LM5088: webench 48V to 12V 8A switcher, excessive heat?

Hello,

Please can you take a scope measurement of the Switch node and also the Gate drive signal? If there are some anomalies with the gate drive signal and the FET is not being driven correctly, this will certainly cause some excessive heating in the FET.

The main components getting hot are the inductor, schottky diode and fet.
I was pondering going towards a synchronous switcher for this to reduce the heat induced by the 0.5v voltage drop across the diode. But before this, I need to see if this circuit can work for now.

Let me see if I can include images in here.. 

Hi, thank you for your quick response!

Below is the layout, ignore that IC up in the far top right hand corner, I've only got the switcher parts on the board so far.

Below is a close up of the gates waveform with no load

here with a 3R3 restive load

The blue trace is the gate voltage.

Yellow is the 12V output

Below with no load

3R3 load

here with a 4R4 load, red trace on fet source pin

From you Gate drive waveform and Source (Vswitch node) waveform, it appears to me the FET is getting enough drive to hard switch the FET. However looking at the waveform with 5V per division for Vout, it looks like you have a huge oscillation. I think your loop maybe unstable? Please can you reduce Rcomp from 38.3k to 10k keeping everything else the same? Thanks.

ok, replaced with 10K

Do you have an actual Schematic of your board? Please can you lift the Inductor and measure the inductor current also? Thanks.

hi, sure I've got a schematic. It (hopefully), is the same as the one generated by webench, though. Please see below.

Thank you for help with this by the way, much appreciated!  :)

Please, don't hesitate to ask if I can provide any other helpful information

Sorry, I mean, lift one side of the Inductor and place a loop there, so a current clamp can be used to measure the inductor current. 

ah, I don't have a scope current clamp
Only a stand alone hand held meter.

Maybe I could lift one side and put a few ohm resistor in series then measure the voltage change across that?
I'll see what I can do.

Thanks

Hi, sorry for the delay in posting, I've been away snowboarding for quite a while :)

Replacing the suggested 47uF X5R SMD ceramic output capacitor (GRM32ER61C476ME15L), with a through hole 22uF electrolytic cured the gate drive waveform issue. See pic below.

Could this be from a too low ESR from the original capacitor?

If so, any recommendations for a ceramic SMD substitute? Or should I add some small about of series resistance?

Many thanks

Hello,

It appears your power supply was oscillating due an input LC filter formed and it required damping.  Long cables make up a parasitic L and if you have an input capacitor at the input with very little ESR, this will cause ringing and cause the downstream supply to be unstable due to the output impedance of the input filter formed as it relates to the input impedance of the power supply.  Damping the LC filter is at the input to a power supply is often implemented to stabilize the power supply.

No problem, schematic is still as before and as per suggested TI design that should be shared on here somewhere on this thread.

I'll get the layers in the following post, thanks

Here we are, four layer board, 2oz copper.

Top copper and screen print:

Inner1 layer and screen print:

Inner 2 layer and screen print:

Bottom copper and screen print:

How's that?

I use Proteus for schematic and layout, let me know if I can send them in a more user friendly format.

Thanks

Reviewing your schematic I see you only have a 1uF input capacitor?  The IRMS current will need to be satisfied by Cin.  Suggest an IRMS rating for Cin to be Iout/2 for 50% Duty Cycle (worst Case).  If you Iout = 8A this will mean you need an RMS current rating of 4A.  The 1uF cap is probably undersized? If you you want a more accurate calculation for CIN IRMS, the formula is as follows. ICINIRMS=Iout*(Sqrt(D*(1-D)), where D = Vout/Vin. For 48V in and 12V out and 8A of load current, the CinIRMS = ~3.5A, Suggest placing 2 x 4.7uF ceramics at the input to your supply. I also would suggest a Cin Bulk Electrolytic with a higher ESR (~0.2ohm or so) for damping the input to the supply.

Now to the layout. I think you Vcc cap (C17) is too far away from the IC, this should be as close as possible to Vcc Pin and Gnd. Also your Feedback Resistors(R28/R29) are too far away from the IC. The FB node need to be as close as possible to the FB pin of the IC.

One final point about the layout is that you CS and CSG routing to the Rsense (R32) needs to be routed differentially.

Sorry, not sure why I got stuck in a previous post.  I am glad you are making progress!