ohm resistor was used, it wasn't exactly flat. My FFT bandwidth was 40MHz, and yes I could see a bumps around in the band width of interest, but the bumps seemed to change positions between multiple runs. I'll also try to run a frequency response check to see if peaking happens.

    Yes, we had not removed the ground planes from beneath Rf, Rg although we had done so beneath the IC IO pins. Thanks a lot, now the issue is clear

    Thanks,

    Anoop