A Digital to Analog Converter (DAC) translates discrete binary numbers into continuous analog voltages. If you are asking what the exact conversion is for a common 12-bit DAC (like the MCP4725) running on a 3.3V reference, the answer is 0.806 mV per digital step. The governing formula is V_LSB = V_REF / 2^N. Substituting our values: V_LSB = 3.3V / 2^12 = 3.3 / 4096 = 0.0008056V. This step size—the Least Significant Bit (LSB)—is the fundamental "unit conversion" of any DAC, dictating exactly how much the analog output voltage changes for every +1 increment in your digital code.

The Baseline Assumption: This calculation assumes an ideal, perfectly stable Reference Voltage (V_REF) and a monotonic transfer function. If your V_REF sags under load, the physical voltage output will deviate from this theoretical conversion.

The Core Conversion: Digital Codes to Analog Voltage

At the bench, understanding what a digital to analog converter does requires looking past the binary math and focusing on the physical output stage. Whether the DAC uses an R-2R resistor ladder or a string-DAC architecture, its job is to map a digital integer (from 0 to 2^N - 1) to a specific voltage between Ground and V_REF.

When selecting a DAC for a project, the bit-depth and the reference voltage are the two variables that fix your conversion scale. Below is a spec-sheet comparison of the most common DAC ICs used in hobbyist and industrial prototyping in 2026, showing their real-world resolution and step sizes.

Table 1: Common DAC IC Specifications and LSB Step Sizes
Part Number Resolution (Bits) V_REF Range LSB Step Size (at Max V_REF) Interface Approx. Price (2026)
Microchip MCP4725 12-bit 2.7V – 5.5V 1.34 mV (at 5.5V) I2C $1.20
Texas Instruments DAC8568 16-bit 2.7V – 5.5V 83.9 µV (at 5.5V) SPI $8.50
Analog Devices AD5686R 16-bit 2.7V – 5.5V 83.9 µV (at 5.5V) SPI $11.00
Maxim MAX5216 16-bit 2.7V – 5.5V 83.9 µV (at 5.5V) SPI $6.00
ESP32 Internal DAC 8-bit 0V – 3.3V 12.9 mV (at 3.3V) Memory Mapped $0.00 (On-chip)

For a deeper dive into the internal architectures of these chips, the All About Circuits DAC basics guide provides excellent schematic breakdowns of R-2R versus Sigma-Delta topologies.

How Reference Voltage Shifts the Output Scale

If you are coming from AC power calculations where answers shift wildly between 120V single-phase, 230V, and 3-phase systems, DAC conversions operate on a different axis. Here, the equivalent shift happens across 3.3V microcontroller logic, 5V legacy logic, and 10-12V industrial 4-20mA control loops.

Changing your V_REF fundamentally alters the "weight" of every digital bit. If you are designing a precision audio or sensor-biasing circuit, you must account for how tolerance and thermal drift in your voltage reference shift the analog output.

The table below demonstrates how the LSB step size shifts across a ±20% tolerance range for a 16-bit DAC (like the DAC8568) targeting a nominal 5.0V reference. This is critical when diagnosing why your 16-bit DAC is only performing like a 14-bit DAC in practice.

Table 2: 16-Bit DAC LSB Step Size vs. V_REF Drift (±20% from 5.0V Nominal)
V_REF Shift Actual V_REF Calculated LSB Step Size Total Full-Scale Range (FSR)
-20% (Brownout) 4.000 V 61.03 µV 3.999 V
-10% (Sag) 4.500 V 68.66 µV 4.499 V
Nominal (Ideal) 5.000 V 76.29 µV 4.999 V
+10% (Overvoltage) 5.500 V 83.92 µV 5.499 V
+20% (Fault) 6.000 V 91.55 µV 5.999 V

Notice that a mere 0.5V drop in your reference rail shrinks your step size by nearly 8 µV. In precision applications, this is why we use dedicated low-drift voltage reference ICs (like the TI REF5050) rather than relying on the microcontroller's 3.3V LDO output.

When Theoretical Conversion Becomes Meaningless

Math on a spreadsheet is perfect; silicon on a breadboard is not. The theoretical bit-to-voltage conversion becomes practically meaningless under three specific bench conditions:

  1. When V_REF Noise Exceeds the LSB: If you are using a 16-bit DAC at 5V, your LSB is 76.29 µV. If your power supply has 150 µV of high-frequency switching noise, your bottom two bits are just toggling random noise. The conversion math holds, but the physical signal is garbage.
  2. When DNL Exceeds ±1 LSB: Differential Non-Linearity (DNL) measures the deviation between actual analog steps and the ideal 1 LSB step. If a DAC has a DNL error of -1 LSB or worse, it suffers from "missing codes." Sending a digital command of 1000 might yield the exact same voltage as 1001. As noted in Analog Devices' RAQ on data converter basics, missing codes destroy the monotonic transfer function required for closed-loop control systems.
  3. When Driving Unbuffered Low Impedance: Many basic DACs (and almost all internal microcontroller DACs) lack an internal output buffer amplifier. If you connect a 12-bit DAC directly to a 50-ohm load, the output voltage will sag dramatically due to the internal output resistance of the DAC, rendering the calculated V_LSB entirely invalid.

Frequently Asked Questions About DAC Conversions

Can I use a PWM pin and a low-pass filter instead of a DAC IC?

Yes, but with severe trade-offs. An ESP32 PWM pin filtered through a simple RC low-pass filter can generate a pseudo-analog voltage. However, the conversion is highly non-linear, susceptible to digital switching noise, and typically limits you to 8-bit or 10-bit effective resolution. For audio or precision sensor biasing, a dedicated I2C/SPI DAC IC is mandatory.

What happens if I send a digital code higher than the DAC's maximum resolution?

If you send a 16-bit value (e.g., 65535) to a 12-bit DAC (max code 4095), the behavior depends on the library and the IC. Most I2C DACs will simply truncate the lower bits or mask the upper bits, resulting in an unexpected and often drastically lower voltage output. Always bit-shift your variables to match the DAC's N resolution before transmitting over I2C or SPI.

Why does my DAC output voltage max out at 3.8V when V_REF is 5V?

This is a classic "rail-to-rail" output stage limitation. Older or cheaper DAC output amplifiers cannot swing all the way to the positive supply rail. They may suffer from a 1V to 1.5V headroom limitation. If you need the full 0V to 5V swing, verify the datasheet's "Output Voltage Range" specification and ensure the DAC features a true rail-to-rail output buffer.