The small logic and driver parts on a board often look like leftovers after the processor has been chosen. They are not leftovers. They decide what happens during reset, how a slow input becomes a clean edge, how a group of loads is driven, how a display is scanned and how old voltage conventions meet a new controller. A design can use a large MCU and still need one tiny gate because a signal must be valid before firmware runs. It can use a modern PMIC and still need a Darlington array because a relay coil or indicator load is not a CMOS input.

P3.23 gathers those supporting ICs. The list starts with logic gates that clean, combine, invert and count signals. It then moves through an analog switch that routes small signals without asking a relay to do the job. The second half is output drive: seven and eight channel Darlington arrays, a dual peripheral driver and an LED matrix driver. These parts sit between neat digital decisions and untidy real loads. Choosing them well keeps the main controller smaller, the firmware calmer and the reset behavior visible on the schematic.

Logic gates are small decisions made in hardware

A logic gate is easy to dismiss when the MCU has spare pins and the firmware team can write another branch. That view breaks down at power up, during reset, during brownout and at any point where the signal must be valid without software. 74HC132D as a quad Schmitt NAND gate is a good example because the Schmitt input gives the gate a defined switching band. That matters for slow edges from RC delays, mechanical contacts, optocouplers, sensor thresholds and power good nets. A plain CMOS input may chatter when the edge lingers in the transition region. A Schmitt gate turns the same slow movement into a cleaner digital event.

The NAND function also gives the designer a compact way to combine enable conditions. A rail may turn on only when a main power good signal is present and a lid switch is closed. A radio may stay disabled until reset has released and a host strap is read. If those conditions are implemented only in firmware, the first milliseconds of power up become harder to reason about. A hardware gate makes the default state visible. The question is not whether firmware can do the same thing later. The question is what the load sees before firmware is allowed to run.

Single gate logic serves a different purpose. 74LVC1G08QSE-7 as a single AND gate can live close to the two nets it combines, instead of forcing long traces back through the controller. SN74LVC1G04DBVR as a single inverter can correct polarity at the point where the signal crosses a functional boundary. Those are small changes, but they simplify the board. A single inverter beside a module connector can be clearer than a note that says the firmware must treat that pin as active low forever.

Counters and state logic reduce timing pressure

Not every sequence needs a timer interrupt. SN74HC161PWR as a synchronous counter is the kind of part that still earns a place when a circuit needs repeatable edge counting, scan stepping, divide down behavior or hardware sequencing. A synchronous counter changes outputs in a controlled relationship to the clock. That can matter when several enables, address lines or LED rows are derived from the same step. It also keeps a simple periodic job out of firmware that may already be handling radio timing, sensor conversion and low power state changes.

The part is not a cure for poor firmware architecture. It is a way to make a small repeated behavior independent of firmware latency. A gateway might use a counter to step through status outputs. A test fixture might need an address sequence that remains stable even while the controller is being reflashed. A sensor board may need a low cost divide or sample marker. In those cases the counter gives the hardware a small amount of memory and sequence without pulling in another MCU.

An analog switch is not the same as a digital mux

HEF4066BT as a quad bilateral analog switch belongs in a different mental bucket from a logic multiplexer. It passes analog or bidirectional signals through an on resistance that depends on the supply and signal level. That makes it useful for routing low frequency analog paths, selecting calibration resistors, muting audio, switching simple sensor outputs, or connecting a test node during production. It also means the designer must check leakage, on resistance, signal range, charge injection and what happens when either side is unpowered.

The switch can save board area and avoid the mechanical life of a relay, but it should not be used as a magic wire. If the signal has high impedance, leakage may matter. If the signal has amplitude close to the rail, headroom may matter. If the circuit is sampled by an ADC, settling time after the switch changes matters. If the switch connects to a connector, protection and fault current still matter. The part is helpful when its limits are part of the signal budget instead of an afterthought.

