LPC18xx has short circuit VDD to GND

[v_serpetauskas]

Hi

I have product with LPC1837JBD144. I sold 3k+ of products to the market.

We get warranty claim from our customers, and we found LPC1837 MCU has short circuit VDD to GND. We have 5 this type of accidents and we need to understand how the LPC1837 fail.

One of faulty LPC1837JBD144 had short circuit, after we applied 3.3V to vdd with higher current it started to work again.

Does anyone had similar problems?

[bernhardfink]

Distributors have a procedure in place to return parts to NXP as a "Quality return". NXP doesn't take back samples directly.

Spikes and oscillations of power supply is not always something you can really see. If you have an on-board power supply which normally works fine, there could also be a condition from the outside power source, which influences the behavior in such a way that the MCU goes into a weird state.

If the power ramp-up is not ok, when you have spikes and drops in voltage areas where the function of the silicon is not completely well defined, then circuits on the silicon could enter into weird states.

Regards,

Bernhard.

[v_serpetauskas]

Hi 

Our product once is powered up, never will stop. It has backup power supply for a long time. All external peripherals of product are powered from our product.   

According to customer pcb stopped to work after installation of product in couple weeks.

We do not have the reason to not believe the customer, we know him very well.

So i only see the possibility to go through NXP distributor to get help.  

[v_serpetauskas]

Hi

Any suggestions?

What to do next and to find how it could happen with MCU?

Vaidas

[Hui_Ma]

Hi,

Sorry for the later reply.

From XC6210B332MR datasheet, this LDO driver current typical value is 700mA.

If your board power source from this LDO? If there with other devices on the boards also power source from this LDO?

And you said " we found LPC1837 MCU has short circuit VDD to GND."

Could you let us know how you detect the LPC1837 chip  short circuit?

What tools you are using during detection?

Have a great day,
Mike

-----------------------------------------------------------------------------------------------------------------------
Note: If this post answers your question, please click the Correct Answer button. Thank you!
-----------------------------------------------------------------------------------------------------------------------

[v_serpetauskas]

Hi

LDO powerup MCU and Si4463 silabs tranceiver (its peak current is 80mA during transmit maximum).

I already mentioned we found short circuit on MCU with ohmmeter. 

Vaidas

[bernhardfink]

Hi Vaidas,

inside the LPC1837 there is no mechanism which could cause the effect you report:

1)  A board shows a very high current, you call it a short circuit

2)  You measure on this board with an Ohmmeter a short between VDD and GND at the LPC1837.
     (Did you also do a measurement on a disassembled chip?)

3)  You operate the same board with a stronger power source and then it's working again.

4)  You take a chip from a failing board and mount it on another board and on the new board it's working fine.

There is absolutely no physical explanation for this observation when looking at the LPC1837. It can well be that you have boards with a wrong power ramp / power sequence, which causes the system to go into such a high power consumption setup. But if you disassemble an LPC1837 from a failing board and measure a short between VDD and GND, then this chip is broken and you can't get iot working again by a stronger power supply or an assembly on another board.

Regards,

Bernhard.

[v_serpetauskas]

Hi

My answers:

1. Yes. at first we saw high current from LDO and LPC1837 was heating up.

2. We measured with ohmmeter short on pcb on 3.3V line. After changing the faulty LPC1837 , product was tested and sended back to client.

3. No. faulty LPC1837 was soldered to new pcb and stronger 3.3V was applied. MCU started to work. Product is tested and works fine now. This LPC1837 is on our office.

4. No. 

I understand its hard to imagine what is happening, but question is what to do next time i will receive faulty product from client?

Vaidas

[bernhardfink]

OK, then my theory fits more or less to the observations:

As I wrote in my first post, I have seen such an effect on the MCB1857 board from KEIL. It has a step down converter from 12V down to the required system voltages.

I had boards drawing more than 2 Amps and the LPC1857 was getting very very hot. Powering the board from the 5V USB still worked fine. At this point in time I didn't look for the reason, I just used these boards furtheron just with USB power.

I think I noticed the heat-up early enough to prevent the chips from destrcution.

My assumption is, that you have a problem in the local power supply. It seems that it is somewhere at the edge of something, which generates a bad start condition for the LPC1837. With the MCU change you did in 2) and 3) you seem to be on the good side again, but it could also have been on the bad side.

