• State
  • Locked Locked
  • Replies 7 replies
  • Subscribers 64 subscribers
  • Views 321 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.

Original question:

TPS63900: Using a DC/DC converter with LiSOCL2 Primary Cell for IOT application

TPS63900: High Ripple voltage in Buck mode

Damon
Prodigy 60 points
Part Number: TPS63900

dear Tao, 

I have started to test the Eval board TPS63900 but the ripple voltage is high in Buck mode 100mV (VIN = 3.6V, Vout = 3.3V, Load Current 100mA) , for VIN <= 3.3VDC , in boost mode, I can see a ripple voltage of 20mV.

For measuring the ripple Voltage, I took some precautions:

- I used a paper-clip board probe to avoid noise on the ground.

- the probe is far from RF sources

I also used the ADC from the MCU to measure the Vout, same results.

On the output voltage, I tried several capacitors with Low ESR ( Tantalum + ceramic) from 47uF to 100uF, I got the same results. 

Do you recommend to use a Common-Mode choke on the output voltage ?

My target is to decrease the ripple voltage to <20mV, is that possible ?

The other solution that I have is to bring the output Voltage to 4.5VDC then clean it with a LDO deliverying 3.3VDC on the output, but this will increase the power consumption of my system.

  • 0
    TI__Genius 12880 points

    Hi Damon,

    Tao is out of office this week. I will help answer your question.

    Would you please provide waveforms of Vout(AC), SW1, SW2, inductor current? What is the part number and quantity of the output capacitors(or datasheet)? Can you show me the schematic?

    Regards,

    Mulin

  • 0 in reply to Mulin Yuan
    Prodigy 60 points

    Dear Mulin, 

    here are my tests done with both your eval board and my deisgn, 

    on page 15, 16 it is our design and on page 17 is the High noise we are facing in boost mode.

    thanks for your guidanceTPS63900 Tests.pdf

  • 0 in reply to Damon
    TI__Genius 12880 points

    Hi Damon,

    Please see my Vout ripple test result with EVM board. Test conditions: Vin=3.6V, Vout=3.3V, Iout=100mA. Output capacitors: 22uF*1(GRM188R60J226MEA0)+10uF*1(CGA5L1X7R1H106KT0Y0N). The effective capacitance is 18.58uF at 3.3V DC bias.

    As result shows, the Vout ripple is 16.39mV.

    Damon said:
    My target is to decrease the ripple voltage to <20mV, is that possible ?

    So it is possible. The most important factor that influence Vout ripple is the effective capacitance of ceramic capacitors. Add more  ceramic capacitors with higher rated voltage can help. You can provide part number of your ceramic capacitors to me and check whether the effective capacitance is too small. Tantalum capacitor helps little with ripple so it is useless to change it.

    Also, I think the Vout ripple you test is not real ripple. It seems coupling much noise. Use mini GND loop to measure VOUT Ripple as below.

    Regards,

    Mulin

  • 0 in reply to Mulin Yuan
    Prodigy 60 points

    Dear Mulin, 

    thanks for your quick answer . I will share with you capacitors reference, but I am using low ESR. Measured values at 3.3VDC are 23uF.

    By the way, have you ever tested with a Radio Module connected to the Eval Board ? I think connecting  a Load with fixed value is not really representative, as the noise varies when the Radio Module switch to sleep mode and wakes up.

    Do you also have experience with a Pi Filter (or other kind of passive filter) connected on Vout ?

    best regards, 

    damon

  • 0 in reply to Damon
    TI__Mastermind 31540 points

    Hi  Damon,

    I think you mean the transient performance of this device. and the target ripple during transient is 20mV with 23uF EFF Cout, right?

    As you know, the actual ripple during transient is depend on the slew rate of Iout from sleep mode and wakes up and the amplitude with these two state. So could you share us these info?

    I am afraid a PI filter will not help, i think more output caps is needed for your case.

    Regards

    Tao

  • 0 in reply to Tao Zhang
    Prodigy 60 points

    Dear Tao, 

    Happy New year  !

    Tao Zhang said:
    and the target ripple during transient is 20mV with 23uF EFF Cout, right?

    Right and currently I am far away from 20mV as stated on my tests results (see tests results shared on this post from 6 days ago).

    Tao Zhang said:
    [So could you share us these info?]

      in sleep power mode : 9uA  @ 3.3VDC, during data processing after wake-up:  6-15mA, during RF Transmission 80-95mA @3.3VDC I will try to send you some curves

    I will do some tests with static load like what you did but it's not representative of real conditions. Your datasheet displays the buck-boost is designed for NB-IOT/LORA or equivalent radio module, have you done some tests about ripple voltage with a Radio module ?

    Tao Zhang said:
    I am afraid a PI filter will not help

    I am interested to know why it will not help, please share your thoughts. I will do more tests by adding more output caps.

    many thanks for your support, 

    damon 

  • 0 in reply to Damon
    TI__Mastermind 31540 points

    Hi Damon,

    Happy new year too !!Grinning

    I am afraid we did not do such tests based on Radio module. DS says this device is designed for NB-IOT or similar applications just because it is ultra low quiescent current when load is close to 0.

    From my understanding, i think if we want a better transient performance, we need a higher cross frequency of loop, but unfortunatly, an additional LC filter will not help, mosly, it will decrease your cross frequency. Below is some calculation results based 1uH+1uF filter.  You can see there is no additional zero or pole before 100Khz, For boost and buckboost converter, the fcross is mostly lower than 100Khz caused by RHPZ, so it will not help.

    From other side, inductor of LC filter need a time to energized from 0 to final current  in your application, which means it will make your final ripple worse during transient.

    Anyway, these analysis are just analysis, i did not do any bench to confirm. Sweat smileSmile -  I think they should be right. haha

    Regards

    Tao