LM5013-Q1: LM5013 Thermal management

Hello Fabio 

Are you testing standard LM5013 EVM  ?  What was the highest IC temperature you observed before the device is  damaged ? 

Adding 2 more layer will be better if you have good via connections from the DAP/EP of the device to the internal & external LARGE ground copper polygons. 

-EL 

I'm testing my design. The top of the device case reached about 85°C

Hello Fabio 

If the temperature is only 85°C, I don't think the device is damaged by a thermal issue. Please share your schematic and V(SW-GND) waveform for more investigation. 

-EL

May be the diode the problem? When the temperature reach 80/85 °C the output start flashing (I use some lamps as load) and after few seconds more it fault!
Please find attached the scheme, layout and some waveforms pictures (CH1: Vout, CH2 Vsw)3527.Waveforms.zip1856.Board.zip

Hello Fabio 

I guess C3 is causing the problem. C3 should be 2.2nF (not 2.2uF) . Also, the diode temperature might be too high. Please check the diode temperature too. 

-EL 

Oh, damn! This is a very poor error!
I tried soon with a capacitor from my laboratory hoping in a better efficiency and therefore a lower functioning temperature, but it seems to work at the same way (with the same problem); let me get a new better capacitor to repeat the test.
Many thanks.
Fabio

Hello Fabio

If possible, please capture a high resolution V(SW-GND) waveform using pigtail method. https://www.ti.com/lit/ta/ssztb25/ssztb25.pdf 

-EL 

Hi Eric,

it seems that with 2,2uF bootstrap capacitors the regulator works better because increasing the load, with 2,2nF it do not regulate well.
I attached some pictures of Vi, Vo and Vsw in different conditions: 2L means 2 Loads and about 1,8A, 3L means 3 Loads with about 2,6A
On Vi-Vo pictures the Vi is the CH1 (lower track) and Vo the CH2 (upper track) 
Vsw seems to work in a similar way
Vsw and Vi-Vo.zip

Hello Fabio 

The waveform looks okay. I wish you can check the inductor saturation at full load. Are you able to duplicate the damage? I guess the device might be damaged by a mis-handling......  

-EL

I got the Vi/Vo waveform from the first modified board (Board 1) with the new Cbst = 2,2nF at 2,6A load, and shows the regulation problem.
Then I got a new board (Board 2) with the original Cbst = 2,2uF with the same 2,6A load and it regulates correctly the Vout.
Then I excanged the Cbst between the two boards and:
- Board 1 whith the 2,2uF return to work well
- Board 2 with the 2,2nF shows the same regulation problem
This seems to confirm that this regulation problem at 2,6A is due only to the Cbst value.
You correctly noticed the wrong value used for Csbt but this seems to open a new problem, not to solve the first problem
Now I don't know how to proceed to investigate and solve the problem, I'm thinking to buy the evaluation board, try it in the same conditions and then, if it works correctly modify it once component a time while reaching my scheme verifying when the problems occurr.
P.S.: I attached the updated scheme because I increased just alittle the Vout value (just to be exactly alligned with the waveforms, I don't think it can represent a significante difference)
8524.Attachments.zip

No, the 2.2uF Cbst is the "initial" condition, the scheme I realized and that shows the damage issue.
When you noted the Cbst wrong value I tried to modify it but it starts to show the Vo regulation problem.
Wher I tester the 2.2nF Cbst I turned off the power when the regulator reached 85°C to avoid another device burned but it seemed to increase the temperature aas fast as before, do you believe to use the 2.2nF can avoid the regulator to burn? I could try

Ok, we are going to assembly a new lot of boards with some improvements for a better thermal dissipation, but how can I avoid this behaviour?
I attached 6 picture of Vout (up CH2) and Vin (down CH1) with different output loads for 22nF and 22uF Cbst:
- with 0,9A and 1,8A Iout both boards works well
- with 2,6A load with 22nF Cbst the Vout is not well regulated when with 22uF Csbt Vout continues to be well regulated (no differences in Vin)
- the hi-frequency signal (Vsw) seems to work at the same way in any condition, even when the Vout is not well regulated
Any idea about the origine of this problem and how to try to avoid it?

Vin-Vout.zip

Hello Fabio 

Please compare V(BST-SW) and V(SW) waveform with 2.2nF(not 22nF) and with 22uF. Please check both zoom in and zoom out waveforms. I guess V(BSW-SW) is not enough when VOUT drops. 

-EL 

When I verified Vsw and Vbst I noted that they was practically identical, but I didn't verified the Vbst-sw. Unfortunally last time the last board burned therefore now I can't verify this signal (I have to wait the new production).
How much shoud be the minimum value of this signal?
If your hipotesys is correct, how can I increase this signal voltage?
Many thanks.
Fabio

Hello Fabio

Please give me the update when the new board is ready. 

BST UVLO spec is in the datasheet. It is 3.4V 

We have to check the V(SW) waveform and find the root cause 

-EL