4
\$\begingroup\$

The "generic error" page features this damaged chip with some kind of cavity:

enter image description here

I guess it's some typical damage, but I have no idea of what it might be.

What's the damage incurred by this chip and how is such damage typically incurred?

\$\endgroup\$

4 Answers 4

9
\$\begingroup\$

Thermal damage. Part of chip, mostly metal, overheated, melted and evaporated, after some pressure build up it exploded. Cause is shorting. Cause of shorting can be anything, from manufacturing defect, to ESD, extraterrestrial particle, overvoltage, mechanical stress, bad design, or even software.

The chip itself is not visible on picture. What fell off is only the mold/casing material.

\$\endgroup\$
6
\$\begingroup\$

Other people have identified what causes this failure mode, I'm going to comment on how it likely happened.

The chip in the picture is a TPA3100D2. This is a high-power, Class D audio amplifier.

Determined using the excellent alldatasheet.com search engine. See: here.

The explosion happened near the corner proximate to the pins 32-42. Pins 32-33, 34-35, 36-37, 39-40, as well as 41-42 are strapped together, and are the power supply and output pins for one channel of the device (this also explains the "solder bridges", as @Thorn described them. They are the same pin, internally).

Considering the amplifier datasheet specifies it's protected against short-circuit, I would imagine that there are two possible things that caused this failure. The power supply voltage range was exceeded or reverse biased, or the amplifier was hooked up to a highly inductive or capacitive load.

\$\endgroup\$
2
  • \$\begingroup\$ "or the amplifier was hooked up to a highly inductive or capacitive load." Wouldn't a typical speaker be classified as an inductive load? \$\endgroup\$
    – jippie
    May 28, 2012 at 11:23
  • \$\begingroup\$ @jippie - Yes, but only to a certain extent. Normal speakers are inductive, but not that inductive. I'm thinking "solid-state tesla-coil inductive", not "speaker-inductive". \$\endgroup\$ May 28, 2012 at 12:28
2
\$\begingroup\$

IC manufacturers call this sort of damage "EOS/ESD" damage - electrical overstress / electrostatic discharge.

The real mechanism for this sort of damage is thermal - something under the packaging got really hot really fast, expanded and blew up. This can be due to many factors, but usually means that the silicon under the package suffered some damage (electrical or thermal), became a non-fusible heat source and went kapow.

It's very difficult to pinpoint the real cause of this sort of damage without knowing the operating conditions and time-to-failure. The reasons for failures can be nearly infinite, and proper failure analysis requires getting as much information about the conditions to pare down the fault tree to a minimal number of possible causes.

Assuming this part is part of a qualified, released assembly:

If the part blew immediately on first power-on, it's likely due to an electrical fault at the IC or elsewhere in the circuit due to a manufacturing defect (missing part/solder bridge) or operator error (reverse polarity applied, etc.)

If the part blew after a few hours, it could be due to process-related latent damage (thermal shock due to reflow, ESD during handling, mechanical damage from pick-and-place, board flex, etc.)

If the part blew in the field, it could represent a design marginality that doesn't easily manifest, or latent damage, or customer abuse, or bad yield from the IC manufacturer...

If this is a first-run or prototype build, not only do you have to rule out manufacturing defects but also design oversights (excessive voltage or load, insufficient cooling, etc.)

\$\endgroup\$
4
  • 1
    \$\begingroup\$ It's almost impossible that ESD can do this. A 200pF capacitance at 10kV holds only 10mJ of energy. That's 100mW for 1/10th of a second. \$\endgroup\$
    – stevenvh
    May 28, 2012 at 13:10
  • \$\begingroup\$ @stevenvh It's what the manufacturers claim, i.e. what they always say when they see this sort of stuff. That being said, I've seen my fair share of random production blow-ups where ESD couldn't be proven, but the yield improved once a broken wrist or heel strap was replaced. (The ESD doesn't cause the crater, but it damages the die so that when power is applied, kapow) \$\endgroup\$ May 28, 2012 at 16:56
  • \$\begingroup\$ That's what I meant: it doesn't cause the crater. \$\endgroup\$
    – stevenvh
    May 28, 2012 at 16:58
  • \$\begingroup\$ We agree to agree then :) \$\endgroup\$ May 28, 2012 at 16:59
0
\$\begingroup\$

Problem started due to overheating. Because the system was running on very low or zero sound for long time. Amplifier heats due to this reason.

\$\endgroup\$
1
  • \$\begingroup\$ Please don't use SMS abbreviations like "bcoz" on EE.SE, and please do use proper English capitalization. \$\endgroup\$ Nov 23, 2014 at 17:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.