LM5176: The regulator don't have the required output voltage.

Hi Romuald,

Thanks for reaching out.

Can you please send your schematic?

Also it would be helpful if you can fill out and attach the excel quickstart calculator: https://www.ti.com/tool/download/SNVC208

I am looking forward for your response.

Best regards

Moritz

Hi Moritz,

Thanks for your reply.
Please find the schematics and the excel quickstart calculator.

Best regards,
Romuald

PCBE1080AAAA_BV700_VIN_24V_3A.pdf

8345.LM5176 Buck-Boost Quickstart Tool.xlsm

Hi Romuald,

thank you for sharing the files.

Is my assumption right that some boards are working without any issues and only a few are failing?

If yes, how many good and how many failing boards do you have right now?

If you say VCC is only 3V - this is the VCC pin of the LM5176 - right?

Is there anything else connected on this pin beside what is shown on the shared schematic?

Can you feed on a external voltage of 1.0V to EN/UVLO and check if on the failing board the VCC is than as expected.

Best regards,

 Stefan

Hi Stefan,

We have 15 prototypes and we have started only 5 of them.
All the 5 have the same issue.

The pin VCC at 3V is the one from LM5176.
There is no other connexion than the one on the schematic for the VCC pin.

We will made the test for the EN/UVLO and come back to you?

Regards,

Romuald

Hi Stefan,

We have set the EN/UVLO pin at 1V and this did not solve the issue.
With EN/UVLO at 3V3: 
VCC : 3V8

With EN/UVLO at 1V

VCC : 4V81.

VCC value is not the same on all board, it goes from 1V3 to 3V8.

Is there something to check on VCC?

Regards,

Romuald

Hi Romuald,

how many boards have you tested and how many of them do show this issue?

I assume there is one or more MOSFETs damaged and this will load the VCC LDO and generate a drop of this voltage to the measured level.

It can also happen that due to electrical overstress, e.g. due to undershoots on the SW the internal LDO for VCC gets damaged. 

If you have a working boards please check the SWx for over and undershoots.

Best regards,

 Stefan

Hi Stefan,

We have tested 5 boards and all have the issue.

How can we check if the MOSFET is damaged or if the internal LDO is broken?

We have also remove the diode that connect VCC and BOOTx. We have connected a 7.40V to BOOT1 and BOOT2 and the consumption is only 3mA.
On this board, only the PGOOD pin is connected to VCC now and the VCC is at 1.85V. 

Have you some other tests to do?

Regards,

Romuald

Hi Romuald,

try to remove all MOSFETs on one board and check if VCC is then OK.

Best regards,

 Stefan

Hi Stefan,

I done a test by deconnecting the VCC from BOOT1 and BOOT2.
But VCC is still at 1.85V.

In this case VCC is only connected to PGOOD pin.
Can we conclude that the MOSFET is not damaged?

Regards,

Romuald

Hi Romuald,

No, there are still the low side MOSFET.

Best regards,

Stefan

Hi Stefan,

Without the MOSFET, VCC is still low at 3.25V.
Removing the MOSFET haven't change the value of VCC.

If i have follow correctly, it seems that the LM5176 LDO have been damaged as it is still low without the MOSFET.
What can have destroy the internal LDO?

The boards have been power with a control supply at 24V and an current limited to 0.5A.

Regards,

Romuald

Hi Romuald,

I also agree that the internal VCC LDO or logic internally supplied by this LDO looks to be damaged.

The LDO or gate driver logic can be damaged by an undershoot on the SW which then leads to an behavior as you have observed.

So you should check the SW voltage.

Using gate resistors and/or a snubber on the SWx nodes can help to prevent this undershoots but unfortunately there are no footprints on the PCB (at least not in the schematic)

Best regards,

 Stefan

Hi Stefan,

You mean added a resistor between the SWx pin and the line between the source and drain of the 2 mosfet?

By experience what is the value recommanded for this application?

From which element this undershoot can be generated?

The undershoot must be large as the component is damaged after some seconds (time between starting the power supply and the measurement).

Regards,

Romuald

Hi Romuald,

gate resistor is in series with the gate line to the MOSFET and a starting value can be 2 Ohm

For snubber design see here: 

SSZTBC7 Technical article | TI.com

The Snubber should be from SWx to GND.

