More about PDM

Created on 2013-02-01 00:15:00

PDM is Pulse Density Modulation, a system for representing a sampled signal as a stream of single bits using delta sigma conversion (sometimes called sigma delta conversion). PDM is the technology used in many oversampling ADCs and DACs, and is the basis of the Sony/Philips commercial digital format and disc trade-named DSD and SACD, respectively.

A one-bit stream is unacceptably noisy, but very high sampling rates and noise shaping techniques are used to greatly reduce the noise in the audio spectrum. This noise energy is moved from the audio baseband into the area of the spectrum above 20 kHz, where it is inaudible. It is also impossible to properly dither a 1-bit stream, so there is always a small amount of distortion due to quantization error.

Why PDM?

Properly done, PDM can digitally represent high quality audio, and is inexpensive and easy to implement. For these reasons, a PDM stream is now commonly used as the data output of a MEMS (Micro Electro-Mechanical System) microphone. MEMS microphones can be made to be very small and are inexpensive to implement on silicon chips, and are found in many small devices such as cell phones or Bluetooth devices.

Analog to PDM to Analog

An analog signal can be directly sampled at a high sampling rate (several megahertz or more) and converted to a PDM stream. The PDM signal can be converted back to analog audio by passing it through a low-pass filter.


A signal that is coded as PCM (pulse code modulation, the coding widely used in digital audio) can be converted to PDM by sampling it at a higher rate (interpolating) and reducing the word length to one bit. The ratio of the interpolating PDM bit rate to the PCM sample rate is called the oversampling ratio.


A signal that is coded as PDM can be converted to PCM by sampling it at a lower rate (decimating) and increasing the word length. As mentioned above, the ratio of the PDM bit rate to the decimated PCM sample rate is called the oversampling ratio.

PDM modulator

PDM modulators have a series of integrators, or accumulating nodes. The “order” of a modulator refers to the number of its integrators.

Higher-order modulators provide improved noise and distortion performance over much of the amplitude range, but suffer from increased instability at amplitudes approaching maximum level. To allow full amplitude levels without driving a PDM modulator into instability, limiting techniques are typically employed. In APx, an Overload indicator (shown in the Status Bar) is activated when the modulator exceeds the defined overload point. For the APx 4th-order modulator, the overload point is about –7.5 dBFS. For the 5th-order modulator, the overload point is about –9 dBFS. Limiting is used above this point, enabling both the 4th and 5th order modulators to operate to maximum level without instability. However, noise and distortion rise for signals above the overload point. In practical applications, such performance is deemed acceptable, since voice signals at maximum levels are typically brief transients. In APx, modulator distortion at maximum level is about 1% THD+N.

MEMS microphones

A MEMS microphone system consists of the MEMS sensor, an analog preamplifier, and a PDM modulator, often implemented on a small silicon chip. DC power and a clock signal at the sampling rate are typically provided by downstream devices.

Manufacturers of such devices typically use 4th-order sigma delta modulators at a clock frequency of 3.072 MHz, which in the receiver is typically decimated by a factor of 64 to a baseband sampling rate of 48 kHz.