74hc595 failure?

Ok, I must have bad luck if you guys remember I had a uln2803 fail, probably my error, I have a circuit with two 74hc595 shift registers driving two uln2803 anyway and this time a 74hc595 failed with the last 4 outputs stuck on, even with the output enable off,
the first one is fine and half of this one but idk why just half doesn't work
any ideas? I had tested to board for continuity between connections and made sure there were no shorts before I applied power, and actually it worked fine two days ago
I had the solder points underneath taped so nothing could touch metal, but maybe static electricity could do this?

I replaced it successfully, which was a pain, but still im curious and kinda afraid why that one broke, I only had one spare and I thought I took all precautions the first time, but apparently something happened that caused it to fail, which an update on that I put a multimeter on the output of the ones stuck on and it was at 4.5v and when it was supposed to be low it went to 3.2v about, kinda odd I would think, especially the output enable part not working to stop it,

Somehow you blew it up I guess. Chit happens but if you do it again I'd start looking seriously at the circuit.


Rob

Static electricity?

Lack of decoupling capacitors?

Long wires picking up interference?

Accidental shorts?

Poorly regulated supply?

Its just two shift registers and two darlington arrays, not much 5v power being drawn but I thought maybe it was my laptops less than 5v usbvcc so I put it on a 5v 3a supply,
I have a 10uf cap on there right in between all the chips,
The design is so simple that's why its on a protoboard instead of a pcb, just 4 ics, no other components
I covered all the uln2803 outputs to make sure it wouldn't short, covered the back
the shift register outputs go 1/4 inch to the uln2803 inputs, all covered so I can't imagine a short there
I am working on a plastic table tho, not that I've ever actually felt a spark ever, but could that do it?

Check the power supply voltage (especially as it switches on) ??

5v, solid
I used to have it on my usb port which is actually 4.2v unloaded, so I thought that was it, but with a real supply it didn't help
Could undervoltage harm it?

Do you wire it up when the power is on?
That would cause you to damage chips.

Its all soldered connections and the only movable connections are the ones to the arduino and those are inputs really, but also done before power is applied
its odd to me atleast that the daisy chained one went bad while the first one, which shares nearly all the same inputs would be apparently fine
How easy is it for static to damage it? I don't hear of it oftern so idk

The HC family are CMOS logic based... known for sensitivity to static... and why most parts are now shipped in anti-static containers.

How easy is it for static to damage it? I don't hear of it often so idk

The correct answer is that you should ALWAYS wear a ground strap when handling semiconductors. I usually "cheat" by touching one of the metal-cased, grounded, pieces of equipment on my workbench just before touching the electronics.

Static damage is rather rare. Most chips have some sort of static-protection, but there is limit the amount of voltage/current you can hit it with. It depends on the parts and it depends on the environment (humidity, etc.) and conditions where you are working. If every time you walk across the room and touch metal, you get a spark, you are at high-risk. But, I'm pretty sure you can damage a chip when the discharge is too small to feel or see.

It's also rather rare to run-across bad chips or unreliable chips. Once your circuit works, it should work "forever" unless you stress it with over-voltage, over-current, or over-heat. If you look-up the failure rate (MTBF), it's a LONG time... The MTBF for the Atmel chip is estimated to be almost 2000 years at normal operating temperatures! Of course, that's just an estimate since it has not existed that long, and it won't be in-use that long...

I assume that you purchased your parts from a reputable supplier, and NOT from some random seller in eBay?

I had a uln2803 fail, probably my error, I have a circuit with two 74hc595 shift registers driving two uln2803 anyway and this time a 74hc595 failed

I hate to attempt failure analysis... I always hate to guess when one of our customers asks what caused a failure...

But if the 595s that failed were both connected to the 2803 that failed, here's one possibility - You did something to overload the 2803 (something that caused too much current to flow). Then, the damaged/shorted 2803 "pulled" too much current from the 595, damaging/weakening it.

Could undervoltage harm it?

No. Over voltage (including static discharge), excess current, or excess heat. But, I have seen a power supply that would put-out a brief ~20V spike when first turned-on, before settling-down to 5V. (That's VERY RARE.)

No if you look a the internals for a 2803 there is a largish series base resistor (2K7 ohms) however IF the software had an error that allowed a whole string of LED's to come on for an extended period of time or many gates on at the same time a failure could happen where the 2.5A substrate Max current could be drawn/exceeded, this would create hot spots which are the beginnings of second breakdown. An indicator of a chip 'broken/damaged' in this manner would be voltage on an open gate input at the ULN2803. go read the data sheet carefully and examine your S'ware carefully as it is easy to forget the total device dissipation. One good pointer is to measure the component temperature after 5 or 10 mins of operation. The chip shouldn't be hotter than perhaps 40 - 60C if it is then that type of failure is a reality. If the emitter junction of the output (driver) transistor failed due to heating it would likely show a positive voltage on an open input. Devices as a general rule fail only when abused and this is your second posted notice of failure. One thing that might have been missed is do you or have you disconnected power immediately before a failure as you might have a ground loop that is biasing the devices into a failure or you have an intermittent PSU regulation issue. also remember that the device ratings MUST include the operating temperature and MUST be DE-RATED for higher operating temperatures. or they will die...

Doc

Yeah this particular project has had bad luck, but the failed 74hc595 wasn't connected to the failed uln2803 from another post, this was a clean slate, entirely new protoboard
im using 15/16 of the two uln2803s, but drawing about 44ma each output so max draw on a chip would be 1/3 amp
also all the leds are only on alltogether for a few seconds at first, then maybe half may be on at once, but cycled around a bit
one thing I just thought of speaking of just before power removal, could I damage it if I remove power in the wrong order?
I have a 12v lead acid battery as the led vcc, a 12v to 5v 3a power supply for the arduino and chips vcc, so it all has acommon ground, but if I were to remove the power in the wrong order could something bad happen?

heres a pic of the situation, thats before the failure and with only 11 leds hooked up, but all working as expected
i hooked up the last 4 then upon powerup realized the second chip wasnt operating properly

heres the project, with everything on except turn signals(shoddy video editing software on my ohone couldnt get that frame of the startup test everything phase)

and if anyones curious, thats the gauge cluster from my 1st car, sadly that was totaled in an acciden
a beautiful 1993 mitsubishi 3000gt sl, this is gonna be part of its "shrine" hanging on my wall next to the smashed license plate and other memorabelia

The base resistor of the 2803 would have limited current to the HC595 to approximately 0.00259259259259259259259259259259A or 2.6 mA, not nearly enough to damage the part. there is one failure I haven't discussed yet, the reason why lightening damaged electrical devices should be discarded even though they appear to work properly... Static damage isn't always immediately apparent, it can manifest tiself days and weeks later... possibly disconnected failures and possible why the device intermittently fails and might be killing the driver too...

Doc

Eek im starting to wonder if maybe I should get one of those ground strap things, especially after something this annoying