In 1983, Motorola released a DC-DC controller in an 8-pin DIP package. Inside: about 50 transistors. The datasheet had a few pages of formulas and three reference schematics. No thermal shutdown. No current-mode control. No synchronous rectification. Not even a fixed switching frequency.
By 1990, Motorola had already marked it "Not For New Designs" and replaced it with the MC34063A - a slightly improved version with marginally better specs. They expected it to fade out within a few years.
It's now 2026. TI still lists the MC34063A as "Active." ON Semiconductor (the Phoenix division that was once Motorola Semiconductor) still manufactures it. A dozen other companies produce pin-compatible clones. It ships well over 100 million units per year and turns up in car chargers, routers, TV power supplies, microwave oven controllers, and set-top boxes.
No chip was ever supposed to last this long. Here's why the MC34063 did.
The MC34063 is unapologetically simple. Inside the 8-pin package: a 1.25V bandgap reference, a comparator, an oscillator (relaxation type, charging and discharging a timing capacitor between two thresholds), a driver stage, and a 1.5A peak-current Darlington switch. That's it.
No undervoltage lockout. No soft-start. No thermal protection - if it overheats, it dies. The switching frequency drifts with load and input voltage because the control scheme is constant-on-time, not fixed-frequency PWM. An EEVblog commenter described the switching waveform as "a huge mess."
And yet.
That same simplicity is the MC34063's superpower. Because it uses hysteretic control instead of a fixed-frequency PWM loop, there is no control loop to compensate. You don't tune a Type II or Type III compensator. You plug numbers into the datasheet formulas for RSC, CT, L, and Cout, and it works. First-time SMPS designers get a working power supply on their first PCB. That alone explains half of its longevity.
In the early 1980s, the electronics industry was going through a power-supply revolution. Linear regulators - 7805, 7812, LM317 - were simple but inefficient. A 12V→5V linear regulator at 1A burns 7 watts as heat. The same conversion with an MC34063 at 75% efficiency burns about 1.7 watts. Not great by modern standards, but 4× better than a linear regulator, in an era when switching supplies were still exotic.
The MC34063 condensed what had previously been a board full of discrete components - oscillator, comparator, reference, driver, power switch - into a single chip. It wasn't the most efficient solution. It wasn't the quietest. But it was universal: one IC, one set of datasheet formulas, and you could build a buck converter, a boost converter, or an inverting supply. No other chip at the time offered all three topologies from an 8-pin package.
The timing was perfect. The PC revolution was starting. Modems, peripherals, automotive electronics, set-top boxes - suddenly everything needed DC-DC conversion, and the MC34063 was the cheapest, most flexible way to get it. Motorola's application notes did the rest. The formulas were good enough. Engineers copied the reference schematics, tweaked the values, and moved on to the next problem.
At volume, the MC34063 costs about 11 cents. That's not a metaphor. One Hackaday commenter calculated it's cheaper than peanuts by weight. The production lines paid for themselves in the 1990s. At this point, the cost of an MC34063 is raw silicon, packaging, and shipping.
This has a specific consequence: once an MC34063 is in a product's BOM, there is zero financial incentive to replace it. A newer chip might be 15% more efficient. It might switch at 10× the frequency and let you shrink the inductor. It might have thermal shutdown and soft-start so the product doesn't fail in strange ways. But it costs 45 cents instead of 11. Over 10 million units, that's $3.4 million in added BOM cost. For what? A product that already passed EMI, already passed thermal qualification, and already ships to customers who don't care what's inside the power supply.
The MC34063 doesn't win on performance. It wins on installed base. The design is proven. The supply chain is infinite. The price is unbeatable. Until one of those three things changes, it stays.
Automotive 12V accessories. The MC34063 is resilient to the voltage spikes, load dumps, and general electrical chaos on a car's 12V rail. Its wide 3–40V input range absorbs transients that would kill a more delicate 5V-only switcher. Every car USB charger you've ever bought at a gas station likely has an MC34063 or one of its clones inside.
Cheap networking gear. Routers, switches, modems - the kind that cost $30 and ship with a 12V wall wart. The MC34063 steps that 12V down to 5V and 3.3V for the SoC and Ethernet PHY. Nobody's opening these boxes to admire the power supply IC. They just need it to work for 5 years without failing. The MC34063 obliges.
TV and appliance power supplies. The standby power rail in a TV or microwave oven doesn't need high efficiency or low noise. It needs to be cheap, reliable, and survive being left plugged in for a decade. The MC34063's hysteretic control - messy as it is - has no control loop that can go unstable after 5 years of capacitor aging.
