By David Herres
Typical real-world phenomena (most sensor output, acoustic information in air-pressure variations, temperature fluctuations, etc.) are analog rather than digital in nature. If translated into an electrical signal, they will vary continuously on a sliding scale rather than being encoded in discrete steps, binary or otherwise.
If digital processing and memory are to be used, the analog information must first be digitized. This takes place via an analog-to-digital converter (ADC), as outlined in a previous article. Effective methods have been developed to accomplish this process at a speed and resolution that is appropriate for the application.
After processing and storage in memory, the digtitized information must be converted back to analog form. (This is ironic, because within the individual human the signal must be again converted to digital form before our network of neurons can process and consign the data to human memory.)
The device that accomplishes the conversion is known as a digital-to-analog converter (DAC). It can be implemented either within an IC or as a discrete device, as represented by a high-performance oscilloscope.
A very basic DAC is the pulse-width modulator, frequently appearing as part of a variable-frequency ac motor drive (VFD). A VFD varies the speed of ac motors (usually three-phase induction) in response to changes in the width of electrical pulses synthesized within the control equipment.
Pulse-density conversion is used in oversampling and interpolating DACs. One example is the delta-sigma DAC. The end result is that noise is shifted out of the low frequency area into a higher frequency zone where it is not an issue.
A very fast DAC is the binary-weighted device, but accuracy is less than required by many applications, and the expense for a high-resolution device may be prohibitive due to the large number of components required.
The R-2R ladder DAC is noteworthy because it is inexpensive and simple to manufacture since it needs only two resistor values. In fact, by means of parallel or series connections, a single resistor value will suffice. Moreover, regardless of the number of bits, the output impedance is consistently equal to the value of the resistor chosen, which simplifies filtering in down-stream analog circuitry. Superposition and Thevenin Equivalent circuit analysis can be applied to make sense of this DAC. These theorems simplify Ohm’s law calculations and make possible determination of the output impedance.
The thermometer-coded DAC consists of a resistor for every possible digitization value. The device is characterized by exceptional speed and accuracy but its high cost precludes most applications.
Hybrid DAC’s combine two or more designs in a single integrated circuit, so this is a widely used solution. Because of the large and growing numbers of DACs in use worldwide, we can expect to see a further decline in the cost of these devices.
For the basics it seems to be enough!
The R-2R circuit is a workhorse but there is a down side and that is a buffer amp is needed. If an op-amp is used, do not expect rail to rail performance rather about .3 v above the negative rail to about VCC-.5 V from the top rail. I have used a MOSFET with the R-2R circuit driving the gate. As you can imagine, this is not even nearly a linear circuit but in some cases a linear circuit is not wanted. With the circuit I am using, the non-linearity is what I wanted because I was controlling an LED. It is not linear but it is repeatable and I can work with non-linearity much better than I can unrepeatable.