TPS54340: Issue at the transient load response of TPS54340.

Hello Virendra

Could you please fill the QuickStart available for the device. You can find it here SLVC452 Calculation tool | TI.com.

Thank you

Regards

Onkar

Hii Onkar,

I have attached the sheet and below is the Section of schematic for reference, 

Also I have tested this circuit with Cout : 132 uF and 176 uF but same behavior is there.

Regards,

Virendra

1050.TPS54360-361FAMILY_CALC_TOOL_REVE.xls

Hello Virendra

Thanks for sharing the schematic and QuickStart.

The L4 after 1st stage of output capacitor degrading the transient performance of the converter. What is the purpose of the L4? Is it there to reduce ripple further?

You can try reducing the value L4 and see if it improves the transient behavior. Moreover, you can increase the crossover of the loop. Currently it is around 10 kHz you can start with that number and try increasing up to 40 kHz see if it improves the performance.

You can use following section of the calculator to get starting values of components.

Let me know if this helps.

Thank you

Regards

Onkar 

Hello Onkar,

Thank you for your response.

1. The L4 is used as an filter. Checked response by reducing the L4 value and by removing the L4 (short L4) there is no improvement observed. W/F is for your reference

Result of remaining i will share soon.

regards,

Virendra

Hello Virendra

What is your target for the undershoot? Currently undershoot is around 10%. 

Let me know when you get results with improved crossover 

Thank you

Regards

Onkar

Hiii Onkar,

Thank you for your response.

Observed the same behavior at crossover frequencies 20khz and 40khz apx.

1. For 0 to 2 A transient load à output voltage dropped from 5.3V to 4.240V the drop is 1.06V
2. Observed Waveform:

         3.  Checked at crossover frequency : 20 khz and 40 khz referred below section for component values

Due to a voltage drop of about 1V during a 0 to 2A transient load, causing the further circuit to restart.

To avoid this, we need to reduce this undershoot to 0.1V, i.e., 1% to 2%.

 Is it possible with TPS54340 ?

regards,

Virendra Sutar

Hello Virendra 

1 - 2 % is pretty tight but should be possible with TPS54340. To get 1% undershoot you would need to output capacitance more than 85 uF. Considering the derating with DC bias, value of required capacitor will be more than 125 uF.

Here I have considered a 22uF, 10V, X7R capacitor from Murata for derating calculation and following is the typical derating curve for it.

To satisfy the ripple requirement you will need even higher capacitance. I would recommend you recalculate the required output capacitance and chose the combination of capacitor accordingly. Moreover, recalculate the compensation component as those will change with capacitance value.

I am assuming L4 was removed while you performed the above test. Kindly confirm if that's not the case. 

Thank you

Regards

Onkar

Hiii Onkar,

I have already tested the circuit with Cout = 132uF and 176uF (22uF in parallel combination)

I have calculated the values by considering below formula from datasheet and parameters

And Compensation values as in below schematic

But getting same response.(L4 removed)

Cout = 132uF

Cout = 176uF

Thank you..

regards,

Virendra

Hii Onkar,

Can you suggest component values for required transient response, with considering current schematic, it will be very helpful 

I have also checked in WEBENCH Tools

regards,

Virendra

Hello Virendra

Could you please conduct following steps in your experiment?

1. Instead of giving load step of 0 to 2A. give load step of 0.05 A to 2A
2. Fix output capacitance around 200 uF and see if it improves the response. Note considering the DC bias effects actual value of required capacitor will be higher than this. You can also try using electrolytic capactor
3. Chose value of c19(Ccomp) around 1.1 nF and vary the value of R4 (Rcomp) from 10k to 70k to see if it improves the response. you can try 5 - 7 values of R4 in from mentioned range. 
4. Add feedforward capacitor (Cff) of around 1.1 nF and see if it improves the response further. Try 2-3 different values of Cff to check its impact on the response. Note Cff should parallel to top feedback resistors. If possible, you can short R3 and connect Cff parallel to R23. in this case R23 needs to be around 66k

Measure the undershoot with each step and variation with different component values. Let me know the results of this experiment.

Thank you 

Regards

Onkar

Hello Virendra

Any updates on this?

I know I suggested lots of experimental steps. However, considering the stringent transient requirement there is no other way around it. 

Let me know once you finish the experiment.

If our application doesn't demand constat switching frequency you can consider COT or Hysteretic control if above experiments doesn't lead to any fruitful conclusion.

Thank you

Regard

Onkar Bhakare

Hello Onkar,

sorry for delay in response.

as per suggestions checked the circuit and below are the observations:

-Condition 1: Fixed capacitor at 220uF cap, (L4 Short) and Cff 1nf and 0.22uF

-observation:

1. in this case, 0 to 5V start up time is more 

2. behavior at transient load 0 to 1 A

result : no improvement

-Condition 2: Fixed capacitor at 220uF cap, (L4 Short) and Ccomp : 2nF, Ccomp 2 : 10pF: 30Kohm, (also tried different combination for Comp section)

-observation: At 0 to 2A transient response. observed drop is 5.235V to 4.771V i.e. 0.464V 

result : Initially at transient response drop was apx. 1V but now it is 0.464V but still not achieved expected drop i.e. 0.100V 

regards,

Virendra

Hello Virendra

Thanks for sharing the waveforms!

What is the Cff value for improved transient response?

Moreover, could you please try applying load step after ensuring there is minimum load present on the converter before. Under no load conditions converter will be operating in pulse skipping and response of converter will be slower. Additionally output of converter will sitting at higher value than nominal value under typical load condition thus you would see higher dip during load transients.

Try implementing this - 

Onkar Bhakare said:

Instead of giving load step of 0 to 2A. give load step of 0.05 A to 2A

Moreover, could please point out the measurement location for Vout? layout and setup picture will be helpful here.

Thank you

Regards

Onkar

Following are the comments and questions on Waveform 1 and 2

1. Most likely part is operating under current limit due to large output capacitor. Thus, startup time is higher than expected soft start time.
2. Output doesn't recover in the 2nd waveform. Is it measured at the load end or right across the output capacitor?

Regards

Onkar