Hello and welcome to the Texas Instruments overview of architectures of precision DACs. In this presentation, we will cover the string DAC, including basic architecture, common specification ranges, and a look at the advantages and disadvantages of designing with one. The string DAC is the most straightforward precision DAC architecture. It is simply a collection of resistors and series with a tap node between each of the resistors.

When a digital input code is applied to the DAC, it is decoded and the output switch is moved to the appropriate configuration. For instance, when a full-scale code is applied the topmost switch closes and the output goes to full scale. As with all DACs, the full scale code of a string DAC produces an output of V REF minus 1 LSB.

Notice in this diagram that the full-scale tap point is after the first single resistor in the latter, producing an output voltage after the gain amplifier of 7/8 of V REF. When a zero scale code is applied, the bottommost switch is closed and the output goes to zero scale. Ideally, every resistor down the chain creates a voltage drop equal to 1 LSB. In a straightforward string DAC design there is one resistor switch pair for every code that can be applied to the DAC.

Because of this, as resolution increases the number of resistors and switches in the string DAC design exponentially increase. Since trimming 2 to the number of bits resistors to match one another is not very practical, delivering good DNL and INL performance from a string DAC is heavily dependent on good IC layout. Generally speaking, string DACs have the worst linearity of all precision DAC architectures.

The input impedance for the reference node is usually high impedance due to the volume of resistors in the string. This makes most string DACs low power devices, but it comes at the sacrifice of increased noise at the DAC output. Layout and design techniques are used to reduce the number of resistors required in the design and keep the package size small. Without an interpolating amplifier, even medium resolution string DACs would probably come in impractically large packages. Finally, the simplicity of the string DAC design delivers the lowest cost precision DAC option.

Let's take a look at an example with a 3-bit string DAC. Suppose the code written into the device is 0.2 or binary 010. First, we close the switch that maps to binary 010.

Calculating the voltage at the non-inverting terminal of the amplifier then becomes a simple resistor divider equation. Make sure to take into account that the R DIV resistor has a value of 8 R. Finally, we apply a gain of two through the amplifier for a final value of one fourth the voltage reference. Looking at a 3-bit DAC transfer function we can see that our calculated answer matches the ideal function.

Next, let's look at some of the common DC characteristics of a string DAC. Here, we will use DAC 7562 as an example. Keep in mind that this is a 12-bit part and some of these specs will scale up as resolution increases.

The DNL of DAC 7562 is typically plus or minus 0.05 LSBs with an max of 0.25. This number will scale approximately with the difference in number of bits. The 16-bit version would then typically have about 0.2 LSBs of DNL, four times the 12-bit spec.

The INL is typically 0.3 LSBs with a max of 0.75 LSBs. This spec scales at 2 to the power of the difference in bits. Again, for the 16-bit case, the number would scale up to 16 times the original, giving about 5 LSBs typical. Remember that this is just a shorthand way to estimate the difference between the specs of two different resolutions.

As we previously alluded to, the difficulty in trimming these parts means that string DACs don't deliver the best INL and DNL performance. The offset, full-scale, zero code, and gain errors are dependent primarily on the output of the amplifier that follows the resistor ladder. You'll notice that the offset and gain errors are extrapolated from the two-point line. What this means is that we measure the output at a few codes above zero and a few below full scale, draw a line between the two, and extrapolate out what the zero and full scale values are. We use this extrapolation instead of taking the actual zero in full-scale codes due to nonlinearities caused by headroom of the amplifier.

Now, we can look at the common AC specs for a string DAC. We again show DAC 7562. Settling time is relatively fast, although not the fastest available. The slew rate of the amplifier does contribute heavily to the length of the settling time.

A benefit of few switches in the resistor ladder changing at any given time is this low glitch impulse. You'll find a spec to hold true across most string DACs. Lastly, the noise spec isn't great. The effective resistance of the ladder causes a lot of noise in the system compared to other architectures.

Lastly, we'll wrap up with the pros and cons of a string DAC. First, the advantages. The high level of simplicity associated with a string DAC design, output buffer design, and low-touch trimming techniques keep the cost of manufacturing a string DAC low. Few switches moving over any given code transition keeps glitch energy low. The structure of the resistor string guarantees monotonicity since negative resistors can't exist.

The large impedance of the resistor string keeps the design low power. And generally speaking, this design can be realized in a small package.

Second, the disadvantages. As the resolution of a string DAC increases we need exponentially more resistors, which would lead to exceptionally large packages to contain 2 to the N resistors. Since string DACs have limited or no trim, the architecture inherently offers worse linearity than others.

The switching structure leads to slow code-to-code transitions. And as a result, the device update rate is limited. An output buffer is required to isolate the resistor string from the point of load. Typically, this buffer is included on silicon, which can be considered either a good or bad thing depending on the design. Finally, most string DACs have higher noise specs due to the noise generated by such a large number of high-value resistors in the network.

Thank you for watching this video on the string architecture for precision DACs. Please watch our other videos on precision DACs to learn more.