In this series of DAC Basics, we start out by briefly discussing three architectures (Figure 1 ). This will give us a good frame of reference as we work through the critical digital-to-analog converter (DAC) specifications and applications.
These architectures are the string DAC, R-2R DAC, and multiplying DAC. In all cases, these devices use a digital input code to generate an analog output signal. Each architecture offers advantages and disadvantages in both the time and signal domains. In this series, we will be unraveling these differences.
String DAC resistor architecture
The string DAC uses a resistor-string architecture (Figure 2 ). One could say that the string DAC looks a lot like a digital potentiometer, and that is true. The string DAC simply has the additional output amplifier in the circuit.
For simplicity, think of the design as a single string of 2N -matched resistors with a reference voltage (VREF ) on top. In this formula, N equals the number of DAC bits. If you have a 12-bit string DAC, theoretically, there would be 4069 resistors. Alternatively, a 16-bit converter would have 65,536 resistors, which is unwieldy! Later in this series we will share some design tricks to reduce the string DAC's resistor count.
As you change the input codes to the DAC, the output amplifier taps into different points on the string. In this manner, the DAC produces the appropriate output analog voltage. Some resistor-string DACs have an option for multiple gain settings. These DACs accomplish different gain settings by switching the internal amplifier's feedback resistor to different values. The user implements these changes with the DAC's control and interface logic.
One thing to notice about the string DAC is that the voltage reference, VREF , is immediately divided by two. The DAC designer does this at the top of the resistor string by using the RDIVIDER resistor. This reference-voltage-division technique ensures that the common-mode input voltage to the internal output amplifier remains within its input common-mode linear range. (These are some references for this section: What does rail-to-rail input operation really mean? and Rail-to-Rail Inputs — what you should know!)
The string DAC is appropriate for a majority of precision DAC applications. In particular, string DACs are found in adjustable references, DC bias control, or closed-loop systems such as digital servos or machine/motor control applications. The string DAC does bring advantages to the table. The output voltage of the string DACs are inherently monotonic. This device exhibits lower glitch performance than the other DACs in this discussion.
R-2R DAC architecture
The R-2R DAC uses a stack of resistors configured in a basic R-2R architecture followed by a buffer amplifier (Figure 3 ). The legs of the R-2R resistors are biased using the ground or a voltage reference in accordance with the input digital code.
This DAC is also referred to as an R-2R ladder or voltage-switching DAC. For each new digital input update, the internal 2R leg resistors are directed to VREF or ground. As the input code increases, more switches are switched from the ground node and to the VREF node. The resulting output voltage is the summing of all the 2R ladder voltages.
The R-2R DAC is most appropriate for industrial applications. With this DAC, each new update involves switching the 2R legs to either the voltage reference or ground. Since the internal resistors and NMOS transistors aren't perfectly matched, the R-2R is more expensive and difficult to manufacture. This is because trim circuits are included and extended test time is required. The gate-switch timing-skews manifest themselves at the output of the DAC as glitches. The glitch is most prevalent when all bits change, or during the most significant bit (MSB) transition, when bits are switching from 7FFFh to 8000h (for a 16-bit DAC). The R-2R DAC typically has excellent low noise, integral nonlinearity (INL), and differential nonlinearity (DNL) performance.
The internal architecture of the MDAC and R-2R DAC are similar in that they use the R-2R architecture. The voltage reference placement in the MDAC device is above the R-2R structure. With this configuration, the output signal manifests itself in terms of current instead of voltage.
The reference voltage input characteristics defines the difference between an MDAC and any other type of DAC on the market. Through this architecture, you can see that a positive reference voltage is applied to the top of the R-2R ladder. In this configuration, the resulting output signal is current.
Most users prefer voltage outputs so, to facilitate the conversion from current to voltage, there is an on-chip resistor that can be used as an external amplifier's feedback resistor. This allows the system designer to configure the external amplifier to their advantage.
For instance, the signal at the output of the external amplifier may be required to range beyond the DAC's supply voltages. This is easily achieved with an external amplifier that has its own power supply. In contrast, the system designer can use a higher speed amplifier that is manufactured in a different high-speed technology.
The MDAC has the same resistor and NMOS matching issues as does the R-2R DAC, making this device equally more expensive and difficult to manufacture. To ensure the best performance, the IC manufacturer uses careful layout techniques to match the on-chip feedback resistor (RFB ) to the R-2R ladder resistors. These matching practices reduce the influence of IC process variations. Like the R-2R DAC, the MDAC typically has excellent low noise, integral nonlinearity (INL), and differential nonlinearity (DNL) performance.
MDACs are found in automatic test equipment or instrumentation. This architecture is capable of a high-voltage output capability. MDAC manufacturers are able to design high-resolution devices (16-bit) with ±1LSB INL and DNL specifications. MDACs require an external current-to-voltage operational amplifier (op amp).
The MDAC, R-2R DAC, and string DAC architectures do not encompass all possible DAC topologies. But, knowing about these topologies will give you a good start on knowing the basics. The rest of this series will compare and contrast these devices in terms of applications and specifications.