L6981(C50DR) dies on hard switching of input power

[HMüll.4]

Hello everybody,

I'm using an L6981C50DR for supplying a small board with 5V. The schematics is attached and should be perfectly conform to what is described in the specification from https://www.st.com/resource/en/datasheet/l6981.pdf as "Figure 4. Basic application (fixed VOUT versions)" on page 3. I'm supplying 24V to it. Everything works well when I turn on and off the power via the switching power supply that provides these 24V. But when I reboot my board the "easy way" by pulling out and reconnecting the +24V wire, the L6981 dies immediately. After this, pin 4/Vout is permanently at GND letting me assume something has melted within the IC.

When measuring the switching power supply with an oscilloscope, it provides a small peak in such a scenario but this peak does not go beyond of 28V - which is far below the 40V the L6981 should be able  to deal with.

So...any ideas what the reason for this problem is and how to solve it?

Thanks!

[Peter BENSCH]

With switching regulators, it is always helpful to include the layout as an image.

However, in this case, I am also certain that your problem is due to an unsuitable coil L3. The keyword here is saturation current, which is 1.3A for your NRS6028T 330MMGJ.

I quickly simulated your design in the eDesignSuite and came up with peak currents of around 1.66A, which are well above the limit. Even if you may not need the 1.5A permanently, the output capacitor C38 must be charged, which the controller wants to do enthusiastically and at full current – until the magnetic field collapses when the saturation current is reached and only the DC resistance is effective.

So you will have no choice but to dimension the coil accordingly (recommendation: saturation current > 2A, better 2.5A).

Hope that helps?

Regards
/Peter

[HMüll.4]

OK, the inductor was not the reason. I replaced the inductivity by a SDR1006-330KL which has a saturation current of 2,9A which should be more than enough.

First try was with 15V power supplied. here randomly connecting the power did not lead to a destruction of the L6981C50DR but a second stage ST1L05APU33R was destroyed immediately. So it seems there is some kind of dramatic peak on VOUT-side of the L6981C50DR 

Second try was with 25V power supplied. Same result as before, the L6981C50DR was destroyed immediately (this time again with a loud nose like a flash-over.

The layout is attached, the L6981C50DR part is highlighted. Top side is red, bottom side is blue.

[HMüll.4]

OK...I'll replace the inductivity, try again and let you know!

[HMüll.4]

After some further testing it turns out the ST1L05APU33R-part is the problem. I replaced the whole 5V-stage with the L6981C50DR by a Traco Power TSR 1-2450E DC/DC converter. Now the situation is, the 5V-part survives random power cycles at the 24V input without problems. But the 3.3V part which is located behind the DC/DC converter still dies! Means even when the TSR 1-2450E still provides 5V after some power cycles, the 3.3V rail is dead.

[Peter BENSCH]

Well, it would probably be quite complicated to trigger a voltage spike that destroys the ST1L105 at its VI and have to replace it constantly, wouldn't it?

The problem can certainly be exacerbated by an unfavourable layout, but it would still make sense to connect a TVS to the VI of the ST1L105.

I would also like to take this opportunity to point out that most LDOs, especially the ST1L105, have quite low noise rejection for higher input frequencies. In this specific case, the L6981C50DR operates at 400kHz, for which it is not even possible to estimate where the PSRR might be in the data sheet, fig. 19 of the ST1L105. If you require high ripple rejection, you should install an LC filter upstream of the ST1L105, which should also attenuate any high voltage spikes that may occur during switch-on.

Regards
/Peter

[HMüll.4]

Quick update: I have an other board which makes use of exactly the same schematics and components for 5V and 3.3V power, which makes use even of the NRS6028T 330MMGJ but has a different layout. This board is working fine, I can apply 24V power randomly and nothing happens. So this seems to be some weird layout effect...