Digital systems, purely analog systems, mixed-signal systems, and many things in between often need a specialty analog interface to collect and condition signals. Analog interfaces are not like digital interfaces; they are often highly customized groups of components because there is not always a simple ASIC solution available. Analog interface design can involve everything from capturing and filtering a signal to feeding a signal into an ADC. No matter what you need to do in analog front-end design, success depends on component selection and successful signal modeling across your front-end.
To do this successfully, we often leverage simulation tools and manufacturer-supplied component models to fully simulate an analog interface. The process for generating and cleaning up an analog signal will be outlined in this article. As most systems operate with a digital processor or embedded application, there will probably be a digital section in the mix. This means an ADC will be needed, so we will also include some advice on using ADCs in the analog front-end.
How to Build Your Analog Front-End
All analog front-ends are big analog circuits that capture and manipulate a signal within some frequency range. Typically at low frequencies, we use the term analog to describe a system, but at progressively higher frequencies we might call it an RF system. There is no specific frequency that divides an analog front-end from an RF front-end, and they operate under similar principles.
That being said, an RF front-end is typically used for broadcasting or receiving a signal via coax or over-the-air. An analog front-end covers just about everything else, such as interfacing with sensor measurements from DC all the way up to MHz frequencies. The main components you'll need in an analog front-end can be either active or passive, and some can be specialized ASICs that perform certain functions. A non-comprehensive list is given in the table below.
The list of parts shown in the table enable typical approaches to designing a linear analog system. Some of the functions that could be implemented include:
Filtering, both active and passive
Signal generation or conversion
Modulation or demodulation
Use of one signal to drive or modulate another signal
Amplification, either with FETs or op-amps
Comparisons between two or more signals
Oscillator circuit excitation
Phase offset or amplitude measurements
Direct or indirect measurement of frequency
The goal in designing the analog front-end is eventually to get the signal to a point where it is sampled by a receiver or it is able to interact with a load. Most often, in mixed signal systems, the analog front-end is designed for the former usage.
Analog Front-End Architecture
The analog front-end architecture used in a PCB should first be planned out. This requires comparing the signal being input to the system with the signal you want to receive at your sampler or load. There are many questions that need to be answered by the designer before creating the signal flow for the analog front-end
What is the input signal level?
What is the receiver’s dynamic range and noise floor?
Is there a noise source that could interfere with the signal?
What is the frequency range for the input signal?
Is the signal a sinusoid or is it a different waveform?
All of the above questions will determine the combination of waveform generation, filtering, amplification, and modulation/demodulation stages required in the analog front-end. Once reaching the receiver, this can put constraints on the sampler if the signal is being converted to a digital signal with an ADC.
Well we can't fully describe every system architecture in this guide, we can describe some of the major functions that would need to be designed and evaluated through simulation. Depending on the frequency range, generic parts in your simulation library might be useful for understanding signal flow in simulation. However, some specialty parts have functions that are not captured in generic parts libraries, and so they need to be included in a simulation with manufacturer part models.
Now, let's jump into some of the major functions involved in constructing an analog front-end.
Filtering and Amplification
Filtering and amplification are often implemented together in an analog front-end and they can comprise most of the system design. The point of filtering and amplification is to adjust the waveform level at the desired frequency range such that it falls within the receiver or load dynamic range.
Select the filtering and amplification stages based on how the two sets of devices create a passband in your desired frequency range while also applying gain. This can be seen quite clearly by looking at a pair of transfer function curves for an amplifier and filter. Assuming both are operating in their linear ranges, the transfer function for the two stages combined is just the product of the two individual transfer functions. This fact can be used with information from a datasheet to apply gain within the desired passband.
Analog front-end design requires taking multiple transfer functions in products to determine the effects of each stage on a signal, such as a cascading of a filter (left) and an amplifier (right).
To evaluate filtering and amplification, it's appropriate to use the two stages together in the same simulation. The reason for this is to examine how the two stages combine to create a transfer function with gain. Since we are gathering a transfer function, we would want to use a frequency sweep and create a Bode plot. From the Bode plot, you can examine total gain within your passband and the phase shift experienced by the input signal.
Modulation or Demodulation
Modulation and demodulation are commonly used in RF signals, but this is not the only instance where these are used. In some Precision measurement applications, the modulating signal is extracted from a received waveform in order to interpret some information. To extract the modulating signal requires some type of filter circuits:
Once the modulating signal is extracted, it then likely needs to go through its own amplification, filtering, and acquisition/read-out stages in the analog front-end. Typically some DSP will be performed in the embedded application to take the acquired signal and pull some meaningful information from it.
DC Level Measurement
DC level measurements may seem quite simple: just put the signal through a filter and run it into ADC. Based on the numeric output from the adc, you can interpret a DC voltage level.
In cases where you are simply trying to measure a moderate DC voltage with high SNR and no overlaid AC signal or strong noise source, this works fine. When there is a specific noise source, you need to measure a DC offset on an AC signal, or the SNR value is low, some other approach maybe needed. A diagram showing each stage in the front-end for a potentially low-level DC measurement is shown below.
One of the keys here is selection of the ADC and resolution. As was discussed in a different article, the ADC May produce higher accuracy if it has lower resolution rather than higher resolution. The reason for this is quite simple: if the ADC has lower resolution, then the voltage distance between quantization levels is larger, and the measurement can withstand more noise overlaid on the DC signal level. This is important when we consider the amplification stage in the above diagram, because the amplifier will increase both the noise and the signal, ideally by the same magnitude.
ADC quantization can allow noise to fall between two quantization levels if the resolution is low enough.
Phase, Amplitude, and Frequency Measurements
Direct measurements of a waveform can be performed in three ways:
Digitally through DSP operations
Through direct measurement and comparison in a lookup table
By comparison with a reference signal or reference oscillator
In terms of the application development required, the third option is most convenient. However, some designers who are not experienced with PCB layout may find the use of a reference oscillator in a complex analog system rather difficult. This is due to issues with crosstalk, noise, and frequency/phase response.
Extraction of one of these measurements via DSP operations also requires cleaning up the signal, such as through amplification, filtering, and averaging. The goal is to clean up the noise so that it does not create excessive errors when DSP is performed on the acquired signal.
Finally, the lookup table approach is a simple way to implement frequency measurements with a microcontroller and filter. With an active or passive filter, the transfer function through the filter circuit will be known, therefore a reading of the amplitude will have a definite relation to the signal’s frequency based on its attenuation or gain in the passband curve. By comparing the measured values (e.g., amplitude) with value in a lookup table, the signal’s frequency would be the closest frequency value in the lookup table.
Tools to Simulate Your Analog Front-End
Analog front-ends need to be thoroughly evaluated in simulation before building an experimenting with circuits, and definitely before spinning a PCB prototype. Board spins are expensive and take a lot of time, while breadboarding leads to potential human errors that are also time consuming. If you have access to a large model library, you can comprehensively simulate your analog front-end before spinning prototypes.
Whenever you want to build and analyze your analog circuits, make sure you simulate your designs with the complete set of tools in PSpice from Cadence. PSpice users can access a powerful SPICE simulator as well as specialty design capabilities like model creation, graphing and analysis tools, and much more.