Best regards,

 Stefan

Hi Stefan,

I have added a 2 Ohm resistor on the gate line to the MOSFET, by the issue is still the same.
VCC is low at 3.3 V.

What can be done to solve the issue?

Another question, we are trying to simulate the LM5176 but after some time the MOSFET stop to work and output power supply did not reach the expected level.
Do you have a working simulation with the LM5176?

Regards,

Romuald.

Hi Romuald,

have you replaced the LM5176 on the board as well - in case it was damaged before.

After replacing the LM5176 and powering up again keep the EN/UVLO at 1V first, this will not start the power stage but enable the VCC LDO.
Check the VCC voltage first now.

Ensure VCC is at the right level of ~ 7V before fully enable the power stage with EN/UVLO > 1V.

Best regards,

 Stefan

Hi Stefan,

We have started a new board (faster than changing the LM5176).
We have added the gate resistor of 2.2Ohms and added a wire to control the EN/UVLO signal.

At startup with a EN/UVLO at 0V, VCC is at -0.88mV.
Increasing EN/UVLO to 1V, VCC goes to 7.31V and VOUT is at 1.36V.
Still increase EN/UVLO to 1.2V, VCC drop to 3.12V as soon as the EN/UVLO activate the LM5176.
Reducing EN/UVLO to 1V VCC is only 5.7V now.

It seems that the VCC LDO is broken when the LM5176 starts.

Do you have any idea of what can cause the issue?
Which test can we do to improve our understanding?

Regards,

Romuald

Hi Romuald,

i have had another look on the schematic you shared above and have seen some strange thinks:

- Why do you use different diodes for D3 and D2

- why are there not connection dots on the 3 marked locations

Are this connected?

Best regards,

 Stefan

Hi Stefan,

For D2 and D3, we have follow the BOM generated by Webench designer.

I confirm that the nets are connected, it is only a display issue.

Regards
Romuald

Hi Romuald,

Interesting that Webench suggests different diodes here - but should not be an issue.

Good to know that the connection points are there.

Based on your measurement results there is something which is heavily loading the VCC LDO when starting to go in active mode.

As this only happens once the device gets into active mode this should not be related to the low side driver.

So I assume that something with the high side driver is causing the issue.

(Just to ensure - can you confirm that once you go back to standby mode (UVLO ~ 1V) the VCC voltage goes back to 7.3V)

If you have gate resistors included now - try to remove them and make the same test again.

Best regards,

 Stefan

Hi Stefan,

Unfortunately, when EN/UVLO goes back to 1V, the VCC is only at 5.7V.
I have try to power off and power on the board again and the VCC is still at 5.7V.
I was not able to have VCC at 7.35V after the first start of the LM5176.

We have tried to increase the capacitor value on VCC from 1µF to 4.4µF.
But this haven't solve the issue.

With EN at 1V VCC is now low.
With EN at 2V VCC is at 6.9V but Vout is 4.5V (expected 24V)
With EN at 2.1V VCC is at 7.1V but Vout is at 4.55V
With EN at 2.5V VCC is at 4.5V and Vout at 3.38V

Regards,
Romuald

Hi Romuald,

so something happens on the boards which damage the LM5176 VCC LDO.

Most properly this is a undershoot on the SW which will then overcharge the Boot Voltage.

Try to add 7.8V  diode from BOOTx to SWx to make sure you stay below 8.5V (or above the BOOT cap).
This can keep your system in a safe state at least to check the undershoots on SWx.

Are you using gate resistors now? If yes, which value?

Best regards,

 Stefan

Hi Stephan,

You mean added a TVS diode between BOOTx and SWx in parallel with the capacitor?
Something like that:

Is the reference TVP06B9V1A-G good for this test?

We have added gate resistor of 2.2Ohms on the PCB.

Regards,
Romuald

Hi Romuald,

no, i mean a Zener Diode and the orientation is opposite to the one you have drawn.

It should prevent BOOTx from overcharging.

Best regards,

 Stefan

Hi Stefan,
I have made the test with the Zener and this haven't change anything.
VCC was at 7.42V with EN at 1V.
Once we rise EN to 1.3V VCC drops to 3.23V.
Setting back EN to 1V VCC is no 5V.

Any other Idea?

Regards

Romuald

Hi Romuald,

Please expect an answer until the end of the week.

Thanks for your patience.

Best regards

Moritz