The buck configuration above steps 24V down to 5V at 500mA. R_SC = 0.3Ω sets I_PK at 1A. L = 220µH ferrite-core with I_SAT ≥ 1.2A. C_T = 470pF gives ~50kHz switching.

The 1N5819 Schottky has 40V/1A rating with V_F ~0.4V at 1A - fast enough for the MC34063's switching speed. Feedback divider: V_OUT = 1.25 × (1 + R1/R2) = 1.25 × (1 + 3.6k/1.2k) = 5.0V.

Boost configuration note: In boost mode, the inductor, switch, diode, and output capacitor are arranged differently - the inductor goes from V_CC to SC, and the diode points toward the output. Component selection principles remain the same: ferrite core inductor rated for I_PK, Schottky catch diode, and tight switching loop layout.

For boost, pay extra attention to the output capacitor voltage rating - you're stepping up, and a 16V cap on a 24V boost output won't survive.

Common Mistakes When Using MC34063

1. Inductor saturation - the #1 failure mode. The MC34063's darlington switch has no internal current limiting beyond the R_SC sense resistor. If the inductor saturates, current spikes are only limited by R_SC and the switch's ~1.5A ceiling - still enough to destroy the IC in seconds.

Per EEVblog and Chinese forum discussions (eeworld.com.cn, 2023–2025): the most common root cause is selecting an inductor by inductance value alone, ignoring saturation current. So how do you know if your inductor is saturating? Audible whistling is the telltale sign - use ferrite-core inductors rated for at least 1.3× I_PK, and if your MC34063 smokes after a few seconds, the inductor is the first thing to check.

2. Using a standard rectifier diode instead of a Schottky. A 1N4007 has t_rr ≈ 2µs and V_F ≈ 0.9V. At 50kHz, the off-time can be under 10µs - the diode spends 20% of that in reverse recovery. Result: massive switching losses, IC overheating, and poor regulation at light load.

Use a Schottky: 1N5819 (40V/1A), SS34 (40V/3A for higher current), or MBRS340 (40V/3A SMC). The diode's V_RRM must exceed your circuit's peak voltage - in a 24V boost to 36V design, use a 60V Schottky minimum.

3. Wrong timing capacitor C_T value. Too small: frequency goes too high (>100kHz), switching losses dominate, and the IC overheats. Too large: frequency drops below audible range (~15kHz) and the inductor audibly whines.

Also critical: use COG/NP0 ceramic for C_T, not X7R - X7R capacitance varies ±15% over temperature and voltage, shifting your switching frequency by the same amount. Per the UTC datasheet (QW-R103-008.P): C_T = 4×10⁻⁵ × t_on, so a 470pF COG cap gives t_on ≈ 11.75µs and f ≈ 50kHz at 50% duty.

4. Undersized output capacitor - ripple feeds back into the control loop. The MC34063 uses hysteretic (bang-bang) control, not fixed-frequency PWM, so output ripple is inherently larger and more irregular than with modern PWM controllers. The datasheet formula C_O = 9 × I_OUT × t_on / V_RIPPLE is not a suggestion - it's the minimum for stable operation. For a 500mA/5V buck at 50kHz: C_O ≈ 220µF.

But why does ripple matter so much on this particular IC? The output ripple couples directly into pin 5 through the feedback divider - if the ripple waveform distorts the 1.25V threshold crossing, the hysteretic control loop oscillates erratically. Use low-ESR electrolytic or a combination of electrolytic + ceramic at the output.

5. PCB layout: long traces in the switching loop. The loop VCC → R_SC → inductor → output cap → Schottky → GND carries the full switch current with sharp di/dt edges. Per EEVblog discussions (2023–2025): even 20mm of thin trace adds enough parasitic inductance to cause voltage spikes at the switch pin that exceed the 40V abs max.

Keep this loop under 25mm total path length. Use copper pours, not traces. Place the input cap, output cap, and Schottky as close to the IC as physically possible. Route the feedback trace (pin 5) away from the switching node - opposite side of the board if needed.

6. Operating near 40V input without protection. At 24V nominal with automotive or industrial transients, the input can easily spike past 40V. The MC34063 has no internal overvoltage protection. Per Chinese forum reports (eeworld.com.cn, 2024): a 24V truck electrical system with a load dump can hit 60V for hundreds of milliseconds.

Add a TVS clamp (e.g., SMAJ33A) and a small series resistor (1–5Ω) between the raw input and pin 6. The UTC datasheet (QW-R103-008.P) specifies 40V as the absolute maximum, not a continuous rating - operating above 30V continuously requires thermal derating or a pre-regulator.

Frequently Asked Questions

Q1: My MC34063 gets burning hot and dies after a few seconds. What should I check?

