What is your most sad chip damage?

[technix]

Let's talk about fried, burnt or damaged chips. What is your most sad chip damage?

I recently damaged a EPM240T100C5N CPLD when trying to take it off an adapter board with my hot air gun. That was sad as the chip is not cheap, and I don't have a lot of them.

[iampoor]

 

[Tomorokoshi]

Sweet!

Potato?

[atferrari]

Not the most but first: a TL 431 melting on a protoboard in front of me. Survived for several seconds. I forgot circuit details and why.

My first of the CD4000 series CMOS (maybe a 4017).

My first micro: 16C57 with pin connected to maybe the next pin, both as outputs. Stupid bad combination.

My last micro: 40-pin 18F? with Vcc and ground pins shortcircuited after a careful wiring in the protoboard. Looked it nice? Oh yes.

[free_electron]

not me, but i saw this happening and pulled the plug...
few years ago a cow-orker had to test power consumption of the core of a new chip. we had provided a jumper on the motherboard to disconnect the rail feeding the core. i explained he needed to connect an external power supply and ampmeter. ( he needed testing over a range of voltages so we would not use the on board regulator. )

5 minutes later : smoke ... burnt socket , dead chip. Dude grabs another board and fries that one also... I walk over to see what happened ...
Dodo had connected the external supply between regulator out, and rail of the chip... instead of ground and rail of the chip. So the chip got 3.6 instead of 1.8 volts...  fortunately i stopped him in time. we only had 3 working parts at the beginning and he just fried two of em......

I w\rote a note to management that he should never be left unsupervised in the lab...

[hans]

Probably one of first times using PICKIT2 when I was still in college.

Was trying out the logic analyzer future. I wanted to debug a circuit on a breadboard powered from it. So I grabbed which I thought was CH1, used it as a trigger source, but got nowhere. It was permanently high anywhere I probed.

So I thought.. let's change trigger to logic state to test GND and Power, because it's really strange.

Poof.

Turns out I was counting short/wrong/derp with the pin-out, and had GND or PWR in my hand as "probe".
Turns out that there is no protection (it normally shuts off in short or overload) when the PICKIT2 is waiting for a trigger.

 

Eventually replaced the N/P mosfet pair, schottky diode plus 1 trace, and it's still working.
Got a CC/CV bench power supply instead.

[PA0PBZ]

About 35 years ago I saved enough money to buy my first computer:



It was a lot of money for me, saved as a newspaper boy and working on Saturdays, almost 1000 guilder or $500.
After a few months I managed to destroy the 6502 CPU, I can't remember exactly how but I still know it did hurt a lot. 

[wraper]

PIC16F84 about 14 years ago. I just notice ESD spark to the frequency counter PCB which I made an hour ago and it's dead  .  they did cost more than $10 at the time in the city where I lived. And I was just a teenager on tight budget (parents almost didn't give me any money, most of the money what I had was earned by myself).

[SeanB]

There were some 6502 bugs on early chips which could do that............ Later mask versions improved the microcode decoder so that an illegal opcode would be translated mostly to EA NOP instead of some which locked up bus drivers to kill the chip. There were some games that used these undocumented opcodes.

http://en.wikipedia.org/wiki/Illegal_opcode

http://visual6502.org/wiki/index.php?title=6502_Unsupported_Opcodes

Or you shorted an address or data bus to either ground or Vcc and blew up those rather underpowered bus drivers on the 6502. Very common as well, they were really only intended to drive a set of octal buffers external to the chip, but were often left out to save the cost of the 6 chips, 2 bus buffers, one octal transceiver and 3 TTL logic to control direction and tristate them when the bus was inactive.

[smjcuk]

<-- transistor damage here. I get through a 100x bag of GP NPNs a year.

I melted a couple of 74LS devices thinking "hey I don't have a 7805 handy and these seem to run quite fine on a 12v DC wall wart for the first 20 seconds so I'll just leave them. Hmm funny smell. Whoops."

I drove 45 miles to get my first PIC16F84 from Cirkit distribution before they went under. Drove 45 miles home and promptly blew the damn thing up by sticking in the ZIF the wrong way round.

[PA0PBZ]

Or you shorted an address or data bus to either ground or Vcc and blew up those rather underpowered bus drivers on the 6502. Very common as well, they were really only intended to drive a set of octal buffers external to the chip, but were often left out to save the cost of the 6 chips, 2 bus buffers, one octal transceiver and 3 TTL logic to control direction and tristate them when the bus was inactive.

I'm quite sure that was the case, unprotected card edge connectors on the unbuffered bus didn't really help on a messy workbench.
Oh well, on the good side that was the first processor I successfully replaced.

[Rick Law]

Not a single chip, but a single event.

I stumbled into the calibration mode of my B3603 PSU, it displayed a cryptic code which I misunderstood to be "loaded preset 2.00V", and upon the second press of OK, it proceeds to output 30V for calibration.