Overall the power supply is a very complex topic, the inductors of the PCB and also from the LQFP package in combination with the involved linear regulators (so also the ones inside the LPC1837) and capacitors need to work together. Especially the LQFP144 package with its few GND pins might be ore difficult to deal with than the BGA256.

If the power supply is the problem, then there isn't a single reason for the failure, it's more how all these components play together. On an existing PCB you can normally just add/change buffer capacitors, the existing routing and layer structure is hard to patch. It's also interesting to test this under temperature conditions, it changes the behavior of the involved components in one or the other way.

Hard to say what could be the best way to debug the issue. In principle you would need a board which shows the problem, but a board where the LPC1837 is still alive. Then you can work with a current limited power supply to prevent a destruction. Then you could apply changes to the PCB and circuit and test again.

Regards,

Bernhard.

[v_serpetauskas]

I thought at first i have issue at power supply side, but received faulty product after MCU change we investigated for power up and down sequence and other-shoots. Seems everything is ok with power sequence. 

On PCB we have ESD protection ON semi UESD3.3DT5G on every VDD lines and Reset line and USB lines.

In case of other-shoot of power supply ESD would be damaged. Now the components are ok.  

Then i will get the faulty MCU, can we send to NXP to understand where is short circuit inside MCU? 

Is this possible to do what?

We use LPC micros since 2008 and did not faced this type of problem.

Vaidas

[v_serpetauskas]

Hi 

MCU is powered from LDO U23 3.3V port.

To provide full schematic we need nondisclosure agreement.

+4.2V comes from stepdown converter. 

[Hui_Ma]

Hi，

If there with obvious/observed damage at LPC1837 chip? We are thinking if there with EMC/ESD damage to the chip.

We assumption all 3k+ boards pass to ended customer passed the manufacture testing.

So, the 5 issued boards was working before issue happened.

We have below two suggestions:

1> Replace a new LPC1837 and check if there with same issue.

      If yes, the board should with another issue affect LPC1837.

      If no, it shows the old LPC1837 exists problem, need advanced check.

2> Remove board other components besides of power related and LPC1837 (Mini configuration board) and check if LPC1837 with the same issue or not. It also need to replace a new LPC1837 chip at first.

Thank you for the attention.

Have a great day,
Mike

-----------------------------------------------------------------------------------------------------------------------
Note: If this post answers your question, please click the Correct Answer button. Thank you!
-----------------------------------------------------------------------------------------------------------------------

[v_serpetauskas]

1. LPC1837 is replaced on the pcb. Product is tested and fully working. 

      And i want to remember you -> the faulty CPU is working too fully on another product pcb.  The problem was it had short circuit inside until we applied strong 3.3V to CPU power supply. 

2. Are you asking to take another working product and remove all components? We will not be able even to try is it working or not? What will show this test?

[bernhardfink]

You always write "short circuit VDD to GND".

Is this proven with an Ohm-Meter or do you "just" see a high power consumption on the main input?

The reason why I'm asking:  I have seen one time on such a MCU board that the power regulators were in a kind of oscillating mode and sucked a lot of current. The fact that it improves when you change the characteristic of the input voltage would fit to such an issue.

You can make also a crosscheck with a change on the MCU software side: take your existing hardware and apply another software which behaves differently with regards to immediate power demand. Differently means most likely in your case that you try to demand for only little power.

There is an easy way to do that: you go into ISP mode and stay in the bootloader instead of entering your application software.

Regards,

Bernhard.

[v_serpetauskas]

Yes it is proven with ohm-meter.

If i will have faulty device i will not be able to write another firmware without VCC applied because of the vcc short to gnd. 

The only faulty MCU i have (after applied "strong 3.3V") now does not have short vcc to gnd. It works on the product perfect now. 

[Hui_Ma]

Hi,

Sorry for the later reply.

From above behavior description, the only faulty LPC1837 chip was latch-up with "Strong 3.3V".

What's the "strong 3.3V" means?

best regards,

Mike

[v_serpetauskas]

Hi

We connected power supply 3.3V with maximum 3A current.  we did not measured current.

[Hui_Ma]

Hi,

Could you post the  board LPC1837 power related circuit here for double check?

I find there with a similar question thread, please check here for detailed info.

Thank you for the attention.

Have a great day,
Mike

-----------------------------------------------------------------------------------------------------------------------
Note: If this post answers your question, please click the Correct Answer button. Thank you!
-----------------------------------------------------------------------------------------------------------------------