LM27403: shutdown an unable to restart during moderate transients and loads

eric garner

Part Number: LM27403
Other Parts Discussed in Thread: CSD17309Q3, CSD17312Q5

We are using an LM27403 in our 12V 3p3V supply, Imax=9A. Starting with the Webench reference design:

we implemented the converter as shown here (C18,C32 andC38 are not stuffed) :

Webench Part	Webbench Part Number	Quantity	Attribute 1 Value	Footprint	Design Refdes	Reach_PN
Css	GRM155R71E393KA88D	1	39nF	402	C35	11-0182-01
Cboot	C0603C104Z4VACTU	1	100nF	603	C17	11-0116-01
Cc1	C2012C0G1H472K060AA	1	4.7nF	805	C33	11-0184-01
Cc2	CL21C471KBANNNC	1	470pF	805	C36	11-0128-01
Cc3	CL21C621JBCNNNC	1	620pF	805	C21	11-0183-01
Cd	GRM033R71A152KA01D	1	1.5nF	201	C2	11-0185-01
Cdd	GRM219R61E106KA12	1	10uF	805	C31	11-0170-01
Cf	GRM188R61E105KA12D	1	1uF	603	C30	11-0186-01
Cin	CL32B106KBJNNWE	3	10uF	1210_270	C20	11-0173-01
Cotp	GRM155R61C104KA88D	1	100nF	402	C34	11-0116-01
Cout	GRM32ER60J476ME20L	1	47uF	1210_270	C32	11-0187-01
Cs	C0603C224Z4VACTU	1	220nF	603	C19	11-0188-01
Csby	TMK063CG101JT-F	1	100pF	201	C29	11-0181-01
Dboot	SBR0230T5-7-F	1	0.58V	SOD-523	D4	08-0104-01
L1	XAL1010-472MEB	1	4.7uH	XAL1010	L1	18-0062-01
M1	CSD17309Q3	1	30V	DQG0008A	Q3	08-0105-01
M2	CSD17312Q5	1	30V	TRANS_NexFET_Q5	Q6	08-0106-01
Q1	MMBT3904T	1	0.2A	SOT-523	Q5	08-0107-01
Rc1	CRCW04023K57FKED	1	3.57kOhm	402	R24	10-0380-00
Rc2	CRCW040254R9FKED	1	54.9Ohm	402	R20	10-0381-00
Rcs	CRCW04023K09FKED	1	3.09kOhm	402	R16	10-0382-00
Rf	CRCW04022R21FKED	1	2.21Ohm	402	R4	10-0383-00
Rfadj	RC0201FR-07105KL	1	105kOhm	201	R22	10-0384-00
Rfbb	RC0603FR-074K3L	1	4.3kOhm	603	R23	10-0385-00
Rfbt	ERJ-6ENF2002V	1	20kOhm	805	R19	10-0379-00
Rotp	CRCW040284K5FKED	1	84.5kOhm	402	R21	10-0386-00
Rpgood	RC0201FR-0756K2L	1	56.2kOhm	201	R18	10-0387-00
Rs	CRCW04028K25FKED	1	8.25kOhm	402	R3	10-0388-00
Rset	CRCW04023K09FKED	1	3.09kOhm	402	R17	10-0382-00
U1	LM27403SQ/NOPB	1		WQFN-24		02-0239-01

The issue is that if the design is started under moderate load, a load transient of more that ~2A is applied or a total load of more that 5A is applied the converter shuts off and cannot restart until the load is removed.

recompensating the design to give more phase/gain margin:

improves the stability slightly but not much. do you have suggestions on troubleshooting what's going on here?

over 5 years ago

0 Mathew Jacob1 over 5 years ago

TI__Expert 5900 points

Eric,

I see that the current limit setting resistor is off and you have low phase margin. Please modify to attached. Other recommendations are increasing input caps by adding one more 22uF cap and maybe a 0.1uF right next to the HS FET. The output cap can also be increased if you need better transient undershoot and overshoot or you can increase the switching frequency to 500kHz and increase the bandwidth. But first try these values and the problem should be resolved.

0 eric garner over 5 years ago in reply to Mathew Jacob1

Prodigy 160 points

I'll swap values and see what happens

0 eric garner over 5 years ago in reply to eric garner

Prodigy 160 points

changing to those values made a big difference, it still drops out on 8A load transients for 4 ms despite 188 uF of output capacitance. It might do better with a stiffer bench supply feeding it.

do you know how the webench values could be so off? usually they are pretty spot on

0 eric garner over 5 years ago in reply to Mathew Jacob1

Prodigy 160 points

after re-implementing with your recommended values, startup under load (0.5 ohm) is still not stable. using either a 0.5 ohm 100W resistor or a Rigol DL3021 load. do you have other recommendations?