Almost everything hooked up to my PSU at the time suffered some damange.

Only remember the "bigger things" that was on
- I2C LCD adapter.
- 20x04 LCD display
- Adafruit ADS1115.
- I2C DS3231 clock
- TP4056 USB charger (two of them)
Arduino was ok.

The POP sound was an expensive one.

[128er]

A few years back, I had to replace a 5V PSU for a CNC controller on a big lathe. Actually my working day was over. But then, the power supply arrived at the end of the day via UPS. And I thought "come on, replace it, then it's done". Done that, turned the large mains switch on and kaaaboooomm. Huge flashes and clouds of smoke. I reversed the polarity  Several tantalums exploded. The controller was build out of four or five double euro cards in a rack. 1980's technology. The boards were jam-full of discrete logic. Trouble shooting was not so funny. Traced all the shorts, replaced the ICs and tantalums. After a few days it was back to life.

Lesson learned. Don't do something in hurry. Double check all things before applying power.

[SeanB]

<-- transistor damage here. I get through a 100x bag of GP NPNs a year.

I melted a couple of 74LS devices thinking "hey I don't have a 7805 handy and these seem to run quite fine on a 12v DC wall wart for the first 20 seconds so I'll just leave them. Hmm funny smell. Whoops."

I drove 45 miles to get my first PIC16F84 from Cirkit distribution before they went under. Drove 45 miles home and promptly blew the damn thing up by sticking in the ZIF the wrong way round.

TTL can survive that.......

crDSC01484 by SeanB_ZA, on Flickr

This board, and the rest of the computer, had the unregulated 16V rail applied to it for a while. Only failure is obvious. PM5403 hex open collector inverter, ironically one with a 30V rated open collector output. The custom Intersil IIL ADC on a separate board was absolutely happy with over double it's rated voltage.

[smjcuk]

The devices were fine and still worked but packaging melted

CERDIPs are a lot harder to bend

[Artlav]

I think it was this one.


What's so sad about it?
I ALMOST made the thing it was in work.
I knew what the problem was.
But it was the last FET i had.
And it was the middle of the night.

[wraper]

I think it was this one.

What's so sad about it?
I ALMOST made the thing it was in work.
I knew what the problem was.
But it was the last FET i had.
And it was the middle of the night.

Doesn't look like genuine International Rectifier part.

[T3sl4co1l]

I think it was this one.

What's so sad about it?
I ALMOST made the thing it was in work.
I knew what the problem was.
But it was the last FET i had.
And it was the middle of the night.

Mine's better...



Note the cratered, melty burned spot where the die arced over (IGBT die -- the other shiny rectangle is the co-pack diode, which was unscathed until the case was cracked open).

I've had much larger devices fail, but I don't have pictures handy.  ISOTOP and industrial module parts.  Those things make *shrapnel*.

Tim

[TerraHertz]

I'd found a *working* Osborne 1 computer for my so-called museum of ancient computers. I'd opened it up and had it laid out on a bench, just to check out the insides. For some reason I needed to fix something in the ceiling, so climbed up on a chair next to the table. When getting down I carelessly let my hand touch the bare main board in the Osborne. Felt the zap. Oh crap....

Sure enough it had become a non-working Osborne 1. The chip I touched was a PIA, something-21 iirc, and those things are notoriously static sensitive. But I was too busy to try replacing chips then. The machine is still in my 'fixit someday' pile.

Oh wait, a worse one... A Tektronix 7104 scope. Arrived in Australia working but badly needing cleaning inside and out. I'd had it powered up shortly before starting to open it up, and I know perfectly well about EHT remnant on CRTs. There's a HV in-line joiner on the EHT cable, so I intended to pull that open and touch the metal plug prong to the chassis. It was a bit awkward to reach. I pulled it open, then butterfingers dropped the connector. As it fell past the vertical deflection amp PCB (covered in Tek custom unobtainium ICs) there was that awful 'SNAP!' sound of HV discharge. Scope is dead. I was never able to even get the power supply working again.

[miguelvp]

Mine was a Z80 for my ZX-81 when I was on my teens. That's when I started playing with electronics and managed to get a connector that matched the pitch of the expansion port and I was sticking all kinds of TTL chips to drive relays without a board, just a could of wires and components attached to the connector.

I managed to short something and killed the processor and back then I had to wait about 3 months to get a replacement Z80 plus the time it took me to save the money to special order it. While waiting I bought a book on the electronic store where I placed the order for the Z80 and it was about assembly language for the Z80 that was on the discount bin.

That was my turning point to go the software route because I of the new found power of what assembly could do, so when I got the ZX-81 running again I started coding in assembly and by that I mean write the program on paper figure out the opcodes and program a basic program to poke the assembly code into the system. I did that for so long that I figured out what the bits on the opcodes did and continued doing assembly when I got my ZX-Spectrum later.

