TPS54540B: Duty cycle instablity at high load.

Lionel U

Prodigy 80 points

Part Number: TPS54540B

Hello

I'm using the TPS54540B to get 3.3V 5A from a 24V nom input power supply.

Vin max=28.8V

Vin min=19.2V

Here is the schematics used :

I've checked the open loop gain and phase margin here is :

And the transient behavior (Vin=nom, Iout from 1A to 5A) :

So for me my design is stable.

But when I look at the duty cycle at Vin=max and Iout=5A, I've got instability cycle to cycle :

(green = Inductor current, yellow = SW pin, purple = Vout in AC)

With statistics to get the sdev of the duty cycle

I exactly follow the layout of the EVM. I don't think it's a noise issue. I've changed my input power supply, used big electro cap to filter it. The behavior is exactly the same.

I think it's due to the internal slope compensation but I'm not an expert.

Can you tell me if this phenomena is normal or not ?

If the output current is reduced to 4A at Vin=max, the stdev of the duty cycle drops to 1%.

If the output current is reduced to 3A at Vin=max, the stdev of the duty cycle drops to 0.3%.

Thanks for your support

7 months ago

0 Marshall_Beck 7 months ago

TI__Mastermind 34790 points

Hello,

Thank you for sharing your collected data.

I do agree that you have a great deal of phase margin. Transient doesn't look to be ringing what is characteristic of an unstable design.

E2E has reduced the resolution of your images but your cycle-to-cycle variation doesn't appear to show signs of instability to me.

For SHO, often you would see a combination of small and wide PWM pules and that would occur at duty cycles >50% which neither of those appear to exist.

The inductance which is important for fixed slope comp is what is also important. It seems reasonable for your output voltage and fsw.

0 Lionel U 7 months ago in reply to Marshall_Beck

Prodigy 80 points

Hello

Thank you for this reply. For me, the duty cycle seems to be not stable. Attached full resolution picture.
Indeed, the stats give DC (mean) = 16% and a stdev of 3%. The variation is quite important.

And the pattern is repeatable. The change is cycle to cycle.

If, I change a little bit the compensation (fc lower) and I've got the same behavior.

Do you think the inductor should be changed ? Or changing the switching freq ?
3% of a switching period (1/699kHz) is close to the delay of the current limit. Is there any relation ?

We just want to fully understand this phenomena before releasing a board.

Thanks for your reply.

0 Marshall_Beck 7 months ago in reply to Lionel U

TI__Mastermind 34790 points

Hi,

You are welcome

Could you please provide high-resolution captures of the data?

Please make sure that they show up correctly in the e2e thread.

If it is an acutal slope comp situation, I would like to emphasis again that it only occurs at greater than 50% or greater duty cycle which doesn't appear to be the case in this instance hence my earlier conclusion. If you have slope comp issue it would be due to too low of an inductance at your given frequency when operating at duty cycles greater than 50%.

https://www.ti.com/lit/pdf/snva555

https://www.ti.com/lit/an/slvae22a/slvae22a.pdf?ts=1712840933319&ref_url=https%253A%252F%252Fwww.ti.com.cn%252Fsitesearch%252Fen-us%252Fdocs%252Funiversalsearch.tsp%253FlangPref%253Den-US#:~:text=Slope%20Compensation%20(Se)%20is%20an,inductor%20current%20falling%20slope%2C%20Sf.&text=Ri%20is%20the%20current%20sensing%20gain.

0 Lionel U 7 months ago in reply to Marshall_Beck

Prodigy 80 points

Hello

Sorry for the delay.

Here are the link to have pictures in Hi Res. It's the only way I found.

https://photos.app.goo.gl/2tAXLX9pXv6wGcDc9.

Thanks for your support

0 Marshall_Beck 7 months ago in reply to Lionel U

TI__Mastermind 34790 points

Hello,

That does look like a slight varying duty cycle, but subarmonic oscillation occurs only at >50% duty cycle. With the result of your bode and transient plot, your design appears to be stable to me.

0 Lionel U 7 months ago in reply to Marshall_Beck

Prodigy 80 points

OK thanks for your confirmation

0 Marshall_Beck 7 months ago in reply to Lionel U

TI__Mastermind 34790 points

You are welcome.

Because of the Thanksgiving holiday in the U.S., TI E2E™ design support forum responses may be delayed from November 25 through December 2. Thank you for your patience.

Power management

Power management forum

TPS54540B: Duty cycle instablity at high load.