TPS65400: TPS65400 Input Capacitor

Hi Henry.

1, What's your 10uf rated voltage and How about your 10uf DC bias curve, if it is not good, when input voltage added into the mlcc, the 10uf cap value maybe have a large derating. so that there is no enough power to provide when converter switching.

2, Draft estimate capacitance value Cin=Iin/(deltaVin*fsw)

3, if you need to get more, please upload your schematic, test waveform and your capacitance datasheet.

BR,

yuchang

Hi Yuchang,

My 10uF are rated at 25V, and I am only using it at 12V.  Besides, the oscillation problem increase with lower voltage, so I do not think the voltage derating is the major problem.

Your equation is for input capacitor requirement is mainly for voltage ripple.  But in my case it is a stability issue.  The fact is that if I turn on each output independently, there is no oscillation.  It only oscillation if there are high loads in at least two channels.

My capacitance being used is Samsung CL31B106KAHNNNE.  Attached is my schematic.

Samsung

CL31B106KAHNNNE

Hi Henry,

Yes, the 10uf u used is 25V but the DC bias looks not good, alomost over 60% derating. the real value maybe is only <5uf@12V and so more caps is needed.

and the draft calculation is for deltaV, if deltaV is very big, then which indicates the Vin will goes down a lot.

Two channels could cause oscillation, because more power need be provided from Vin, so that make sense.

Finally, I see there is a current sense resistor in the PVin loop, did you try to remove or short it and re-test to check if there is any promoted?

also remind that make sure the input power has large current limited ability to provide enough current instantly.

yuchang

Hi Henry,

Add three more advice,

1, Did you try the same cap as our EVM mounted, if the phenomenon is same?

2,Did you disconnect R251, U202 and add external 12V power supply to PVIN to double check the oscillation?

3, BTW,  the PVIN cap should be placed colse to PVIN pin,

and if convienient, please upload your waveform, both yours and our EVM you tested.

BR,

yuchang

Hi Yuchang,

You got it!  Once I connect the power directly to the PVIN pin, the oscillation goes away.  Then I suddenly realized that I had an input LC filter to this circuit before the R251 resistor, and I was supplying power through the LC filter.  The LC is 10uH and 40uF which calculates to a 8kHz resonant frequency, then once it coupled with the negative input resistance of a buck converter, it became a perfect oscillator.  Now even if I connect the power directly to the LC filter output, there is no oscillation.  This means that it is the 10uH input inductor that caused the problem.  But for the design, I have to keep the inductor, so I have to really push down the input filter corner frequency by replacing the 40uF with something much bigger.  Maybe you can point me to some application note to show how to select the proper input filter to avoid this problem.  Again, Thanks a lot for your suggesstion.

BR, Henry

Hi, Herry

:), it really is LC oscillation.

1. What is the function of inductor?
2. To avoid this problem, the best way is increasing input capacitor.

thanks.

zhao

Hi Zhao,

The inductor is for another input stage of a forward converter.
So increasing the input capacitor (output capacitor of the LC filter) does help. But it still needs to be a very large. What I wish to get is the quantative capacitance value to get enough margin to avoid the oscillation. As I realized that a buck converter behaved like a negative resistor so I should set the LC filter corner to be much lower than the negative resistor of the buck. Is there a theoretical calculation to give a rough calculation of the frequency response of the input impedance of this buck converter?

Hi, Herry

The buck converter input impedance = -Rload/(D^2). (Rload is the load of buck.)
so the model can be simplified to LCR circuit.

Zhao

Hi Zhao,

Your equation of Rload/(D^2) seems to be a DC resistance equation if we assume D does not change.  But the fact is that D will change with Vin.  What I am looking for is a small-signal equation for the impedance of this buck converter.  I think this equation should be affected by the design of the compensation netwrok and by the design of the chip itself.  Once I have this equation, I can figure out how to design the input LC filter to avoid oscillation.

Henry

There is a good reference here: Literature Number: SNVA538 - Input Filter Design for Switching Power Supplies

Yes, you are right. The minimum input impedance is when D is 100%, so 4V/12A = 0.33ohm. According to another paper I found in the internet, the input capacitor must be greater than L / R^2, so with my 10uH input inductor, the capacitor must be at least 100uF.