Analog to Digital Converter
An Analog to Digital Converter is used to convert an analog signal into a stream of digital values. This is done by sampling the original signal and using a reconstruction filter to reproduce the original signal from these discrete values.
Digital signals are well defined and orderly, making them easier for electronic circuits to distinguish from noise, which is chaotic. The resolution of an ADC specifies the number of distinct, i.e. binary coded, output voltage levels that it can produce over the allowed range of input analog values.
Multiplexed ADCs
One of the main reasons customers choose to use multiplexed ADCs is that they can be configured to have different references, gain values, and dynamic ranges. This configurability extends into the signal conditioning on the analog front end, allowing for more accurate matching between channels.
The performance that can be achieved with this setup varies widely based on the quality of the ADCs and of the input/output circuitry. To achieve a high-resolution ADC, the converters must be capable of handling large steps in the input voltage. This requires a wide bandwidth, fast-settling amplifier to settle the step amplitude and reduce the crosstalk error between channels.
Another important consideration is that the input multiplexer and any amplifiers must be capable of settling quickly to their new values before sampling can occur. This will Analog to Digital Converter limit the error that can occur between channels due to aliasing or crosstalk, as well as ensure that the digital filter accurately reflects the new input value.
Delta-Sigma ADCs
Delta-sigma ADCs are renowned for their outstanding resolution, particularly for high-resolution applications such as sensor data acquisition and industrial process control. Their excellent resolution ensures that even tiny signal variations are accurately captured, while their noise shaping technique spreads quantization noise over a wider frequency spectrum to improve the overall Signal-to-Noise Ratio (SNR).
The key to Delta-Sigma’s performance lies in its intelligent design, which features an integrator, 1-bit Digital-to-Analog Converter (1-bit DAC), and comparator. Through a looping feedback mechanism, it orchestrates a distinct modulation of the quantization noise distribution, curbing out-of-band quantization error while emphasizing higher frequencies.
A low pass digital filter (typically with decimation) then mitigates the oversampled output of the Delta-Sigma Modulator, transforming it into a coarsely-quantized representation suitable for signal transmission and storage. This final filter also serves to curtail the output data rate by reducing its data rate through decimation, which in turn engenders augmented signal resolution within the limited bandwidth of relevance. The overall system’s speed limits are inversely proportional to its oversampling ratio, so this trade-off is crucial for choosing the right OSR to suit a given application.
Delta-Delta ADCs with SAR
ADCs must be able to convert all the signals that arrive at their inputs. However, the frequency of these signals can exceed the sampling rate (fs) of the ADC. In these cases, the ADC must be able to sample the signals at a higher frequency (fs+2) and then digitally represent the result. This process is known as decimation filtering.
In order to perform well, the ADC should also have a high signal-to-noise ratio and an appropriate dynamic range. This is because the ADC must be able to detect and reproduce all the signal nuances.
A common choice for a ADC architecture is the successive-approximation register (SAR). This topology offers medium resolution and power efficiency and is very easy to design. However, it is limited by internal comparator noise and DAC linearity. Therefore, it’s difficult to achieve a high resolution beyond 12 bits without trimming or calibration. This limit is also a constraint in applications that require both wide bandwidth and high resolution. This is where a delta-sigma ADC can provide an excellent solution.
Succeeding Approximation ADCs
While analogue signals can have a continuous range of voltage values, digital circuits only work with binary signals, which consist of only two distinct states, a logic 1 or a logic 0. This makes it necessary to convert these analogue signals into a form that can be processed by the digital circuits. This is where Analog to Digital Converters come into play.
A successive approximation ADC utilizes digital logic to converge on a value closest to the original analog signal using a comparator and DAC. These types of ADCs can provide high accuracy and high speed, making them a good choice for applications that require precise measurements or fast conversion speeds.
During conversion, the comparator circuit outputs a binary code to the DAC, which is proportional to the current bit being tested. This code is then compared with the sampled input voltage by the SAR logic. If the DAC output is greater than the sampled input voltage, the current result bit is set, otherwise it is cleared. This process is repeated for each bit, until all of the bits in the result are approximated.
Dithering
The analog input signal to an ADC is filtered, then quantized to represent discrete values. This adds error, known as quantization noise. It is highly correlated with the signal and therefore can be detected at very low power levels. The noise can also modulate the signal causing harmonic distortion. The best way to reduce the error is to add dither, a noise signal generated pseudo-randomly. The ideal dither is minimal in power, does not introduce noise modulation and eliminates the harmonic distortion due to quantization.
Dither can also increase the ADC resolution for repetitive inputs by averaging over many repetitions. Depending on the frequency of the dither and how Electromechanical component manufacturers many repetitions are averaged, the ADC can achieve a resolution proportional to the number of possible dither values and/or dither shapes.
Dither might seem counterintuitive at first, as it trades distortion noise for other low noise, but it’s needed when reducing bit depth, such as going from 32-bit float to 16-bit CD. It’s also needed when exporting to compressed audio formats such as MP3 and AAC.