74HC14 IC explosion

[trurle]

TL;DR Summary

Searching mechanism and prevention for observed anomaly with rapid unscheduled disassembly of 74HC14 IC

Recently i am trying to put in use a stock of old 74x series microchips manufactured around ~2000 AD.

Unfortunately, one of microchips 74HC14 (hex schmitt trigger inverter) in SOIC-14 package have exploded in test setup in the moment the power was applied.
Loud "popping" sound, flash of blue light, and corner of package around pin 14 (Vdd), measuring around 2mm, shattered. Exposed lead frame from pins 14 and 13 has no visible damage. Chip appears to be "open circuit" for supply current afterward though.

Did anybody experienced similar behavior before? Any ways to avoid it?

Operating conditions were "normal" - AC-DC voltage converter with 5V output (which turned out to be still ok after incident), and either weak pull-ups or ground ties at schmitt trigger inputs.

 

[Tom.G]

Most likely was inserted in the socket backwards.

2nd most likely, had been previously damaged electrically.

3rd most likely, replace 'backwards', above, with 'incorrectly.'

4th most likely, (the generic answer) manufacturing defect. (This one is especially useful when you don't want to take blame. )

Cheers,
Tom

 

[trurle]

Most likely was inserted in the socket backwards.

2nd most likely, had been previously damaged electrically.

3rd most likely, replace 'backwards', above, with 'incorrectly.'

4th most likely, (the generic answer) manufacturing defect. (This one is especially useful when you don't want to take blame. )

Cheers,
Tom

This is SOIC, not DIP. Connected by reflow soldering without socket.
I checked the orientation and soldering. All pins were connected correctly. This leave electrical (ESD?) damage or manufacturing defect as possible culprits. Never before observed ESD damage to lead to actual explosion though. Usually it results in just non-responsive input.

 

[Rive]

Singular event or it happened for more than one IC the similar way?
It may be:
- accident
- since it is from an older stock, maybe popcorn effect? Usually it is just popping, but long term storage might lead to degradation too (=> short circuit => bond wires acts as fuses). Maybe worth checking the other chips from the same stock for cracks.

 

[trurle]

Singular event or it happened for more than one IC the similar way?
It may be:
- accident
- since it is from an older stock, maybe popcorn effect? Usually it is just popping, but long term storage might lead to degradation too (=> short circuit => bond wires acts as fuses). Maybe worth checking the other chips from the same stock for cracks.

Singular event in the batch of 5 prototypes.
Regarding popcorn effect - possible indeed. The storage conditions were far from optimal, and duration was too long. Popcorn effect would be more likely at reflow rather than at functional test stage though..

 

[berkeman]

Summary:: Searching mechanism and prevention for observed anomaly with rapid unscheduled disassembly of 74HC14 IC

AC-DC voltage converter with 5V output (which turned out to be still ok after incident)

What kind of supply? Does it have current-limit capability? Maybe when bringing up old circuits (or even new ones) for the first time, turn the current-limit down and slowly ramp up the voltage to 5V while monitoring the current to look for any issues. That's usually how we first bring up new ASIC prototypes...

 

[Rive]

Popcorn effect would be more likely at reflow rather than at functional test stage though..

For the 'pop' part, yes. But it can blow only/if it's powered up. I would check the other ones for cracks and try to find a way to bake the remaining stock before use anyway.

 

[trurle]

What kind of supply? Does it have current-limit capability? Maybe when bringing up old circuits (or even new ones) for the first time, turn the current-limit down and slowly ramp up the voltage to 5V while monitoring the current to look for any issues. That's usually how we first bring up new ASIC prototypes...

Current limiter on 700mA is present, but it is not controllable (hardwired inside power brick). I indeed tested 1st prototype only with workbench power supply at low current limit (20 mA) and it has no issues.

 

[essenmein]

Another thing to consider is oxidized leads, we had built a batch of prototypes with 1 year old MCU's that were apparently not sealed properly, everything looked ok till we powered units up and they would randomly roman candle on us. Mind you the MCU didn't blow, it popped the power modules it was driving.

Turned out the problem was oxidized leads on the SMT package combined with a not so aggressive flux meant some of the leads "looked" soldered, but were barely connected and would disconnect with minimal coercion causing many very strange failures till we identified the problem was various MCU leads not connecting properly intermittently.

 

[trurle]

Another thing to consider is oxidized leads, we had built a batch of prototypes with 1 year old MCU's that were apparently not sealed properly, everything looked ok till we powered units up and they would randomly roman candle on us. Mind you the MCU didn't blow, it popped the power modules it was driving.

Turned out the problem was oxidized leads on the SMT package combined with a not so aggressive flux meant some of the leads "looked" soldered, but were barely connected and would disconnect with minimal coercion causing many very strange failures till we identified the problem was various MCU leads not connecting properly intermittently.

Yes, problem with oxidized leads (cold joints) do happen. Usually "roman candle" happens when chips are driven while power rails are disconnected. I think it can be ruled out for my particular case because all inputs of exploded chip were connected through series 1 MOhm resistors.
Anyway, will try to make more prototypes in February after parts will arrive.

 

[essenmein]

If there are such large impedance's to the I/O then it probably was just a dead chip to begin with, what's that rule, have to fail at the same thing three times to ensure you have a statistically significant result? Another thing to check though is if turning on the supply causes an initial overshoot or something that may have cooked it.

 

[trurle]

If there are such large impedance's to the I/O then it probably was just a dead chip to begin with, what's that rule, have to fail at the same thing three times to ensure you have a statistically significant result? Another thing to check though is if turning on the supply causes an initial overshoot or something that may have cooked it.

The second prototype batch of 3 (including reworked power supply brick of module which contained exploded chip) have survived qualification. Therefore, the preliminary decision is "a singular accident due dead chip".