Driver arrays keep loads away from fragile pins

CMOS output pins are not load drivers. They can toggle a logic input and light a small LED under controlled current, but relays, solenoids, buzzers, lamps and long indicator harnesses need a stronger device and a clearer fault path. ULN2003D1013TR as a seven channel Darlington driver array and ULN2803ADWR as an eight channel Darlington driver array are common because they pack several low side drivers into one package and include clamp paths for inductive loads. They let a small controller command a bank of outputs without pretending its pins can absorb coil energy.

The Darlington structure has a voltage drop. That drop matters when the load voltage is low or the current is high. It also dissipates heat inside the package, so channel count alone is not the current budget. A relay bank that looks fine one coil at a time may become warm when several channels are on together. The review should include load current, duty cycle, clamp energy, package temperature, ground return path and what the load does when the MCU is reset. Pulling the input low by default can prevent a relay from clicking during boot.

SN75472P as a dual peripheral driver fits cases where a smaller number of outputs needs a dedicated driver rather than a full array. The same thinking applies. The load current, voltage drop, clamp path and input threshold must fit the circuit around it. A driver part is often where a low cost board becomes serviceable because the expensive controller is no longer the device that takes the hit from the output wiring.

Display drivers turn many LEDs into one managed load

LEDs look harmless one at a time. A matrix is a different load. It needs row and column timing, current control, brightness matching, refresh rate, fault behavior and EMI awareness. LP5860TRKPR as an LED matrix driver is the kind of IC chosen when the display should not consume an unreasonable number of MCU pins or timer slots. The driver can handle scanning and current control while the main controller sends display data at a higher level.

That does not remove board level choices. LED current still comes from a rail with limits. The matrix still creates switching current. The package still needs thermal margin. The interface still needs a clean reset state so the display does not flash random patterns during boot. A driver also changes the service story. If the display fails, the designer can separate a data issue, a supply issue, a damaged LED string and a driver fault more cleanly than with a pile of GPIO pins.

These parts should be reviewed by state, not by package

A useful review of logic and driver parts starts with states. What is the output during power off, ramp up, reset, firmware update, brownout, sleep and fault recovery. Which loads must stay off until the controller has declared the product ready. Which input edges are slow enough to need hysteresis. Which signals cross a connector or cable. Which outputs can create heat or inductive kick. After those questions, the choice between a Schmitt NAND, a single gate, a counter, an analog switch, a Darlington array and a matrix driver becomes clearer.

The same review should include manufacturing and service. A gate that combines two enables should be probeable, or at least explainable from two nearby nodes. A counter used for scan stepping should have a reset path that the production fixture can force. An analog switch used for calibration should have a known off state when the tester disconnects. A driver array should leave enough measurement access to tell whether the load, the driver input or the driver output is at fault. These are not large additions to the schematic. They are the difference between a board that can be diagnosed and a board that only says it failed.

Thermal behavior is another place where small packages hide real work. A Darlington array may be chosen because the load current per channel looks modest, but the package sees the sum of voltage drop, duty cycle and ambient temperature. A LED matrix driver may have current control, yet the board still has to move heat away from the package and keep the display rail quiet enough for nearby sensing or radio circuits. A single gate may draw little current, yet its input can sit at an undefined level if a connector is unplugged and no default pull state was planned. Small parts do not remove engineering review; they focus it into places that are easy to miss.

There is also a procurement angle, but it should not drive the design before the electrical role is clear. Logic families differ in supply range, input thresholds, output drive, tolerance of slow edges and power off behavior. Driver arrays differ in clamp structure, voltage drop and channel current. LED drivers differ in scan method and fault reporting. A pin compatible option may fit the footprint and still change reset behavior or output polarity. For that reason, the order is first function, then electrical limits, then package, then sourcing options. Reversing that order turns a small support IC into a future change request.

These parts rarely decide the product headline, yet they decide how the product behaves during all the moments users do not notice unless something goes wrong. A clean reset gate prevents a false enable. A small inverter removes a polarity trap. A counter keeps a scan sequence independent of firmware load. An analog switch routes a test point without a relay. A driver array absorbs a coil event that should never reach the MCU. A LED matrix driver keeps display current and timing out of the main control loop. That is why filling out a circuit with the right small logic and driver parts is not tidying. It is system design in the parts of the schematic that are easiest to underestimate.