Signal processing has come a long way from the days when early computers used processors consisting of bit-slice chips. Now, in the 21st century, the widespread use of integrated circuits and standalone digital signal processing (DSP) chips have radically revolutionized electronics and, in turn, the fabric of societies around the world.
Once the size of entire rooms, powerful computers that read, process, and store data from all over the world can now easily fit in our pockets. While the implementation of DSP has certainly changed the world to date, it will only continue to further technological advancement. But what exactly is digital signal processing and how is it used? This piece will review what DSP is, its different types, and key considerations when choosing a DSP for an application.
Definition of Digital Signal Processing
DSP is carried out by specialized DSP chips that convert real-world signals – such as audio, video, temperature, pressure, or position signals – and digitize them to be mathematically manipulated. These chips can be thought of as like calculators that rapidly produce complex results. Mathematical operations like addition, subtraction, multiplication, and division are applied to analog signals to produce an output.
DSP is constantly taking place – after all, it has formed the foundation for the digital world today. An example of DSP in practice is whenever a video is recorded with a smartphone. Analog-to-digital converters (ADC) first convert analog signals, such as audio or video, into a digital format consisting of 1’s and 0’s. A DSP chip then processes those now-digital signals and feeds that information back to the user for practical use in the real world through the means of a digital-to-analog converter (DAC). Additionally, the digital signals manipulated by the DSP chip are stored in memory and can be played back on demand. Those signals are simply taken from memory and passed through the DAC again for the end-user. What results for the end-user is a replayable video complete with the captured sounds and images.
This is only one example, however. The potential applications of DSP are expansive – from video and audio recording to facial recognition, to sonar and radar tracking.
7 Steps to Digital Signal Processing
7 key steps occur in digital signal processing to produce the desired result:
- An analog input signal is received.
- A pre-filter removes unwanted high frequencies, or noise, from the analog input signal.
- An ADC converts the filtered analog signals into digital signals.
- A DSP chip analyzes, processes, and commits the digital signals to memory.
- DAC devices convert the digital signals back to analog signals.
- A post-filter removes noise from the generated analog signal.
- An analog output signal is provided.
Fixed-Point or Floating Point DSP?
Digital signal processing is split into two categories – fixed-point and floating-point DSP. The type of DSP used dictates how signals and data are stored and manipulated.
Fixed-point DSP chips are designed such that integers are represented and manipulated using a minimum of 16 bits. This yields up to 216 possible bit patterns, or 65,536. Duly referred to as fixed-point because those numbers are represented by a fixed amount of digits either before or after the decimal point. The accuracy and precision fixed-point DSP chips can support are typically less than that of floating-point.
Floating-point DSP chips are designed to manipulate and represent rational numbers through a minimum of 32 bits – effectively yielding up to 232 possible bit patterns, or 4,294,967,296. The floating-point variety allows the decimal point of a number to “float” relative to other significant figures within the number. While the fixed-point variety can be used to represent the numbers 123.45 or 1234.56, the floating-point variety can represent numbers such as 1.234567 or 0.001234567, etc.
As a result, floating-point digital signal processing chips can support a higher dynamic range of values when compared to fixed-point and can represent both exceptionally small and large numbers. Therefore, whenever computationally intensive applications are required, floating-point DSP chips are ideally suitable.
What to Consider When Choosing DSP
For any designer, it’s important to minimize the cost of a design as much as possible while retaining the required functionality and safety. It’s no different when designing a system that will use DSPs. Fixed-point DSP chips are typically priced less than their floating-point counterparts due to their ease of manufacturing. As systems process larger pools of data and/or require greater accuracy and precision, the use of floating-point digital signal processing chips is warranted, making the end result more expensive.
The speed at which a designer can develop a product determines whether or not that product enters the market before the competition. As such, the type of DSP used in a product requires much thought. Floating-point DSPs are generally easier to program and develop algorithms for due to their ability to obtain more-precise values when compared to the fixed-point variety. Fixed-point DSP chips require more handiwork to compensate for and remove noise. Whereas, floating-point DSPs are much easier to manipulate but come at a higher cost.
The performance of a digital signal processing chip is an important factor in the overall success of a product. Designers desire DSP formats that efficiently process algorithms – and while both formats can obtain desired results, designers should evaluate how well the formats reach those results. For example, having a floating-point format solve fixed-point tasks can lead to greater power draw – impacting both power and cooling requirements. Ideally, designers should balance cost with performance while edging out their competitors to obtain a marketable product.
Innovate and Create with Digilent
The widespread implementation and use of DSP in electronics have played a fundamental role in the transition to the digital world known today. With DSP, the data that can be processed is nearly limitless, which has led to an exponential increase in computing power to deal with these large pools of data and consequently a better understanding of the world and universe at large.
Digilent has enabled engineers, researchers, scientists, and students with innovative FPGA and SoC-based hardware-software systems. From competitive pricing and comprehensive documentation to the portability of our products, we have the right solution for you. For more ideas and tutorials on getting your DSP application up and running, visit our Reference Center.