LM2576HVT-ADJ at >50 inputs voltage - chips are damaged

An important update - at up to 48.5 Input voltage, all works good... I commited about a thousand power-ons and offs under load over a week and it works just fine.

Thank you!

Hi guys,

an update to the case... in the Datasheet for the LM2593 I found the following in regards to the 'Inductor selection':

"...

The Energy values shown on the nomographs apply to steady operation at the corresponding x-coordinate (rated maximum load current). However under start-up, without soft-start, or a short-circuit on the output, the current in the inductor will momentarily/repetitively hit the current limit ICLIM of the device, and this current could be much higher than the rated load, ILOAD. This represents an overload situation, and can cause the Inductor to saturate (if it has been designed only to handle the energy of steady operation). However most types of core structures used for such applications have a large inherent air gap (for example powdered iron types or ferrite rod inductors), and so the inductance does not fall off too sharply under an overload. The device is usually able to protect itself by not allowing the current to ever exceed ICLIM.

But if the DC input voltage to the regulator is over 40V, the current can slew up so fast under core saturation, that the device may not be able to act fast enough to restrict the current. The current can then rise without limit till destruction of the device takes place. Therefore to ensure reliability, it is recommended, that if the DC Input Voltage exceeds 40V, the inductor must ALWAYS be sized to handle an instantaneous current equal to ICLIM without saturating, irrespective of the type of core structure/material.

..."

So guys, the LM2576 operates at three times as low frequency compared to LM2593 and also has a more primitive overcurrent protection... so can it be, it is not 'fast enough' to recognize overload condition due to inductive spikes at saturation?

Consider the following scenario:

- No 'soft start'

- 1000 uF 100V LowESR Capacitor

- High Vin

On the power-up, the switch is closed and current inrushes into the inductor. Since it is rated 1.8A, it is easily saturated by >3.0A switch current and after saturation it behaves like simple resistor of very low value. At this point the output capacitor is still at low voltage (compare energies stored in the capacitor and the inductor), so difference in voltage with very small resistance produce overcurrent at so hight dI/dt that 50 kHz is not fast enough to sense before it actually burns out...

So, if this scenario is right (your opinions and suggestions?) the solutions might be:

- move to a higher frequency

- use IC with Soft-start (LM2593 etc.)

- or use a massive core inductor with lots of ferrite :)

Your feedback will be appreciated!

Sergiy 

Comments? A word? Anything? ;)

Hi Sergeii,

So sorry for the delay in getting back to you. We are looking into your post and will get back to you shortly.

Thanks,

Anston

I would try using an inductor that has a current rating at least equal to the current limit of the device.

FD

...sure, I would say even more - exceeding the maximum switching current of the chip. However, I faced certain problems in getting coils rated above 1.8A @ 330 uH here. So I used 2 in parallel, naively thinking it will help. Now I use powedered iron toroid core, wound manually, which should be by factor 2.1-3.5 resistant to saturation and distributed air gap. The problem is, I don't have enough expierence with formulae and lacking of info about core (or both), so I cannot estimate the I sat mathematically.,. and cannot measure it experimentally, because this would require quite powerful pulse generator.

The most practical advice you could give is how do I combine standard coils values with rated I sat to get the value I need, with I sat satisfactory to fit the application. The problem is, HOW the standard coils are rated. For example, in my case when I first tried M33 ferrite power coil rated 1.8A, I discovered later that in fact, they have at this current only half of their inductance! Is this normal way to rate? So how many of standard 1.8A 330 uH I have to use in parrallel and in sequence, to get 'virtual' 3.5A 330uH which actually has this inductance at this maximum 3.5A?

Thank you for coming back to my topic!