LM25148: unstability

pw06042023

Part Number: LM25148
Other Parts Discussed in Thread: LM62460, TPSM63606

Hi,

I've designed a new board with LM25148 . I follow the Figure 9-17. Application Circuit from datasheet to do a 3.3VDC/4A power rail.

I choose the following component :

Self : 2.2µF/15A / 7mOhms max
Cinput=2*47µF/50V chimical low esr and 2*2.2µF/50V ceramic
Coutput = 3*47µF ceramic 10V
PFM/SYNC=VDDA= diode emulation mode
VCCx=0
Compensation = approximatively same as evaluation board (10K+4.7nF)//18pF.
Mosfet : QG=15 to 20nC / Cg=1100pF / VDSS=40V / Rdsmax=30mOhms
Feed back = RFH = 150K ; RFB=47K (tested with 100K to VDDA too) for 3V3 output
Cboot = 100nF/50V
Cvdda=1µF/25V
Cvcc= 10µF/10V
En = High = 2.8VDC
CNFG=41.2K
RT=41.2K for 525KHz switching
no low pass filtering on ISNS+ (only kelvin connection)

For Iout = 0A to 250mA everything works fine (3V3 with 40mVpk ripple).

For Iout > 250mA, I've a Vout from 2.8VDC to 3VDC with 500mVpk ripple which seems to be unstable (unlike VDDA and VCC seems to be ok).

I try with the internal compensation too (with 100K to VDDA)

I got the following HO output (blue) and LO (yellow).

I use the excel snvr504b to check design and I didn't see any mistake.

The design is similar to the design generated by the TI webench tool.

Do you've an idea on my mistake ?

Thanks.

over 2 years ago

0 Orlando Murray over 2 years ago

TI__Mastermind 22302 points

Hello,

Can you attach your quickstart calculator and schematic?

Can you confirm stable operation at no-load with FPWM?

Also what is your current sense resistance?

Any reason why you are using a controller instead of a converter with integrated FETs?

3.3V 4A output should easily be achieved with LM62460 converter (integrated FETs), or even TPSM63606 module (integrated FETs and inductor).

Hope this helps,

-Orlando

0 pw06042023 over 2 years ago in reply to Orlando Murray

Prodigy 10 points

LM5148-LM25148 Quickstart Calculator_rev3.xlsm

You'll find attached the excel calculator.

current sense resistance : 10 mOhms.

I try with 525KHz, 1MHz and 2MHz switching frequency.

Due to large Vin, high current and high operating temperature, integreated mosfet solution is not possible

unstable operation with load in FPWM.

0 Orlando Murray over 2 years ago in reply to pw06042023

TI__Mastermind 22302 points

Can you attach your schematic?

Also can you take the scope shot of no-load FPWM?

0 pw06042023 over 2 years ago in reply to Orlando Murray

Prodigy 10 points

3v3_rail.pdf

attached the schematic of the 3v3 rail

0 pw06042023 over 2 years ago in reply to Orlando Murray

Prodigy 10 points

Scope at noload FPWM (HO:yellow LO:blue)

0 Orlando Murray over 2 years ago in reply to pw06042023

TI__Mastermind 22302 points

Hello,

Your switching frequency is coming in correct, however the early HO termination in FPWM should not be happening.

Right now your FPWM looks like DEM, which implies that the device may think current limit is being hit.

Also note the device starts up in DEM, the FPWM waveform should be taken after 3ms startup.

Can you probe the inductor current with a current loop?

I want to see SW node, inductor current, VOUT and maybe COMP.

Your schematic looks OK, maybe CNFG resistor should be closer to 30kΩ.

Please share your layout as well.

Did you try your design on the EVM layout?

https://www.ti.com/tool/LM25149-Q1EVM-2100

-Orlando

0 pw06042023 over 2 years ago in reply to Orlando Murray

Prodigy 10 points

comp :

vsw :

vout :

0 Orlando Murray over 2 years ago in reply to pw06042023

TI__Mastermind 22302 points

Hello,

Are you able to have COMP, SW, and VOUT on the same plot?

It looks like your COMP is not stable, it should be a smoother waveform. Double check your COMP components.

Are you sure this is FPWM?

Is this taken after soft-start?

Please also share your layout.

-Orlando

0 pw06042023 over 2 years ago in reply to Orlando Murray

Prodigy 10 points

Hello Orlando,

I bought an evaluation board and It seems we have to add a filter on sense pin... -:) .

But I have an another question:

when we cut the power supply with a simple 4A resistive charge, sometimes LM25148 is destroyed. In the datasheet page you say "Care must be taken in applications where there are fast input transients that cause the voltage at VIN to suddenly drop more than 2 V below the VOUT setpoint. The LM25148 has an internal ESD diode from the VOUT to the VIN pins that can conduct under such conditions causing the output to discharge. To prevent damage to the internal ESD diode under the said conditions".

Could we add a schottky diode (anode connected to VOUT and cathode to VIN) in order to stay below 2V in reverse and to assure the current flow outside the esd diode ?

In datasheet, you said "add Schottky diode in series with the VIN pin" but are you sure ? (due to very high power dissipation)

Could you explain me what we have to do to avoid the LM25148 destruction ?

Thks,

Kind regards,

+1 Orlando Murray over 2 years ago in reply to pw06042023

TI__Mastermind 22302 points

Hi,

No problem on Schottky in series with VIN pin, the IC does not sink that much VIN current, only up to the VCC current which is ~20-50mA and also current limited to ~170mA.

The majority of the power dissipation is in powerstage / MOSFETs.

I think the series schottky to the IC is the best. A schottky diode from VOUT to VIN could work, but it would be difficult to estimate current capacity required.

Hope this helps,

-Orlando

0 pw06042023 over 2 years ago in reply to Orlando Murray

Prodigy 10 points

OK Many thanks Orlando for your reply. I wish you a good day.

Power management

Power management forum

LM25148: unstability