0
\$\begingroup\$

A question came up about common failure mode of a transistor. Searching past posts I actually found varying answers, either open or short. Maybe it depends on the application, and I'm missing something?

"The output of a TTL NAND gate is being used to drive a relay ON by pulling the source low. The diode for the relay was installed backwards, effectively shorting the gate's output to the power supply. Which component has most likely failed?

  • The gate's output transistor has failed open
  • The gate's output transistor has failed shorted"

schematic

simulate this circuit – Schematic created using CircuitLab

The book provides that 'open' is the correct answer, not 'short', but doesn't explain why.

\$\endgroup\$
5
  • 5
    \$\begingroup\$ In my experience almost everything fails silently short first, and then quickly with smoke or bangs gets opened. \$\endgroup\$
    – PlasmaHH
    Commented Mar 16, 2017 at 22:21
  • 1
    \$\begingroup\$ In my ESD lab I've seen them fail in all crazy ways... \$\endgroup\$
    – Arsenal
    Commented Mar 16, 2017 at 22:34
  • 3
    \$\begingroup\$ As @PlasmaHH said, if it shorts it will cook really fast, eventually it will either burn open or the bonding wire will fuse. \$\endgroup\$
    – Trevor_G
    Commented Mar 16, 2017 at 22:38
  • \$\begingroup\$ Without proper diode clamp, the excessive flyback voltage and current would short out Vce. \$\endgroup\$
    – D.A.S.
    Commented Mar 16, 2017 at 23:05
  • \$\begingroup\$ It would be good to look at the involved gate design, e.g. zh.wikipedia.org/wiki/File:7400_Circuit.svg. 1N4148 can tolerate 2A peak current for 1us, I did not see consumer chips having similar specifications. The circuit must eventually fail open per physics - burning things, even metals, become oxides which most probably do not conduct current. \$\endgroup\$
    – Anonymous
    Commented Mar 16, 2017 at 23:24

4 Answers 4

0
\$\begingroup\$

In my experience, your specific example will have two distinct failures.

I'm assuming that the power supply can supply several Amps of current when I describe the following:

First: the transistor fails SHORT

Second: the bonding wire that runs from the die to the package pin burns open.

Depending upon the energy levels involved, the second part of that failure could be silent, smelly, noisy, or both smelly and noisy.

\$\endgroup\$
0
\$\begingroup\$

If the transistor failed short, it (and the diode) would form a direct path from power to ground, and would experience a large current flow. Between the two of them, one would eventually fail, and the transistor is probably the more likely.

\$\endgroup\$
0
\$\begingroup\$

Failures can be analyzed, and a lot of different mechanisms exist. The particular properties of this circuit (the diode carries excess current, as does the NAND, if the gate goes LOW) don't really determine if the diode would fail open or short, or if the NAND would fail open or short, but because currents are unlimited (except by the power supply impedance) a fault IS expected.

A likely scenario is an open gate; the aluminum metallization atop the gate chip would melt before the diode (assuming it is a plug-type construction) would overheat, or the dopants in the gate would rediffuse. If the diode were MISSING, a current surge from the relay might dump enough heat into the chip to make it fail short (and the melted aluminum might pool to connect the pin to that short).

That's only 'likely', though: the surge of current could also damage the power source, or trip a protection circuit that hasn't been diagrammed. Or, the diode (also stressed beyond its specifications) could prove less robust than expected.

\$\endgroup\$
0
\$\begingroup\$

I would think the NAND gate failed since it was sinking 50 mA which exceeds the the max IOL for any TTL logic gate.

Logic output transistors are not the same as a bus driver like an 74xS244

I would expect the output transistor to fail open as the transistor likely burned.

When the current would flow through the reversed diode the TTL gate would fail. The diode can take 300mA where the logic gate is 10mA or less.

There is no way a logic gate failure could take out a power supply. Not likely to take out a 300 mA diode either.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.