DAC or digital pot?

Which type is suitable for a given application and what problems these devices cannot solve

BY REINHARDT WAGNER
Maxim Integrated Products
Sunnyvale, CA
http://www.maxim-ic.com

D/A converters (DACs) and digital potentiometers (digipots) are very similar at first glance. Many similarities of DACs and digipots allow one to use both device types in some applications. The requirements of a given system, however, often limits the choice to one or the other. This article discusses what they have in common, how they differ, and what problems these devices cannot solve.

Digipot characteristics

The wiper of a digipot, as for an electromechanical potentiometer, can be set anywhere between the two ends H (high) and L (low) of the resistor string. Wiper settings are made via a digital interface (see Fig. 1) , either directly, with a digital value (D), or incrementally, via up/down pulses. Unlike an electromechanical pot, the digipot imposes restrictions on the voltages applied to its pins.

Fig. 1. A digipot used as a voltage-output D/A Converter requires an output buffer amplifier

The range of voltages at the resistor ends (L and H) must be within the range defined by Vcc and Vss/Gnd. For a digipot with N setting levels and no wiper load, the wiper voltage is given by Equation 1:

Vo =VL +D/N(VH -VL ) (1)

Output resistance Ro at the wiper connection equals the sum of the wiper resistance and the equivalent string resistance, and depends strongly on the wiper position. Wipers have an intrinsic resistance of a few hundred ohms. If the H and L ends are driven with zero source impedance, the following equation gives the output resistance, RO :

RO = RW + RE [D/N – (D/N)2 ] (2)

Because the total end-to-end string resistance (Re ) of a typical digipot is between 10 kΩ and 100 kΩ, its wiper resistance (Rw ) is significant only at the extreme wiper positions. Otherwise, Rw can be disregarded. The normalized output resistance (excluding the effect of Rw ) varies with D/N as shown in Fig. 2 .

Fig. 2. The normalized output resistance Ro /RE (excluding the effect of Rw ) varies with D/N.

DAC characteristics.

Although current-output DACs such as the MAX5494 are readily available, most DACs now come with an output-voltage buffer that provides a low output impedance. The output buffer in Fig. 1, for example, allows the digipot in that figure to function as a voltage DAC. Thus, the output voltage of a DAC with resolution n is proportional to the product of the reference voltage Vref and the applied digital value D:

VO = D/(2n -1)xVref (3)

The output voltage of a DAC, as for a digipot, is proportional to both the programmed digital value, D, and the reference voltage (Vref or Vh ). A DAC is defined as a multiplying type if its reference voltage can be changed dynamically, and a multiplying DAC is called a 4-quadrant type if it can accept bipolar D and Vref values in accordance with Equation 3. When D and Vref are both negative, the 4-quadrant multiplying DAC generates a positive output voltage (VO ). That capability allows the device to invert Vo in response to the combination of negative D values and positive input voltages. An integrated amplifier ensures a low output impedance for the voltage DAC.

Similarities between DACs and digipots

Digipots offer the same basic function as do DACs. If the L terminal of a digipot is connected to ground, it operates as a voltage DAC but with high output resistance, and has the same transfer function as a DAC. That connection allows a digipot with output buffer to be used as a voltage DAC, as shown in Fig. 1.

Digipots and DACs are also available as nonvolatile devices, which means they retain their digital input value when the supply voltage is removed. Because digipots and DACs are often used to “calibrate out” the non-ideal characteristics of an analog signal path (offset and gain, for example), these input compensation values should not be lost during a power failure or system restart. Despite the DAC’s lack of an L terminal, the DAC has features that are precluded in a digipot due to its inherent structure.

Differences between DACs and digipots

The number of wiper taps on a digipot is limited by layout and technology restrictions. A high-end digipot might have 1,024 taps, equivalent to a DAC with 10-bit resolution, but DACs are currently available with resolutions of 24 bits and more.

The tap spacing of today’s digipots can be linear or logarithmic. Logarithmic potentiometers are commonly used for controlling audio volume. By compensating for the nonlinear relationship between voltage and the perceived sound intensity, they produce an apparent linear relationship between knob position and sound level. Linear multiplying DACs can also be used to achieve a nonlinear relationship between D and VO , but they need higher resolution for that purpose. Because DACs with higher resolution are more expensive, logarithmic digipots tend to be more popular.

Digipots generally support only relatively slow signals, but DACs can pass and generate extremely fast signals. DACs are often used for fixed or slowly varying voltage generators, while specialized DACs are also available for high-speed and RF applications.

Speed in D/A conversion is defined by three parameters: digital update rate, output settling time, andfor digipots and multiplying DACsbandwidth. High bandwidth is necessary for video-signal multipliers and other fast multiplying applications, while high update rates and fast settling times are required in applications such as direct digital syntheses (DDS), used in digital signal generators and in modern RF transmitters.

The low update rate and long settling time of digipots makes them unsuitable for DDS applications. Fast DACs used in DDS systems accept their digital inputs as parallel words, whereas digital pots usually have a relatively slow serial-bus interface.

Because the structure of a digipot ensures that its differential nonlinearity (DNL) is always greater than –1, its output is guaranteed to be monotonic. On the other hand, the DNL of a DAC can be less than –1, so monotonicity is guaranteed only for DACs whose DNL is equal to or greater than −1.

A digipot can be used as a variable resistor if one resistor end is connected to the wiper. Such digitally settable resistors are relatively inaccurate, however, because the total resistance of the digipot is subject to a manufacturing-dependent tolerance of 20% to 25%. In contrast, voltage-output DACs cannot be used as resistors. Current-output DACs can be used as variable resistors, but with the restriction that the DAC output must be connected to a virtual ground such as the summing node (inverting input) of an op amp.

The output impedance of a voltage-output DAC is small and fairly constant, whereas the output resistance of a digipot is much higher, and varies strongly with the wiper setting, as shown in Fig. 2. The output resistance of a common 100-kΩ digipot, for example, ranges from about 100 Ω to a maximum of 50 kΩ.

Because digital pots are set with upward/downward pulses or via a serial interface, they require few pins and usually come in small packages. Slow DACs often come in small packages for the same reason. A fast DDS DAC, on the other hand, has parallel pins for fast data loading, and therefore resides in a larger package.

One can obtain multiple DACs or digipots in a standard package. Six digipots in one housing are currently the maximum, but substantially more D/A converters per package are possible. The current state of the art is 32 DACs in a single package. Digipots are generally more cost-effective than DACs, however. Their small chip surface and small package size have a direct effect on the cost of manufacturing.

When an application calls for resistor or voltage dividers with freely defined H and L voltages, digipots are essential. One example is a difference amplifier with digitally controlled gain. The gain factor for this circuit can be programmed via the serial interface of the DS1867 dual digipot or the newer MAX5494. Thanks to wiper positions that are stored in the EEPROM of that device, the differential-gain value is retained even after a loss of power.

The many similarities of DACs and digipots allow one to use both device types in some applications. The requirements of a given system, however, often limits the choice to one or the other. ■