So you could say that it wasn't the most sad chip damage but a life changing one. But at the moment it was quite devastating because it meant I couldn't use my micro computer for about 4 months, but that extra time allowed me to really read that book back and forth.

[Armxnian]

GTX 690 die that cracked after I removed the water block. $1000 down the drain   

[smjcuk]

Ouch. That stings.

[Rerouter]

120A 48V regulator that i was helping a friend assemble boards for, i was lucky enough to complete mine before him, so off we run to hook it up to test, and between large electro's exploding, power mosfets exploding, inductors and resistors fusing and ceramic caps popping, was probably the most noise filled failure i'd ever heard,

Long story short he had every transistor pin-out rotated in some way, about 40 different transistors and IC's failed shorted, across the board all at once,

[technix]

GTX 690 die that cracked after I removed the water block. $1000 down the drain   

Whip out that hot air gun and desolder all those high current inductors, so you can roll a multi-phase SMPS capable of the same 100+ amperes if you happen to need one

Those GDDR5 chips can be recycled too for your FPGA-based projects after reballing. Tantalum caps are valuable too.

[linux-works]

with the chip carrier, its about $30.

I lovingly soldered that small chip, had it working in my audio preamp (its a really low-noise 8ch volume control chip; the best one out there that I've been able to find).  but when I did some work on the temp chassis (which has unmarked jacks on the back; being a proto and all) and reconnected the rca cables, I swapped inputs and outputs and the blue smoke came out of the chip ;(

I learned how sensitive that sucker is; and why people are putting output buffers on boards with that chip, to isolate and protect the chip if you reverse the inputs and outputs by mistake.

not allowing myself to be defeated (this is 1am in the morning) I took another chip and carrier, soldered that and made SURE to never plug the wrong cable into that box again.

$30 is not a huge amount but it still stings.

[wraper]

Whip out that hot air gun and desolder all those high current inductors, so you can roll a multi-phase SMPS capable of the same 100+ amperes if you happen to need one

Those GDDR5 chips can be recycled too for your FPGA-based projects after reballing. Tantalum caps are valuable too.

If it was my graphics card, I would order GK104-385-A2 GPU on taobao and repair the thing. Though not everyone have the needed equipment to do this. Almost hopeless to replace a chip of such size with just a hot air gun.

[technix]

Whip out that hot air gun and desolder all those high current inductors, so you can roll a multi-phase SMPS capable of the same 100+ amperes if you happen to need one

Those GDDR5 chips can be recycled too for your FPGA-based projects after reballing. Tantalum caps are valuable too.

If it was my graphics card, I would order gk104-355-a2 GPU on taobao and repair the thing. Though not everyone have the needed equipment to do this. Almost hopeless to replace a chip of such size with just a hot air gun.

Actually if you know how to work it, desolder the bad chip and you may still have what is effectively a GTX 670 left.

[smjcuk]

I'd end up with a dead graphics card and a two dead GPUs doing that

[wraper]

Actually if you know how to work it, desolder the bad chip and you may still have what is effectively a GTX 670 left.

This could work only if second GPU is dead. In this particular case it is the first one.

[Sbampato12]

Some years ago, I was trying to play with TV transmitter, but after a while looking for the IC when I finnaly got one, and just one (I was a teenager, with little money) when visiting my grandma in another city. When I came back, I've done a pcb soldered everything and let the magic smoke escape due to wrong power connection. -12V instead +12, was enough to make some capacitors pop, and fry the IC. 

I never tried to do that circuit.... 

[TheWelly888]

At work I once came across a handheld battery powered pulse oximeter that was damaged with fried ICs because someone had mistakenly ordered a boxful of 3.6V AA lithium batteries instead of the standard 1.5V AA alkaline batteries for ward stock!!!  The pulse oximeter uses 4 AA batteries

[LektroiD]

Wait until you burn out a CEM/Curtis chip. You will cry!

[linux-works]

I am reminded of the HP-25 (and similar) calcs that are RUINED by ebay sellers when they plug them in (gotton at estate sales for 'flipping' on ebay) and the battery is either dead or missing.  they put unfiltered ac (iirc) into them when there is no functional battery there; its part of the circuit (bad idea, HP!).  once ruined, the calc is useful only for spare parts, but its major chips are toast after that one mistaken plug-in.

I bought one that was like that.  another that did work, and I love that old HP calc style.

[SeanB]

I have somewhere the remains of a dual 200A power darlington inverter block, where the one side blew up when the control board failed. Replacement was in the $2000 range, and the inverter in question was thus scrapped, as it also needed a new set of batteries. A new unit with batteries cost less than the transistor block alone.

A 200a die is pretty small for the power it handles, it is only 1cm in diameter, and was soldered to a nice aluminia heat spreader, with a copper tab soldered to the metallised area on the top as the one lead.

