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.

TLV2211: source resistance affection to the output noise of low noise OPA

Part Number: TLV2211
Other Parts Discussed in Thread: OPA333

Hi,

Using the following circuit to build a board to evaluate the output noise mainly in 0.1~10Hz. Power supply single +3.3V is provided by  battery. Finds that R1~R4 resistance has serious affection to circuit for both TLV2211 and OPA333 chips.

When R1~R4 are all 120Ohm, and Rf=220K, the gain is 1666. got the following output noise by clean data capture tool  only 2LSB convert to 1.6mVpp noise. it is a little less than simulation result due to avr fitler.

when R1~R4 is all 30Kohm and Rf=20M, the gain is 1333.  the output noise is 10LSB , about 8mV.  only changed 5 resistors while gain is lower, output noise is 5 times worse. 

The circuit has a 16Hz cut off freq LPF, so noise of 0.1~10Hz is definitely dominate. The noise density of  15Kohm(30K||30K) is about 16nV/rtHz, it can be neglected when comparing with op amp's noise density at low freq tens Hz.    We got the similar result eitherusing OPA333 or TLV2211. we only see that 20% noise level increase by doing simulation or calculation when comparing the above 2 case. what's the possible cause?

PS, the data capture system runs at 100SPS, and sample-hold time is configured to long enough to eliminate the error.

BRs,

JH

  • JH,

    In order of likely issues.

    Is your test circuit completely enclosed in a solid faraday cage? To keep external noise out. Metal paint can with sealed lid works well.
    Is there a VCC bypass capacitor in circuit? Yes, prefered.
    Try adding a small capacitor (10pF) in parallel with R5, to make sure any parasitic capacitance at inverting input doesn't affect phase margin.
    If none of that helps, try a 10k resistor from op amp output to ground to get op amp output to be class A.
  • Hi Ron,

    Thanks for your comments.

    The board testing is shielded like your suggestion as the following pic show, the metal shield is connected to the board GND. It works effectively to remove the air flow or human motion affection, but not works for the 5 resistors value changing.

    Adding a 10K resistor to GND at the output terminal of OPA, it has peak-peak noise reduction around 10% in a little long time observing. It helps a little.

    All other condition keep the same while only changing 5 resistors, the similar noise change when using OPA333, we suspect charge injection of chop cause the noise worse. But with TLV2211, the same result, is TLV2211 a internal switching op amp? 

    BRs,

    JH

  • JH,

    In a dream last night I thought the increased noise came from the battery itself and noise varied with battery load current.
    It is easy to prove or disprove this thought. For high value resistor test, just add 120 ohm load on battery to replicate the battery load when the four 120 ohm resistors were used.

    Yesterday, I wanted to blame the resistor thermal noise but that wouldn't be close to 8mV.

    TLC2211 data sheet is nice enough to include a device schematic. Nothing odd is apparent in that schematic.

    Input current noise for FET input can't be the problem.

    Am I correct to assume that noise change is same with both op amps? This suggests issue is outside of op-amp itself.
    I'll give this more thought, and post if anything else comes to mind.
  • JH,

    This is how I would test 0.1Hz to 10 Hz noise.

    C5 has two functions. First is a 0.1 Hz high pass filter. Second is stopping Vos error gain which might make output on the supply rail. C1 and C2 make two low pass filters. 

    Split battery supply is recommended. It can be converted to single supply however, the charging of C5 will take more time to stabilize. 

  • Ron's circuit will work well.  If the noise is lower you may consider this reference design which describes how to measure the 0.1-10Hz noise in roughly the same way we do during device validation.