 Hello, and welcome to this introduction to microphones. In this video, I'm going to cover the details of a PDM microphone, including PDM signal features and the microphone interface. As defined by Wikipedia, Pulse Density Modulation, or PDM, is a form of modulation used to represent an analog signal with a binary signal. In a PDM string, specific amplitude values are not encoded into codewords of pulses of different weight, as they would be in Pulse Code Modulation. Rather, the relative density of the pulses corresponds to the analog signal's amplitude. Let's go in deep on the definition and have a look at how this standard is implemented in a PDM microphone. Let's focus on two keywords of the PDM definition. PDM is a binary signal. It consists of a high-frequency stream of one-bit data, which can be either one, representing a positive pulse, or zero, representing a pulse of negative polarity. The relative density of the pulses corresponds to the analog signal's amplitude. So a run consisting of all ones would correspond to the maximum positive amplitude value. All zeros would correspond to the minimum negative amplitude value, and alternating ones and zeros would correspond to a zero amplitude value. The sampling frequency of a PDM signal typically ranges from 512 kHz to 4 MHz. If we have a look at the frequency response of a typical PDM output, we can see that the sigma-delta modulator pushes the noise at the high frequencies outside the audible range. This kind of signal must be then filtered and decimated in order to remove the high-frequency noise and reduce the sampling frequency to a typical audio frequency. The output of this process is a digital signal in PCM format. In practical applications, a PDM microphone interface is implemented by means of three signals. A clock signal, which is generated by the host, is an input for the microphone. It provides the timing and defines the sampling frequency. A data signal, which is the actual output of the microphone, is synchronized with the clock provided, which means that the microphone's binary stream output will have the same frequency as the clock it receives as input. A third pin, called left-right, is available in MEMS microphones. It's an input for them, and it's used to choose on what edge of the clock the microphone will provide the data. If this pin is set to a high logical level, the microphone is in its right configuration. This means that it will provide data on the following edge of the clock, releasing the data line on the rising edge. If this pin is set to a low logical level, on the contrary, the microphone is in its left configuration. This means that it will provide data on the rising edge of the clock, releasing the line on the following edge. This left-right mechanism allows to use a single data line to transport the data coming from two PDM microphones, if they are configured with opposite left-right pin. This would allow to further reduce the signals to be routed and simplify the board layout. In the image, you can see the typical connection of two microphones sharing the same data line. Please note the opposite left-right configuration, the common clock input, and the common data line, controlled by each microphone on its clock edge. For further information, please refer to the following documents. You can also note 5027, interfacing PDM digital microphones using STM32, and user manual 2372, STM32 Cube PDM 2PCM software library for the STM32-F4-F7 H7 series. Modern PDM microphones can provide several modes of operation to find the right trade-off between power consumption and acoustic performances. Here, there is an extract of the data sheet of the MP23DP01HP multi-mode microphone, which highlights the concept of multi-mode. Depending on the clock provided, the microphones work in a low-power mode or in a performance mode. Generally speaking, the lower is the clock frequency provided, the lower is the power consumption. The acoustic performances are affected as well, and you can usually obtain better signal-to-noise ratio when the clock frequency is in its high ranges. One common use case for this feature is for always-on applications. The microphone and the whole system runs in low-power mode for the most of the time. Waiting for some acoustic events to happen. For example, waiting for the average acoustic level to exceed a certain threshold. At that point, the clock of the microphone can be increased to the performance mode to be able to acquire a better signal that will be further processed for trigger-word recognition. In the following slide, we will start analyzing the basic parameters concerning digital microphones. The first one we are going to approach is the acoustic overload point, or AOP. In simple terms, AOP is the loudest sound a microphone can acquire without too much distortion. It's measured in DBSPL, which is a measure of the physical pressure generated by the acoustic wave in the real world. To have an idea of what we are talking about, here are some examples of the sound pressure level generated by certain events. The hearing threshold is at 0 DBSPL, a normal conversation at 1 meter is roughly 50 DBSPL, a measure road at 10 meter is around 80-90 DBSPL, a jackhammer at 1 meter, about 100 DBSPL, a rock concert would be around 110-120 DBSPL, the threshold of pain would be around 130-140 DBSPL. The typical acoustic overload point for a MEMS microphone ranges from 120 to 135 DBSPL. On the opposite side, we have a measure of the residual noise of the microphone. This is the intrinsic noise generated internally by the microphone, and it's present even if the device is in a complete silent environment, such as a full isolated and unequivocal chamber. This noise, despite being internal, can be converted to an equivalent noise, as if it would be present in the physical world. We will measure this noise in DBSPL. It defines the level of the quietest sound in the real world we can acquire, which will be discernible from the microphone internal noise. The typical equivalent input noise for a digital MEMS microphone is around 130 DBSPL. Using the two parameters we just defined, we can now approach the dynamic range, which basically defines the listening range of the microphone, and it's computed as the difference between the AOP and the equivalent input noise. For now, we have covered the physical domain and how we can measure the sound pressure in the real world. We now introduce a different domain, the digital one, which represents the signal after the analog-to-digital conversion. In this domain, a different unit is used. The DBSPL is a decibel relative to full scale, and it measures the amplitude levels in the digital system. What is the full scale? You can think of the full scale as the maximum value that you can represent by a digital signal without saturation. This will be our reference value, and it will correspond to zero DBSPL. In this scenario, the amplitude of a signal can be represented as the ratio between the measured value and the full scale. To make it simple, you can think of this as a measure of how much more quiet is the signal, respect to the maximum value represented in this digital domain. Here are some examples of a digital signal and their DBSPL value. This is for minus 6 dBFS, minus 12 dBFS, minus 20 dBFS. Let's go a little bit farther and put everything together. We are showing the two domains we spoke about so far. On the left, we have the physical acoustic world, where the measure is made in DBSPL and represents the pressure of the acoustic wave in the real world. On the right, we have the digital world, where the measure is made in DBSPL and represents the amplitude of the microphone digital representation. The mapping between physical and digital world is specific for each microphone and depends basically on the acoustic overload point of the microphone. A microphone with an AOP of 120 dBSPL, for example, will output its maximum digital value when in the physical world the sound pressure is equal to 120 dBSPL. Another one with an AOP of 135 dBSPL will output its 0 dBFS for 135 dBSPL pressure. To better understand this concept, you can think that a specific microphone can be able to listen to louder sounds, respect to the other one. Hence its 0 dBFS will be mapped to a different value in DBSPL in the physical domain. How those parameters are measured? The standard reference signal used for the measurements is a 1 kHz tone with a sound pressure level of 94 dBSPL at the microphone. At first, the sensitivity is measured. Sensitivity is the level of the digital signal expressed in dBFS that of the microphone outputs for the standard reference signal. Then the internal noise is measured inside an isolated anechoic chamber. We will have value in dBFS, which corresponds to an equivalent input noise in DBSPL. The signal to noise ratio or SNR is the difference between the sensitivity and the residual noise. The acoustic overload point is the maximum acoustic signal which the microphone can capture with acceptable distortion. It's measured in dBSPL and as we saw it generates a digital signal around 0 dBFS. The dynamic range basically defines the listening range of the microphone and it's computed as the difference between the AOP and the equivalent input noise.