I did see somebody who forgot to put the CPU heatsink on top of a newly built Pentium computer ( at the time that was a $500 chip as it was the first gen, complete with Fdiv bug) and powered it up. About a half second later there was a crack, and smoke poured up from the processor socket. Needed a new processor and motherboard, as the socket was also melted by the fire.

[notsob]

I once was called out to a computer site after a storm. High voltage power lines had shorted outside the building due to high winds, the computer operators heard a loud bang type noise, a great flash of light made their body shapes silhouette against the wall. A train printer had taken the brunt of all this, the metal mains filter (approx 100mm square) had a hole blown out thru its metal top. Most chips in the printer had holes blown out their centres or the tops completely blown off. Lifted, burnt tracks on boards all over the place and the interface card on the printer controller in the mainframe had chips blown and tracks lifted. The printer did get "repaired" every board was replaced.

[marshallh]

Didn't bake out parts that were acquired from Shenzhen

[m100]

Not my design or responsibility but seeing the remains of 200 circuits boards containing a total of around 2-3000 74 series logic chips all fed off a very large power supply that had no effective overvolt crowbar protection.  After pulling a few of the modules out of a backplane and powering them up on the bench the conclusion was that every chip was toast, not just any toast but very burnt toast, the kind that cracks and splits and lets all the magic out.

[Red Squirrel]

Not sure if most sad, but was a pretty stupid oopsie.  When I was super super noob with electrical/electronics I tried to power a hard drive with a 17 volt wall wart.  12 volts, 17 volts... close enough right?  I also did not know the difference between AC and DC.  The wall wart was AC.

Nothing happened, then I decided to plug it in a real computer PSU like I should have done from the start. No noise or signs of it spinning up.  Then suddenly, a few chips got red hot and caught on fire and let the magic blue smoke out.

[Vgkid]

When I was first starting out I got a broken GenRad1689.  When I powered it up, I got almost every error code(4 less than the max). Started checking the power supply late at night. Unhooked the ribbon power cable, so I could get to the rest of the board. hooked it back up before going to bed. Powered it on the next morning, only to see the display decimals flash, and nothing. So I pulled it apart, I was one off on the connection...
lets just say it is still in a box.

[ivaylo]

As a teenager in the 80s built a MAA723 (LM723 in the West) regulated, variable, 2-35V, 1A, blah, blah, blah, power supply (used Tesla KD503 as power transistor,  the 2N3055s we were getting  were considered mediocre). No need to mention how proud I was, but... there is so much room in the box. I wind another transhormer and stick another channel in. Wow, I got a two channel PSU just like those in the brochures, with 70V when in series and all... Hmm, can I get 2A when in parallel?

Nope, you get a cracked chip (literally, the chip cracked and blue smoke came out, so that's my visual when someone says "the magic smoke came out"). My heart still aches...

[TerraHertz]

At least this one wasn't my fault.
The saddest part was to realize that my most admired electronics instrument company (the old Hewlett Packard) had made this very stupid design mistake.
The circuit below is the power supply of a HP 8018A Serial Data Generator. Can you see what went wrong?
Hint: a connector pin failed. Which pin? Why the *worst* one, of course.

Design rule: connector pin failures should not completely destroy equipment.

[niflheimer]

Feedback?

[T3sl4co1l]

Eww... no default resistor on the +5 or -5.2 remote sense...

[TheWelly888]

Not me but one of my colleagues was re-assembling a volumetric infusion pump (it provides a steady rate of intravenous infusion of a solution to a patient) when he got one of the internal IDC ribbon plugs out by just one pin. He only discovered his mistake when the magic smoke was released when he re-connected the main battery!! Several hundred pounds of board replacement was needed.

[TerraHertz]

Eww... no default resistor on the +5 or -5.2 remote sense...

We have a winner!

[technix]

Eww... no default resistor on the +5 or -5.2 remote sense...

That sound like a smoke generator.

[TerraHertz]

Eww... no default resistor on the +5 or -5.2 remote sense...

That sound like a smoke generator.

Well it's current limited. So it could only blow up one or two chips at a time. They tend to fail open circuit, and so...

[niflheimer]

Eww... no default resistor on the +5 or -5.2 remote sense...

Yeah , that's what I was refering though , picked the wrong term for it

[T3sl4co1l]

Eww... no default resistor on the +5 or -5.2 remote sense...

Yeah , that's what I was refering though , picked the wrong term for it

We can split the prize, just as soon as TerraHertz shows us what it is... (I hope it's fake internet points, I love those!)

Tim

[SeanB]

12V on TTL, they sometimes survive..........

I did post my wall picture earlier, one of the last boards I serviced, and as i had plenty of stock I did not bother fixing it. Must look up that certificate I got on the last day.....