A: Inductor saturation, in that order. First: verify the inductor's I_SAT rating - it must exceed your calculated I_PK by at least 30%. Second: check that R_SC is installed and correctly calculated - no R_SC = no current limit = instant destruction on the first switching cycle. Third: verify the catch diode is a Schottky, not a standard rectifier. Per EEVblog and Chinese forum reports (eeworld.com.cn, 2023–2025), inductor saturation accounts for the majority of MC34063 thermal failures.

Q2: What frequency should I run the MC34063 at?

A: 25–60kHz is the sweet spot. Below 20kHz, the inductor audibly whines; above 80kHz, the darlington switch's slow turn-off causes switching losses to dominate and efficiency drops sharply. Per the UTC datasheet (QW-R103-008.P), most practical designs run at 33–50kHz with C_T = 470pF (~50kHz) or 1nF (~25kHz). The frequency varies with duty cycle because the MC34063 uses constant on-time control - don't expect a rock-steady fixed frequency.

Q3: Can I use electrolytic capacitors for the timing capacitor C_T?

A: Absolutely not. C_T must be a stable, low-leakage ceramic capacitor - COG/NP0 dielectric. Electrolytics have terrible tolerance (±20%), high leakage, and their capacitance changes with temperature and age. X7R ceramics drift ±15% over temperature, which shifts your switching frequency. For a reliable design, COG/NP0 is the only correct choice. A 470pF COG cap costs less than $0.02.

Q4: The output voltage is correct with a load, but floats high with no load. Why?

A: Minimum load current not met. The MC34063 has a minimum on-time (and therefore a minimum energy per switching cycle). If the load draws less current than the minimum energy per cycle can supply, the output voltage creeps up. Per the UTC datasheet (QW-R103-008.P): the minimum on-time is set by C_T and the internal oscillator. The fix: add a preload resistor across the output - 1–5mA is usually enough (e.g., 1kΩ for 5V output = 5mA preload). This burns a few milliwatts but keeps the output in regulation at no load.

Q5: Why does my MC34063 circuit fail when I switch from a bench supply to a battery?

A: Input impedance and bypassing. A bench supply has low output impedance. A battery, especially with long leads, has significant series resistance and inductance. When the MC34063 switch turns on and pulls a current pulse, the battery voltage momentarily sags. Without adequate input bypassing (100µF electrolytic + 100nF ceramic within 5mm of pin 6), this sag causes the internal reference and oscillator to glitch, leading to erratic switching and potential latch-up. Per Arduino forum and StackExchange discussions (2022–2024): always test switching regulators with the actual power source, not just a bench supply.

Q6: Can I parallel two MC34063s for more current?

A: Don't - use a different IC. The MC34063 is a hysteretic controller with no synchronization pin. Two MC34063s on the same rail will beat against each other, creating sum-and-difference frequency ripple that's nearly impossible to filter. For >1A output, use a modern synchronous buck converter with integrated MOSFETs (e.g., MP1584, TPS5430) or upgrade to the pin-compatible NCP3063 (ON Semi, 1.5A, 150kHz, with external synchronization pin). The MC34063 is a 1.5A switch IC - 1A continuous output is already pushing it. For higher currents, move to a newer part.

Q7: The feedback resistor divider values - how do I choose them?

A: Keep the total divider resistance between 1kΩ and 10kΩ. V_OUT = 1.25 × (1 + R1/R2), where R1 is the upper resistor (V_OUT to pin 5) and R2 is the lower (pin 5 to GND). Start with R2 = 1.2kΩ and solve for R1. For 5V: R1 = 1.2k × (5/1.25 − 1) = 3.6kΩ. Too low (<1kΩ) wastes power; too high (>50kΩ) picks up switching noise at the feedback node. Per the UTC datasheet (QW-R103-008.P), the CII bias current is ~20nA, so divider currents above 100µA are sufficient.

Q8: What's the difference between MC34063, MC34063A, and MC33063?

A: Temperature range and minor spec differences. MC34063: commercial, 0°C to +70°C. MC34063A: commercial with tighter spec limits on V_REF and current limit. MC33063: industrial, −40°C to +85°C, otherwise identical to MC34063A. All three are functionally interchangeable with the same pinout, same formulas, same external components. Per the UTC datasheet (QW-R103-008.P) and ON Semi cross-reference: buy the grade your temperature spec requires - the silicon is the same.

Related Products

• MC34063AG-S08-R - UTC MC34063A product page, SOP-8
• UC2844G-S08-R - UTC UC2844 PWM Controller, for isolated flyback/forward designs
• MC34063 vs UC2844 - DC-DC Converter vs PWM Controller selection guide