TPS54531: TPS54531 Startup Issues

Brian Wissel1

Part Number: TPS54531

Hello,

We are having sporadic startup issues with a design that implements a TPS54531, this design takes 5V to 3.3V with an average load of ~2A.

Cin = 9.4uF (x2 4.7uF in parallel) ; Css = 8.2nF ; Rcomp = 51.1k ; Ccomp = 820pF ; Ccomp2 = 15pF ; L = 3.3uH ; Cout = 200uF

The startup wave form will ramp as expected up to 3.3V then after a short period of time (~1ms) the output of the regulator becomes unstable and starts oscillating between ~2V and ~3V every 500ms or so.

If I change the L to 2.2uH from 3.3uH then everything functions as expected, all else remain the same. Using web bench set to balanced I get the following bode results:

Low F gain Cross F Phase Marg. Gain Marg Inductor Val

29.83 dB 24.02 kHz 47.09 deg -17.11 dB 3.3 uH

29.98 dB 26.82 kHz 54.99 deg -17.39 dB 2.2uH

Is there something we are violating by using the 3.3uH inductor? We are seeing the failure in roughly 5% of the boards that come off the line.

Thanks,

Brian

over 3 years ago

0 Ruby Li over 3 years ago

TI__Expert 7520 points

Hi Brian

Would you mind sharing me the waveform of the Vout, PH, and inductor current when using 3.3uH? I would like to see what happens here.

Thanks.

Ruby

0 Brian Wissel1 over 3 years ago in reply to Ruby Li

Prodigy 30 points

Ruby,

Here is the Vout (C1) vs PH (C2) capture, might take me a minute to get the L current capture.

Regards,

Brian

0 Ruby Li over 3 years ago in reply to Brian Wissel1

TI__Expert 7520 points

Hi Brian

Thanks for the quick feedback. Would you please change the boot cap value to 470nF when using 3.3uH inductor and check whether it works?

Ruby

0 Brian Wissel1 over 3 years ago in reply to Ruby Li

Prodigy 30 points

Ruby,

I have better results but the output at startup still drops as before. However instead of oscillating it becomes steady after the initial droop.

Thanks,

Brian

0 Ruby Li over 3 years ago in reply to Brian Wissel1

TI__Expert 7520 points

Hi Brian

Thanks for the feedback. I suspect that when using 3.3uH inductor, the inductance resonates with the diode's capacitance after high side FET being off. The resonate frequency is smaller than that using 2.2uH, which may make the PH voltage not got down to zero. This will cause the charge of boot cap to be not enough to keep high-side FET on. If you could share me the PH voltage waveform detailly when there is oscillatiion between ~2V and ~3V, I think we may see the ring after high side off. You may also make Vin a little bit higher to check whether the device work properly.

If what I think is correct, to handle the problem, we can enhance Vin a little bit up to 5.5V, or enlarge the boot cap (1uF), or change the diode with smaller capacitance or reduce the inductor value.

Hope this helps.

Ruby

0 Brian Wissel1 over 3 years ago in reply to Ruby Li

Prodigy 30 points

Ruby,

What you are saying makes sense based on what I have seen. The attached images show the part trying to start up every ~500ms with a zoomed in view (ph_1.jpg) and a close up view of the oscillation (ph_1.jpg). If you need something more specific let me know, or if you want that referenced to anything else.

These were captured with the original SS cap re-installed, 8.2nF and 3.3uH L.

Thanks,

Brian

0 Ruby Li over 3 years ago in reply to Brian Wissel1

TI__Expert 7520 points

Hi Brian

Thanks for the feedback. I think we get the root cause now.

Ruby

0 Brian Wissel1 over 3 years ago in reply to Ruby Li

Prodigy 30 points

Ruby,

You mentioned increasing the boot cap value to 1uF, would that be better to implement along with the lower value inductor?

Thanks,

Brian

+1 Ruby Li over 3 years ago in reply to Brian Wissel1

TI__Expert 7520 points

Hi Brian

Increasing the boot cap can make 3.3uH also be applied to the application if the requirement of output ripple is strict and customer does not want to reduce the inductor value. But if customer can accept the larger output ripple, the 2.2uH inductor can be a good solution here.

Hope this helps.

Ruby

0 Ruby Li over 3 years ago in reply to Brian Wissel1

TI__Expert 7520 points

Hi Brian

I think I misunderstood your question. I would recommend to add the boot cap value to 220nF or 470nF when using 2.2uH. No need to 1uF.

Ruby

+1 Brian Wissel1 over 3 years ago in reply to Ruby Li

Prodigy 30 points

Ruby,

Thanks for the clarification.

Regards,

Brian

Power management

Power management forum

TPS54531: TPS54531 Startup Issues