Two-Tone Alarm Circuit

Project overview

An individual assignment: design, simulate, and build a battery-powered alarm circuit that produces two distinct audible tones with a flashing LED, meeting a tight power budget and bill-of-materials cost target, manufacturable on a PCB. The project is complete: simulated, built, and verified on real hardware.

System architecture

The circuit is built around a hex Schmitt trigger IC configured as a pair of relaxation oscillators: one slow oscillator controls how often the tone switches, and a transistor switches an extra capacitor in and out of the second oscillator’s feedback loop to flip between the two audible frequencies.

Full circuit schematic of the two-tone alarm — Full circuit schematic

Selected engineering challenges and decisions

Designing around where the datasheet model breaks down. Component values were calculated from the datasheet’s oscillator equations, then refined through LTspice simulation and bench testing, because the real IC behaviour didn’t perfectly match the simplified datasheet model, its oscillator constant assumes a fixed supply voltage that doesn’t quite hold at the circuit’s actual 3V supply. This mattered because a design that trusts a simplified datasheet equation at face value, without checking it against simulation and a real bench measurement, can end up with the wrong oscillator frequencies on the actual hardware.

Getting a free current-limiting resistor out of the LED. The alarm’s LED doubles as both the visual indicator and the current-limiting element for the switching transistor’s base, removing the need for an extra resistor. Under a tight bill-of-materials and PCB-size budget, every component removed without losing function is a real win, not just a minor simplification.

PCB layout of the two-tone alarm — PCB layout, sized to 50.8 x 36.8mm

Verification or evidence

The design was simulated in LTspice, then built and verified on a breadboard with an oscilloscope, checking the control oscillator frequency, the tone-switching transition, and the LED/transistor bias point all matched what was simulated.

Breadboard prototype of the two-tone alarm circuit — Breadboard prototype

The buzzer’s tone switching, compared between LTspice and the same point measured on the real circuit, confirmed the design worked on real hardware, not just on paper:

LTspice simulation of the buzzer switching between tones

Oscilloscope capture of the buzzer switching between tones on the real circuit — Oscilloscope capture of the same behaviour on the real circuit

The control oscillator’s frequency matched too:

LTspice simulation of the control oscillator frequency

Oscilloscope capture of the control oscillator frequency on the real circuit — Oscilloscope capture of the same oscillator on the real circuit

As did the falling-edge transition that triggers a tone change:

LTspice simulation of the falling edge triggering a tone change

Oscilloscope capture of the same falling-edge tone change on the real circuit — Oscilloscope capture of the same falling-edge behaviour on the real circuit

Current status

Completed: simulated and measured results matched closely across every check, on real hardware. This was a closed university assessment; no further development is planned.

What I learned or am proud of

The datasheet’s oscillator constant not quite holding at this circuit’s actual 3V supply is a small but real example of a bigger habit: a datasheet equation is a model with its own assumptions baked in, not ground truth, and it’s worth checking against simulation and a real bench measurement before trusting it at the edge of its assumed conditions.