0 Mathew Jacob1 over 5 years ago in reply to eric garner

TI__Expert 5900 points

Eric,

Are you powering up with the load connected? Can you try to disconnect the load and then connect it after power up.

You may want to try increasing the soft start time.

Regards,

Mathew

0 eric garner over 5 years ago in reply to Mathew Jacob1

Prodigy 160 points

doubling the SS time (changing Css from 47 nF to 94nF) does not resolve the issue

0 eric garner over 5 years ago in reply to eric garner

Prodigy 160 points

powering up with no load, no issues

0 Mathew Jacob1 over 5 years ago in reply to eric garner

TI__Expert 5900 points

Okay once you power up with no load, increase load current and till what current can you load before shutdown happens. Not a transient load step but increase load linearly. Please take SW pin waveform at 5A with no bandwidth limit on scope. I am interested to see waveform and particularly if there is any overshoot exceeding ratings. Also how far is the VDD cap away from the IC VDD pin? It needs to be right next to it.
OTP pin can be connected to ground to rule out any noise issues.
Regards,
Mathew

0 eric garner over 5 years ago in reply to Mathew Jacob1

Prodigy 160 points

increasing load in 0.1A steps, it drops out at 10.2A

SW pin waveform:

as measured here:

Cdd location:

It's not close. I'll move it closer to the pin and hang it off that via

0 Mathew Jacob1 over 5 years ago in reply to eric garner

TI__Expert 5900 points

Yes the SW node exceeds abs max so need to control overshoot. You can try a 3.3ohms in series with the Cboot.
The SW node peak maybe higher at load currents closer to 10A. Looks like there is no place holder for a RC snubber from SW pin to ground.
Regards,
Mathew

0 Mathew Jacob1 over 5 years ago in reply to eric garner

TI__Expert 5900 points

0 eric garner over 5 years ago in reply to Mathew Jacob1

Prodigy 160 points

no 3.3 ohm 603s in house, so used 2 ohm in series with Cboot:

it definitely damped the amplitude of the ringing

at 10 ohms its way better:

0 Mathew Jacob1 over 5 years ago in reply to eric garner

TI__Expert 5900 points

Eric,

The 10ohms looks good. If you have a 6.8ohms maybe okay but that can be done later. It would be nice to damp the ringing with a snubber of 1nF and 2.2ohms from SW pin to ground.

At this point we are just trying to reduce the noise on the board. Did you also move the VDD cap closer?

If the OCP trip point is not getting better with gradual increase in load then that needs to be focused on first to get it to trip as per the calculations.

Regards,

Mathew

0 eric garner over 5 years ago in reply to Mathew Jacob1

Prodigy 160 points

VDD cap was moved closer with no reduced sensitivity to load transients.

added snubber and ringing was improved yet again:

response at 6A transients is still poor, but this is probably good enough for this board spin since this needs to be out the door soon

0 Mathew Jacob1 over 5 years ago in reply to eric garner

TI__Expert 5900 points

Eric,

Did you try increasing Csby from 100pF to higher values? I would recommend taking waveforms at the CS+ and CS- pin with a differential probe to measure the levels.

Regards,

Mathew

0 eric garner over 5 years ago in reply to Mathew Jacob1

Prodigy 160 points

Matthew,

just received the 2nd board spin yesterday. load transients are improved but ringing is much more pronounced even after re-tuning the RC on SW.

at 9A load:

0 Mathew Jacob1 over 5 years ago in reply to eric garner

TI__Expert 5900 points

Eric,

I think this looks okay. If you can send the updated schematic can advise. It is quite normal to have a bit of ringing and overshoot. More you damp it the lower the efficiency. Key is to make sure it does not exceed the SW pin rating and the FET ratings.

Regards,
Mathew

0 eric garner over 5 years ago in reply to Mathew Jacob1

Prodigy 160 points

0 Mathew Jacob1 over 5 years ago in reply to eric garner

TI__Expert 5900 points

Eric,

Your values look okay. R26 value of 1ohms can also be tried. See that you don't have a high side fet gate resistor but its okay since you have a boot cap resistor. Basically the design is for a 14A current limit set point. If you use 2.2uH inductor you will get higher efficiency due to lower DCR and the transient response also would be better. The compensation values can stay the same for trial purposes. R16 and R17 can be cut in half in case you switch to 2.2uH.

Regards,
Mathew

0 eric garner over 5 years ago in reply to Mathew Jacob1

Prodigy 160 points

i'll try 2.2 uH on monday and see how it goes.

Power management

Power management forum

LM27403: shutdown an unable to restart during moderate transients and loads