Electronics > Beginners

Another one bites the dust, ESD killed IC.

(1/3) > >>

RogerThat:
Hi,

Just wanted to share my recent finding and frustration. After searching the web there are very little first hand experience of ESD killing IC so I thought I would share my story. I've been a believer that ESD so seldomly damage ICs that for hobby stuff it's no point caring too much about it. Until now, I've never had this issue. :-BROKE

I'm building a PCB which holds a 24 bit ADC. The interface to the card is a 2.54 pitch row of straight connector (same way Arduino connects to shields). During my prototyping time I've just soldered cut off resistor legs to these places so that I could easily connect the board to my protobord, these lead directly to the chips SPI interface without any protection. Very smart I thought.

Ok, so in my first design the ADC never wanted to communicate. I concluded that I probably killed it during soldering (I bought two and non of them ever worked). They were TSSOP-24 parts, first time I used them. I switch my design to thru-hole to avoid this problem, putting the ADC in a socket instead of soldering.

Two months later with the new design, the socketed ADC is dead after one week. It did survive long enough for me to do some testing of the system, maybe because is used 200ohm in line resistors on the SPI interface.

So, where does the ESD come from?
My soldering station is in the basement and my measuring /computer is indoor in the heated house. Everytime I solder I walk to the garage and then back, good way to build up static electricity. My comfy sheep skin loafers not helping much. When I get back I push the connectors (with my fingers) in to the protoboard, effectivly discharging my static electricity to the SPI interface. During no time I have feelt/heard/seen a spark working like this in my house. I've also found that my computer(Macbook pro) was connected to a non-grounded outlet and thus effectively charging me up.

So how does an ESD damage IC behave?
It becomes unpredictable and the signals doesn't switch properly. I've attached a screen shot from my scope showing SCLK(yellow) and MISO(blue) signal. As you can see the blue signal (in the beginning and the end) doesn't switch properly between 0v to 5v, but kind of 1/5 way. This video show that transistors don't die in a digital way (work or don't work) but in a gradual way:

I hope someone finds this useful.

Mr. Scram:
I've killed a few parts with sparks I could actually noticed, and have had other misbehave in similar ways without seeing sparks. Our animal brain doesn't seem to be very well equipped to handle things it can't directly relate, but that doesn't mean things aren't connected.

CJay:
Until now, you've never realised you had this issue.

ESD is a thing, it's fairly simple to avoid but so many people don't because they just don't link cause and effect and there's a mythology grown up around it saying that ESD is fiction.



.

rstofer:
If I lived somewhere where static discharge was a thing, like anywhere it gets cold, and, after heating, the relative humidity is near 0%, I would be worried.  I noticed this on a couple of trips to Minneapolis, Minnesota, where we could draw an inch long spark from a file cabinet just by walking across the carpet.  Fun for a couple of minutes, a PITA by the end of the work day.

OTOH, I live in the Central Valley of California and the humidity is never much less than 50%.  Today it is raining and the  RH is about 83%.  There is no way I can get a static discharge in this environment.

I do have an anti-static mat on my bench but I put it there when I was working on a motherboard with a CPU chip costing multiple hundreds of dollars.  For my ordinary projects, there doesn't seem to be a problem.  Things that worked a long time ago are still working.

I haven't experienced ESD damage but I have no doubt it can occur.

T3sl4co1l:
It might not even be ESD, if you're supplying power through that header it's likely that some power and signal pins made contact before the ground pins did.  Even more likely if you don't have tons of ground pins in the header.

Merely hot-plugging a connector can develop large voltages (2-3 times supply) at low impedance.  Ample current to fry ICs, especially when that current goes through signal pins rather than the intended supply/ground path!

Hot-plugging can be protected with a TVS across the supply, and series resistors between connector and IC.

If you do this, then still get occasional failures from handling, it most likely is ESD.  You would then add TVS or diode clamps to the connector pins, which absorb most of the ESD pulse, and clamp it to a modest voltage (10s volts), again at low impedance (~10A available during the spike!).  The series resistor, between TVS and IC, again limits this current, protecting the IC.

There are more approaches, depending on pin characteristics (output pins probably can't afford much series resistance), range (analog inputs might need lower leakage than a BAT54S or SMAJ5.0A has), or voltage domain (you don't want to use clamp diodes on a shared bus, on a device where its supply voltage is different, or where its supply can be shut down to save power -- clamp diodes would parasite-power it from the bus itself!).  But this is the basic outline, absorb the transient then deal with the residual.

Tim

Navigation

[0] Message Index

[#] Next page

There was an error while thanking
Thanking...
Go to full version