MLCC flex crack?

[zoltan]

Hello all,

After Dave's latest video I've got to do some damage assessment on work. The service guy said that it was a TVS failure after overvoltage condition on the input terminals (green 3-pin on photo), but I am sure that nothing was connected there, never. Not even in error.

So here is my conclusion about the damage. Did I missed something important, or concluded wrong?

On the following picture I marked the failing point with a yellow arrow - the inner terminal of the grey device. This is the source of thermal energy that damaged the board.
Whoever did the first diagnosis, correctly marked the point of failure, although I'm not sure the device on the picture is a TVS device (transient voltage suppressor) as shown in the schematics - I need one working module to confirm this. The packaging and termination of the device is characteristic of MLCC capacitors, never saw a TVS with this kind of termination.

The green terminal with B, Y, A markings, according to factory documentation is the BUS terminal (RS485) with only one pin connected in some way to this device (GND/Shield). The other two signals are RS485 (+) and (-) and the traces are too small (around 0.5-0.8mm wide) to carry any significant energy through them. If the energy came through them they would show major discoloration and heat damage (burned and exposed/evaporated copper). No such damage is visible in any of the pictures.

So, knowing this, let's try to analyze the cause.

Option 1)  - The device is TVS
    - The typical failure modes of the TVS devices is open, short and degraded mode. If the TVS fail short, the heating would be visible in the center of the device and both ends (and probably desolder itself from the PCB) , not on a single end. If the energy is high enough the device will explode.
    - When designing a TVS protected device, it is important to provide a fusing element to open up the circuit when the TVS device does short. Judging by the amount of damage, there is no fuse in the circuit. This design is flawed!

Option 2) - The device is a MLCC capacitor
    - The damage is consistent with the typical damage caused by the failed MLCC device.
    - The failure modes of the MLCC is open or short. In both cases the major cause of the failure is mechanical stress induced by the PCB bending, vibrations or thermal expansion of the PCB.
    - The location and orientation of the device on the board indicates that the cause of the failure is mechanical stress causing flex cracking. The device is close to a large "overhang" in the PCB with a massive connector on it. If the device (MLCC) is moved a little bit back and rotated by 90° the risk would be much lover. This shows that designer was not aware of the risks of flex cracking in MLCC. This design is flawed!

My conclusion, based on the pictures is that the energy did not come through the green connector because the thin traces on the PCB are intact. The damaged device is an MLCC capacitor - i don't see any reason why would anyone protect the inexpensive and in any case non-critical module from voltage surges coming from inside of the main device.
The large MLCC capacitor suffered flex cracking due to poor design of the PCB and failed short. This in turn caused burning on the module and probably damage on the device that powered the module.

Please keep in mind that this is only a "educated guess" based on the photo of damaged device. For anything more I need a working module to check the identity of the failed device.

I'll have a new, working module in a week or two, then I can do the measurements and reverse engineer the schematics. But until then, I would like to know if I am wrong, and if, in what part.

Z

[T3sl4co1l]

There are MOVs that look like that.  Could also be a ferrite bead, and the real short is somewhere else.

Tim

[zoltan]

There are MOVs that look like that.  Could also be a ferrite bead, and the real short is somewhere else.

Tim

I focused only on TVS, because the service guy explicitly stated that the TVS died due overvoltage (>14V).  I went trough the farnell TVS catalog images and all I found was the familiar terminal design, nothing like this one. I really don't care if it is something inside the main equipment, as the conclusion was that our guys put excess voltage by connecting something to the module and burned out the whole thing (which is not true). The point is to prove the fault is in the module.

All I know about the burned part is that it is connected parallel with the two smaller caps on the right side, and they are connected in series across the power supply (probably 5V).