7.4.1 Transient and Overstress
Electronic components are often prone to damage by
short-duration voltage transients, caused by switching of loads, capacitive or
inductive effects, static electricity, power supply ripple, testing, etc.
Small semiconductor components are particularly vulnerable, owing to the very
low thermal inertia of their wire bonds. MOS devices are very vulnerable to
static electricity, and require special protection.
The subject of electrostatic discharge (ESD) is treated very
thoroughly in other sources, and will only be summarized here. It is becoming
an increasingly important and recognizable problem with the trend toward the
development of integrated circuits of greater complexity and higher component
densities. Some of today’s microcircuits can be damaged by ESD voltages as low
as 20 volts. The smaller the part, the less power it can dissipate or the
lower the breakdown voltage, and the more likely it is to be damaged by an
electrostatic discharge (ESD). Certain parts are considered highly susceptible
and their chances for damage are great. These include metal oxide
semiconductor (MOS) parts with a direct access to the MOS junction, high
frequency parts produced by the Schottky barrier process, many bipolar and
field-effect microcircuits like RAMs, ROMs, and PROMs utilizing small active
area junctions, thin dielectrics, metallization crossovers, and N+ guard ring
structures, precision film resistors and similar parts. A detailed list of
electrostatic discharge sensitive (ESDS) parts and their voltage sensitivity
ranges are provided in MIL-STD-1686 and MIL-HDBK-263. They also describe
control programs that can be applied to minimize component failures due to
In addition to ESD, the designer must cope with the other causes
of transient generation described in the first paragraph.
Semiconductor device circuit malfunctions can arise from two
general sources: (1) transient circuit disturbances and (2) component burnout.
Generally, transient upsets are the controlling factors, because they can
occur at much lower energy levels.
Transients in circuits can prove troublesome in many ways.
Flip-flop and Schmitt triggers can be inadvertently triggered, counters can
change count, memory can be altered due to driving current or direct magnetic
field effect, one-shot multivibrators can pulse, the transient can be
amplified and interpreted as a control signal, switches can change state,
semiconductors can latch-up, requiring reset, etc. The effect can be caused by
transients at the input terminals, output terminals, on the supply terminals,
or on combinations of these. Transient upset effects can be generally
characterized as follows:
||Circuit threshold regions for upset are
very narrow. That is, there is a very small voltage amplitude
difference between signals which have no probability of causing upset
and signals which will certainly cause upset.|
||The dc threshold for response to a very
slow input swing is calculable from the basic circuit schematic. This
can establish an accurate bound for transients that exceed the dc
threshold for times longer than the circuit propagation delay (a
||Transient upsets are remarkably
independent of the exact waveshape, and depend largely on the peak
value of the transient and the time duration over which the transient
exceeds the dc threshold. This waveform independence allows relatively
easy experimental determination of circuit behavior with simple
waveforms (square pulse).|
||The input leads (or signal reference
leads) are generally the ones most susceptible to transient
Logic devices which interface with inductive or capacitive
loads, or which "see" test connections, require transient voltage protection.
This can be provided by a capacitor between the voltage line to be protected
and ground to absorb high frequency transients, by diode protection to prevent
voltages from rising beyond a fixed value (clamping), or by series resistances
to limit current values.
The transient voltage levels which can cause failure of
semiconductor devices are referred to as VZAP. VZAP values depend upon
transient duration. Passive devices can also be damaged by transient voltages,
but the energy levels required are much higher than for small semiconductor
devices. Therefore, passive devices do not normally need individual