LM5088: LM5088-1 Failing, needing to be replaced.

Hello Jerome,

Looks like you are switching at a frequency of ~500kHz which yield gate drive currents of 24mA.  with a Vin of 48V the power dissipation of the Device will be greater than a Watt of power.  I suggest lowering the switching frequency to ~300kHz. 

Another idea is to supply a voltage to the VCC externally through a Diode to reduce the power dissipation in the device.

Can you please take some scope shots of the Switch node before failure?  Want to make sure your design is stable.  Thanks.

David.

So do you think this is purely a over heating issue?
If so why is that? And does the LM5088 not shut down in the case of over temperatures rather than sustain permanent damage? 

I'm just working on getting better scope images.

Here is one I've captured while doing the previous test, it's not great quality.

I've made a video of the last test, is there a way I can send this to you?



This was just before LM5088-1 failure 
I'll work on getting a better picture 

Here we show the four fans running at minimum 

Instead of the PWM fans, here is a purely resistive 3R3 Load. But as mentioned, increasing this purely resistive load, (reducing resistance to 1R65), also causes LML5088 failure

Here's HG with no load

Here I've changed RRT to be 20K instead of 10K7 (this should be 300Khz?)

Hi, i've made the change of these two components as shown in below scope images, but unfortunately the LM5088 now seems to die quicker, (I'm getting through quite a lot of these devices now).

I could send you a video of the scope when the LM5088 fails if it helps?

Here I'm going to include screenshots of the point at which the device fails from minimal load to 3R3 and then 1R65 resistive loads.

When the 1R65 load is applied the device becomes terminally damaged within a second.


Thank you for your help with this.



Minimal load.

3R3 Ohm load

1R65 Ohm load

LM5088 beginning to fail within seconds

LM5088 continuing to fail (a few frames later)

LM5088 failed completely

 

So as you can see that seemed to make the LM5088 more prone to failure. Is there any idea as to why these LM5088's are failing?
Thanks

The Scope Shots show that the loop is still unstable, even though you triggered and stopped on a stable region?  I still think you have a stability issue.

Can you obtain a large Electrolytic capacitor with a high ESR (say 0.5ohm) and place it at the input.  I am suspecting you have long leads adding parasitic inductance to the input that may need to be damped.

use a 100uF or so.

Hope this helps?

David.

Hi, apologies the test circuit didn't match the schematic (although it does match the Webench output).
Where it showed the input capacitors as 2.2uf + 2.2uf + 1uf it was only 1uf. (this matches the Webench output)

When adding the 100uf electrolytic the output became less stable if power was removed then replaced.
This is what happened

Noticing this I removed all input capacitors
This resulted in the LM5088 physically exploding with the application of input voltage.

And so I replaced the 1uf along with the two 2.2uf capacitors and the LM5088 has not failed since, so far.

1R65 load

3R load

minimal load

Should the minimal load gate waveform look like that?
Should it be a clean square wave on there instead of what you can see here?

Regarding the output capacitance. I notice Webench keeps changing its design. 
If I go through the same procedure now, it shows much larger output capacitance than it showed originally.