WEBENCH® Tools/LM5085: Webench-Design gets hot at little loads

Dominik Lorenz

Part Number: LM5085

Tool/software: WEBENCH® Design Tools

Hello there!

I am debugging my design(s) for a while not and still have no clue why it is getting so hot.

There are 2 circuits i designed with the webench tool, both utilizing the LM5085. The first, Vin=43-58V Vin and 24V, 2A out got too hot and isn't working anymore. So this thread will be more about the 2nd one, also 43-58V Vin and 4.3V 1.8A Vout.

All in all, i sticked closely to the parts suggested from the Webench exept the Mosfet, which is the automotive-Version of the suggensted one. Also the 1uF Cin is a different Part number, but of course also with similar specs.

Now, some parts are getting really hot immediately after turning it on, on a load about 130mA. "Some parts" in this case are: The Diode and the Mosfet, way too hot to touch. The LM5085 get also warm after 20sec or so, but im not sure if its becuse of the good thermal coupling of it and the diode of if it gets hot by itself. When letting it run for some minutes, the Inductor an other parts are also getting hot, but again, I'm not sure if its the heat from the diode and the Mosfet.

Here the schematic: (I put the part numbers next to the parts, the text in German is not important)

I think the most important piece of information for you are the waveforms:

Upper Pic: Vdiode in yellow and VCout in red

Upper pic: Vds

Upper Pic: Ufb

Upper Pic: Ushunt (Voltage across the Shunt, not to GND)

My question is, why the parts are getting so hot and what I can do about it. I feel like I'm missing something obvious. Im pretty sure that a higher load would burn the circuit.

Thank you for your help!

Dominik

over 7 years ago

0 Samuel Jaffe over 7 years ago

TI__Mastermind 25590 points

Dominik,

I have that exact same scope at home! Good choice!

The schematic looks good. The waveforms show a steady switching frequency/duty cycle which is good. The Vds goes down to -10V which indicates to me excessive inductance in the switching transient loop which would be caused by layout. Thermal issues point to either improper operation or a layout not optimized for thermals.

Please share the layout so I can take a look.

Thanks!

0 Dominik Lorenz over 7 years ago in reply to Samuel Jaffe

Intellectual 255 points

Samuel,

thanks for the reply! Following the layout. As the Datasheet suggests, i tried to stay on the component side as much as possible, so on the other 2 signal layer are only 1 trace each. There is a Ground-plane to which the Vias under the LM5085 are connected to, but it doesn't carry any currents. Not sure if that would be a good option or not.

Components side:

Inner layer 1:

Bottom layer:

a 3D view:

Again, thanks a lot for your help. You guys are always very helpful!

0 Samuel Jaffe over 7 years ago in reply to Dominik Lorenz

TI__Mastermind 25590 points

Dominik,

Thanks for the layout.

I took a look back at the schematic and the diode datasheet doesn't say it's good for switching regulators. Did WEBENCH recommend this diode? And can you try swapping it with an ultra-fast recovery schottky diode?

The layout looks pretty good. I have a few comments:

The GND plane can be much larger. The IC will dissipate its heat through the thermal pad and we want a large unobstructed GND plane for that heat to spread. You can extend the plane more to the left, much more to the right, and on other layers (especially the bottom layer).
The GND return path from C77 could go directly to a large plane instead of a long trace.
Transients during SW rise/fall time should have as small a loop as possible. The path from COUT back to CIN is long and could be shorter if we utilize other layers or shifted components.
You could route the VOUT (-->FB) trace farther from the SW node, or add GND planes on mid layers for shielding.

Other than that I think it looks pretty good.

-Sam

0 Dominik Lorenz over 7 years ago in reply to Samuel Jaffe

Intellectual 255 points

Samuel,

thanks a lot for the advice! Your're right about the Diode. I try swapping it. It was suggested by Webbench, I attached a screenshot that may show the problem.