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.

TPS74801: TPS74801: questions for PSRR test

Part Number: TPS74801
Other Parts Discussed in Thread: THS3120

Hi Expert,

I have some questions need you help to confirm as below:

1. I saw typical value show in the DS as below, could you help to know the Min & Max value as well? because customer would to verify PSRR on their board and make sure the test result is pass our criterial.

  

2. For PSRR test , customer will add VAC from function generator, the VAC should be sine wave of signal, right?

3. Due to customer not prefer to use "AP" for PSRR test, which can be scan Frequency domain VS Gain, so do you have recommend frequency to check PSRR? just like 1kHz, 10kHz ..or others you suggest.

4, Do you have PSRR test report could to share with customer?

Regards,

Mark 

  • Hey Mark 

    Unfortunately no we do not normally provide max and min values for this as this can vary with frequency, board layout and set-up. I can check on the other recommendations and get back to you in 2-3 business days to help with the other questions. 

    Thanks, 

    McKyla

  • Hi McKyla,

    Ok, look forward to hear from you .

    Regards,

    Mark

  • Hey Mark, 

    Apologies for not thinking of this earlier. We have an applications note on how to measure PSRR that goes into detail on different ways to measure PSRR. Method 2 is the recommended https://www.ti.com/lit/an/slaa414a/slaa414a.pdf?ts=1649091415698

    Correct, they need to use sin wave. 

    Thanks, 

    McKyla

  • Hi McKyla,

    Customer try to calculate PSRR as below, could you help to check the result is meet TPS74801 spec or not?

    Vin= 3.3V

    Vout= 1.8V

    Iout= 0.04A

    Customer measure the PSRR and VAC = 1V (p-p) at 1kHz, then get below result.

    Ripple(input) = 8.8 mV

    Ripple(output) = 4.8 mV

    PSRR = 20 log10(Ripple(input)/ Ripple(output)) = 20 log10(0.0088/0.0048) = 5.264dB

    Can you please help judgment does the result of PSRR is meet TI TPS74801 design?

     And do you have PSRR test report can share with customer?

    Regards,

    Mark

  • Hey Mark, 

    Does the scope have an fft function? If so please send the Ripple (input) and Ripple (output) fft scope shots. 

    Best, 

    McKyla

  • Hi McKyla,

    To check with customer, the scope do not support FFT function.

    Regards,

    Mark

  • extra two questions need you help to check:

    1. customer to add 1V peak to peak AC signal to the input, but only 8.8mV measured show in the scope, do you know why? I saw the document use THS3120 for measurement, does it means we need to tool for PSRR testing? do you have other recommend method to to measure PSRR when customer without THS3120?

    2. if the 1V peak to peak AC signal exist on input side, according to the formula the PSRR = 20log(1V/4.8mV)= 46db, looks it is not align with 60db showed in datasheet. do you have any comments?

    3. what's the prefer method for PSRR measurement? i know you share the document as above, but we still have some questions here.

    Regards,

    Mark

  • Hi McKyla,

    Do you have any comments ?

    Regards,

    Mark

  • Hey Mark, 

    1) For PSRR you need to do small signal analysis like 100mV max. The THS3120 it is a high speed amplifier used to help gain up the signal, if you were not using this then the measurement can not test PSRR this way using a scope. It would be challenging to see the small signal differences here. 

    2) 1V is not small signal analysis, hence this was not a valid PSRR test. 

    3) Do you have a Bode 100 that you can use? That would be the preferred method. 

  • Hi McKyla,

    Q1: So the THS3120 is required for PSRR test if customer follow the document(LDO PSRR Measurement Simplified (Rev. A) (ti.com), right? 

    Q2: But i saw the 1V(p-p) @1kHz is  test condition in the document. what's the right signal for PSRR test? 

    Q3: if use bode 100, do you have guide line to share with customer how to measure the PSRR, how to connection? 

    Regards,

    Mark

  • Hey Mark, 

    There needs to be some sort of amplifier to gain up the signal if you are using a scope to get an accurate measurement. Yes, it says to 1V because they used an amplifier to gain up the signal. 

    This pdf and diagram should help the customer, there needs to be a summing amplifier, filter before the LDO, and then an amplifier into a network PSRR Measurement.

    Hope this helps! 

    McKyla

  • Hi McKyla,

    Customer  used “Bode 100 network analyzer” & “J2120A line injector” to measurement PSRR of TPS74801 last week.

     

    LDO(TPS74801) design as below.

    Vin= 3.3V

    Vout= 1.8V

    Iout= 0.04A

    Below is the test result for PSRR from our verify, please help confirm  the result is ok for TPS74801? And please also let us know how to judgment the PSRR curve is pass or fail?

    Regards,

    Mark

  • HI McKyla,

    May I have you comments?

    Regards,

    Mark

  • Hey Mark, 

    Apologies for delay. This graph looks relatively similar to the datasheet graph given the differences in Vin, Vout, ect. Do you know what output capacitances the customer used? This could account for differences in the higher frequencies. 

    Thanks, 

    McKyla

  • Hi McKyla,

    Customer removed all the Input capacitor of LDO, and the output capacitor are 2*22uF + 1*0.1uF.

    Could you please help to confirm the test result is PASS or FAIL.?

    Regards,

    Mark

  • Hey Mark, 

    I am not sure what kind of answer the customer is looking for, because PSRR changes based on different the variables and at different frequencies, it's not a PASS or FAIL test. It's a graph that we characterize based on a sample of units for a typical metrics and is not something that is tested at final test. The caps the customer used are different then the test conditions defined in the datasheet (see below). 

  • Hi McKyla,

    Below is customer would to confirm with you:

    1. Do we have necessary to verify PSRR when we used TI LDO solution for our design application?
    2. If yes, how to judgment our PSRR test curve whether have any problem or concern from design view?
    3. If not, please let us know what the reason?

    Regards,

    Mark

  • Hey Mark, 

    Can you clarify this question? 

    Are you wondering how to get the PSRR curve to be a closer match to the one found in the datasheet? The same output capacitance would help. If you notice where the knee of the graph is and where the bend turns it is similar to one found in the datasheet. Looking at the knee (or where the graph turns) to compare two different PSRR graphs is a good way to see similarities and verify between two different parts are functioning the same if capacitances different. With different capacitances the higher frequencies are going to differ more with lower frequencies are going to be more similar. 

    If you are wondering if the PSRR might be an issue for your specific application this will be something that the customer will need to access. I can't give guidance from a systems point of view as this is not our expertise. 

    If you are wondering something else please clarify. I hope this information helps.  

    Thanks, 

    McKyla

  • Hi, McKyla

    I agree with you that different test condition will be have a little bit different test result. 

    According to your experience, what kind of application will require to test PSRR in general? 

    For customer application, they are use TPS74801 on rack sever project, I never seem to test PSRR on sever application before, do you thing this is necessary to test?

    Regards,

    Mark

  • Hey Mark, 

    This is going to have to be something that the customer determines, it is not something that we can say one way or another if it is necessary to test. I can help with any LDO related questions on functionality and/or issues experienced with LDO, but I can not make a call on things of this nature this will be left up to the customer. 

    Thanks, 

    McKyla