Hobbyist bench supplies. The DIP-8 package is the secret weapon. You can breadboard an MC34063. You can solder it onto perfboard. The 50–100 kHz switching is slow enough that layout is forgiving - one Arduino forum user reported their breadboard MC34063 performed nearly identically to their PCB version. Try that with a 1.2 MHz synchronous switcher and you'll spend the afternoon debugging oscillations.
You're making 50,000+ units of a cost-sensitive product. At 11 cents per IC, at 200mA load, at room temperature - the math works. The efficiency penalty is negligible in absolute watts. A 2% efficiency gain from a modern chip doesn't pay for its 4× price premium.
Your product has a 12V or 24V input that's dirty. Automotive, industrial, anything with long cable runs and inductive loads. The MC34063's 40V absolute maximum and hysteretic control tolerate input transients without latching up or entering strange failure modes.
You need a negative rail for an op-amp or LCD bias. The MC34063 in inverting configuration is the simplest way to generate −5V or −12V from a positive rail. Two resistors, an inductor, a diode, and a capacitor. No exotic magnetics, no charge-pump IC with a specific output voltage.
You want something that will still be available in 10 years. The MC34063 isn't going anywhere. No single-supplier risk. If TI discontinues it tomorrow, UMW, XINLUDA, Diodes Inc., Taiwan Semi, and ON Semi will still have production lines running.
You're designing a battery-powered product. 75% efficiency means 25% of your battery energy becomes heat. A modern synchronous switcher at 93% will nearly double your runtime from the same battery.
Board space is tight. At 50–100 kHz, the inductor is large (100µH for a typical 12V→5V buck). At 400 kHz with an XL6009, it's 22µH. At 1 MHz with a modern synchronous part, it's 4.7µH - smaller than the MC34063's timing capacitor.
You have sensitive analog or RF circuits nearby. The MC34063's variable-frequency, hysteretic switching produces a noise spectrum that's broad, unpredictable, and hard to filter. A fixed-frequency, current-mode PWM controller puts the noise at a known frequency where a notch filter can kill it.
You need more than ~1A output. The 1.5A peak switch current translates to maybe 0.5–0.75A of continuous output, depending on topology and input voltage. For 2A+, use an LM2596 (buck), XL6009 (boost/SEPIC), or a modern synchronous part.
Nobody set out to make an immortal power supply IC. Motorola's engineers designed a chip that was good enough at the right price at the right time, and then the installed base took over. Every company that has ever swapped an MC34063 into a product created one more reason for the supply chain to keep making it.
There's a lesson here for IC designers: simplicity compounds. A simple design invites second-sourcing. Second-sourcing drives down price. Low price drives adoption. Adoption creates an installed base that resists change. Forty years later, a chip that was "Not For New Designs" in 1990 is still in every car charger on Earth.
The MC34063 will probably outlive most of the chips designed to replace it. Not because it's better - it isn't. But because it's done. The design is frozen. The supply chain is permanent. The price has nowhere left to go but slightly up with inflation. And until every car USB charger and cheap router on Earth switches to USB-C PD with integrated GaN converters, the MC34063 will keep doing exactly what it's been doing since 1983.
A: Motorola's bipolar IC design team in the early 1980s. No individual engineer's name appears in public records - it was a corporate product from an era when IC designers didn't get bylines. The earliest known datasheet is Motorola Rev 5 from 1996, by which point it was already a mature, widely second-sourced part.
A: Yes, actively. Texas Instruments lists it as "Active." ON Semiconductor (the former Motorola Semiconductor) still manufactures it. UMW, XINLUDA, Diodes Inc., and Taiwan Semi all produce pin-compatible versions. Combined shipments are estimated at well over 100 million units per year.
A: It was designed in 1983, when thermal protection circuits added die area and cost. The MC34063 targets cost-sensitive applications where the designer is expected to ensure adequate cooling. If you need thermal shutdown, the NCP3063 (pin-compatible upgrade) has it.
A: ON Semiconductor is the direct successor. ON Semi was spun off from Motorola's semiconductor division in 1999 and continues to manufacture the MC34063A on the original process lines. TI's version came through its acquisition of National Semiconductor's portfolio.
A: No official number exists, but a reasonable estimate puts it in the billions. At 100M+ units/year for at least 25 of its 40+ years in production, the cumulative total is somewhere north of 2–3 billion chips. It's in the same league as the 555 timer and the 7805 regulator.





