• State
  • Locked Locked
  • Replies 7 replies
  • Subscribers 49 subscribers
  • Views 556 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.

OPA2836: How to measure PSRR by using an oscilloscope ?

user5802412
Intellectual 300 points
Part Number: OPA2836
Other Parts Discussed in Thread: OPA847

Hello TI Member :

I am trying to use an oscilloscope to measure the PSRR of OPA2836,  but I am not sure whether the test environment can verify it, could you help me to check whether my test setup and method are correct? Thank you!

1. set the OPA2836 as an unity- gain buffer 

2. let VS+ = AVDD = 5.0V, VS- = AVSS = 0V, VIN1+ = AVDD/2 = 2.5V , VIN2+ = AVSS = 0V , Rloading = 2.2 kohm

3. add AVDD with a noisy source (  1 KHz  sin wave, Vpp = 500 mV  )

4. let the probe tie to ground and record the wavefom of probe ( I think this can be the noise floor caused by probe, so I need to exclude it, you can see the scope capture that named " VOUT_Floor " ) 

5. use a 1:1 probe to measure VOUT1 's ripple  

6. use the math function to get VOUT1 's ripple which is named " VOUT - floor  "

7. record the value peak-peak of ripple 

8. PSRR = 20 Log ( 501.30 / 1.046   ) ~ = 53.6113 dB 

Thank you for your any suggestion!

B.R

Chen

  • 0
    Guru 140200 points

    Hi Chen,

    I know there are more sophisticated circuits but for a simple estimation this circuit should do:

    chen_psrr_1.TSC

    The OPA2836 must run in linear operation. A non-inverting gain of 101 increases the non-ideally rejected supply voltage variation at the output of OPA2836. More gain is no good idea because of the offset voltage of OPA2836. The circuit is followed by a gain = 1001 amplifier to get better scope results.

    It's important that VG1 and VG2 are synchronous to each other but 180° phase shifted. Otherwise the measurement is flawed by the finite common mode rejection (CMRR) of OPA2836.

    A PSRR of 80dB at 1kHz would result in an output signal "VF1" of

    0.2Vp / 10000 x 101 x 1001 = 2.02Vp

    Kai

  • 0 in reply to kai klaas69
    Guru 48395 points

    Incidentally Kai and Chen, 

    The DC PSRR and CMRR are part of the final test sweep using the classic servo amp approach. Rarely do we try to measure the AC PSRR and what shows up in the datasheets are usually simulations. 

  • 0 in reply to kai klaas69
    Intellectual 300 points

    Dear Kai:

    It's very appreciate that you provide a simulation file for me! I think I need to follow this test circuit and implement it on a new PCB. Do you have any recommend OPA product as pre-amp ( IOP1 ) that followed by OPA2836 ( OP1 )? Thank you !

    B.R

    Chen

  • 0 in reply to Michael Steffes
    Intellectual 300 points

    Dear Michael & Kai:

    Sorry for delay reply, this approach seems vary powerful to solve the noise limit of oscilloscope! I'll try it! and It's appreciate that you provide the info for me! Thank you.

    B.R 

    Chen

  • 0 in reply to user5802412
    Guru 48395 points

    When I used to try and bench generate an AC PSRR sweep, the immediate problem you run into is how to generate the signal into the supply line while holding the device stable. The device wants to see a low impedance looking out, normally provided by the decoupling caps, but if you add those caps how does that impact your signal injection and the stability of whatever you use to inject th signal - you almost need to tap off that input signal as a reference signal to know what level you are injecting, a network analyzer calibration step 

    Assuming you can get past this (and I never did very well), your post amp should be high gain, low noise and not bandlimit in your sweep of interest - I would look at the OPA847. 

  • 0 in reply to user5802412
    Guru 140200 points

    Hi Chen,

    do you want to measure the PSRR only at 1kHz?

    In this case you could add tight high pass and low pass filters to limit the noise.

    Also, the shown gain =1001 amplifier is a bit simplified. You can take two and more amplifiers in a row and evenly split the gain over them.

    Kai

  • 0 in reply to kai klaas69
    Guru 48395 points

    Incidentally, I was checking the datasheet for what to expect (simulated data) and this plot is inverted - PSRR should be a positive number (rejection ratio) whereas this is the raw gain from the power supply to the apparent input referred offset voltage error. put that on the list to fix along with the other supply voltages,