From Analogue to PIC16F72: How We Put a Microcontroller in the Inverter

This is Part 01 of Source Code of an Industry — a series from real production-era inverter control technology, written so the engineering story is not lost.

In 2000, under my leadership at Su-Kam, we designed our first MOSFET inverter. That first generation was analogue — there was no microcontroller. The same year we committed to MCU control. I was actively involved in selecting the Microchip PIC16F72. In 2001 we launched our first MCU-based inverter. That is where the microcontroller journey started for our products — and where much of the Indian inverter industry followed.

I learned Assembly myself for practical knowledge of how the machine worked. Production software was written by the software team. Understanding the code made better technology decisions possible.

The public timestamp — December 2001

On 10 December 2001, The Times of India (New Delhi) published a short piece titled “Digital wonders.” It described Su-Kam’s new digital inverter as fitted with a “unique micro-controller based intelligent control circuit” and a “unique micro controlled based charger,” listed at Rs 9,400.

That clipping is the mainstream press timestamp for the leap this series documents: policy moved from analogue pots into firmware.

What analogue control could — and could not — do

Before the MCU, a typical low-cost inverter brain was a hardware state machine:

• Comparators and dividers for battery cut-off and mains detect
• Potentiometers for production and field trim
• Discrete timers for delays and buzzers
• Hardwired relay logic
• Fixed or lightly modulated gate drive

Limits that blocked the roadmap:

1. One PCB spin ≈ one behaviour
2. Pot drift — units differed in the field
3. Weak fault language for dealers
4. Soft-start, multi-stage charge, UPS vs AVR personalities were expensive in discrete parts
5. Behaviour lived in tribal knowledge, not a versioned listing

The 2000 analogue MOSFET inverter proved the power topology. The 2001 MCU inverter proved the control topology.

Why PIC16F72

We needed low BOM cost, deterministic AC timing, ADC for Vin / Vbat / current / temperature, GPIO for relays / LEDs / buzzer / fan, and watchdog robustness — without a DSP price the retail market would not pay.

PIC16F72 sat in that window: 8-bit MCU, HS oscillator, WDT, ADC, Timer0/Timer1, external interrupt. Enough to own mains policy, charge stages, modified square-wave timing, and protection language on one chip.

Design architecture — MOSFET power, MCU brain

The MCU did not replace the transformer or the MOSFET bridge overnight. What changed was the brain.

Advantages over inverters without microcontrollers

The commercial advantage was repeatable intelligence at retail BOM — not “MCU for MCU’s sake.”

Operating modes became software

Analogue boards implemented modes as wiring. The MCU implemented a state machine: boot → ready → mains mode ↔ inverter mode → fault sleep.

Core technology unlocked by the MCU — modified square wave

A fixed 50% square wave is easy in hardware. Regulating it under load is not. Timer interrupts made modified square wave (MSW) practical: each AC half-cycle is a controlled-width pulse on one bridge leg, then off; the complementary leg takes the next half-cycle. Pulse width tracks load and battery state.

That pattern — ISR-timed pulse width on a square carrier — is why technical teams across the industry could ship retail “square wave” inverters that still held output reasonably well, without sine tables or DSP math.

Sensing map

Five analogue quantities (mains, battery, charge current, load current, temperature) plus switch, zero-cross, short-circuit interrupt, fuse sense, and UPS/AVR select fed one policy engine that drove charger relay, changeover relay, charge gate, inverter PWM legs, fan, LEDs, and buzzer.

Why the industry followed

Once a low-cost PIC could own mains policy, charge stages, MSW timing, and fault language on a familiar MOSFET + transformer stage:

1. Dealers got more consistent products
2. Factories got variants without new analogue nets
3. Competitors could see the pattern on the bench and replicate the architecture
4. The market standardised on MCU + MOSFET + transformer MSW, then later moved to sine — but control never went back to pure analogue

Limits — and honesty about roles

MCU MSW was not the end of history. Waveform is not pure sine; 8-bit ISR PWM has finite headroom; monolithic Assembly is powerful to ship and harder for large teams to extend. Later generations addressed waveform with newer MCUs and SPWM.

Roles: I led the MOSFET inverter generation and the decision to adopt MCU control, including involvement in selecting PIC16F72, and learned Assembly for practical understanding. Production firmware was written by the software team. This series will show real listings as engineering evidence, with credit where the lines of code belong.

What comes next

Part 02 — deep dive: modified square-wave inverter control from production source (timer reload, half-cycle length, pulse-width law).
Part 03+ — charger algorithms, protection policy, and later MCU generations from the archive.

The hinge year remains clear: analogue MOSFET inverter in 2000 → PIC16F72 MCU launch in 2001 → industry follows the microcontroller path.

— Kunwer Sachdev