Causes of CMOS gate failure?

[Top Top]

I've had two devices I've sent out come back with bad/failed input gates on CD4070 chips.

In both cases one of the inputs on one XOR gate was reading about .85v under the supply voltage (+12v supply) even when no signal was present at the input (causing a constant "on" at the gate output). The other input of the XOR pair was functional in both cases. There is no direct short from the misbehaving pin to supply.

It was a different gate on the IC in the two cases (though same IC position on the board), and there's nothing obvious wrong with the design (inputs have pulldown resistors and simply take input signals from other CMOS gate outputs).

Any thoughts?

[Grumble]

It could be electrostatic discharge while handling before mounting the ic's on the pcb.
An Analog Devices engineer once said during a talk that the best thing that could happen to an opamp when exposed to electrostatic discharge is to get defective immediately, because the electrostatic discharge can (and will) damage the opamp but the effect could be immanent but not immediately noticeable!
And what goes for opamps also goes for cmos gates...

[Top Top]

[Grumble]

What also is possible in case of an eurorack system that the electronics draws a higher current of the +12 volt line, so the -12 volt line will keep its voltage longer as the +12 volt line which could introduce a negative voltage to the inputs when driven by an opamp that uses both voltages.
This will especially occur when the main switch just switches off the mains supply to the psu, so it is always better to have a switch between the psu and the rest of the system.

[JovianPyx]

Overvoltage, reverse polarity, excessive heat, cracked or broken surrounding package and static discharges.

When I worked for DEC we watched a movie made by DEC engineers to demonstrate static discharge.
Two engineers stood a few yards apart, each has a transistor tester box.
Neither engineer wore any static discharge protection gear.
One had a CMOS transistor.
He tests the transistor which shows good.
He lightly tosses the transistor to the other engineer who catches it and tests it.
The transistor was tossed back and forth about 4 times and it would test good until the next time it was tossed at which time it tested bad and did not recover. There was no carpet shuffling or other overt static electricity generation. The failure occurred due to static that might have come from any number of things, such as a polyester shirt or even simply waving the hands in the air and the fact the transistor traveled through the air when flying. This was a very dramatic demonstration and I never forgot to wear a static strap afterward. ESD diodes are really just parasitic diodes that form as a part of the manufacturing process and can protect from only a very low level of static discharge. I highly recommend that any time CMOS parts are handled that the handler use a grounded static protection strap. After the movie, a guest environment engineer showed us a device that can meaure static charge. We pointed it at the carpet between the cubes and it showed 300,000 volts of charge. Static electricity is everywhere and since we can't see it, it is really important to always wear your static protection gear.

Also, static damage isn't always full fail. Static might punch a hole through the glass gate insulator and form a high value resistor. Stuff like that can make the device work sometimes, or work weirdly.

[Top Top]

Wow, I had heard of their static sensitivity before but have never experienced failure of anything before this, so I never took it too seriously in terms of caution measures. I suppose making things in volume you are bound to come across failures more quickly.

I find it odd that it was a 4070 that failed in each case, and that it was the same 4070 in both cases (there are two of them in the design, out of six ICs total). Until I can find any other reason, I am going to guess it is something with static, and I will make sure to take better grounding precautions.

[CHRISKELLY]

On the topic of static discharge straps - is it advisable to have a large resistor on it within the discharge path to ground? Say 1Mohms?

Im sure I heard that said before as a safety measure.

[Grumble]

[Grumble]

[CHRISKELLY]

Wow - reading that article Grumble makes me wonder how any of my circuits still work!

[Steveg]

Modern CMOS has built in protection diodes which makes it a lot more robust. Resistors don't do a great to shunt static discharge but they can block it. Use diodes to shunt high voltages to V+ and neutralise low voltages by drawing from ground.

[Grumble]

I agree, the article I point to in one of my previous posts is from 2001 and a lot have changed since, but the majority of cmos components used are produced in the previous century even