Pulse-density modulation, or PDM, is a form of modulation used to represent an analog signal with digital data. In a PDM signal, specific amplitude values are not encoded into pulses of different size as they would be in pulse-code modulation (PCM). Instead, it is the relative density of the pulses that corresponds to the analog signal's amplitude. The output of a 1-bit DAC is the same as the PDM encoding of the signal. Pulse-width modulation (PWM) is the special case of PDM where all the pulses corresponding to one sample are contiguous in the digital signal.
In a pulse-density modulation bitstream a 1 corresponds to a pulse of positive polarity (+A) and a 0 corresponds to a pulse of negative polarity (-A). Mathematically, this can be represented as:
- where x[n] is the bipolar bitstream (either -A or +A) and a[n] is the corresponding binary bitstream (either 0 or 1).
A run consisting of all 1s would correspond to the maximum (positive) amplitude value, all 0s would correspond to the minimum (negative) amplitude value, and alternating 1s and 0s would correspond to a zero amplitude value. The continuous amplitude waveform is recovered by low-pass filtering the bipolar PDM bitstream.
Two periods of a higher frequency sine wave would appear as:
In pulse-density modulation, a high density of 1s occurs at the peaks of the sine wave, while a low density of 1s occurs at the troughs of the sine wave.
A PDM bitstream is encoded from an analog signal through the process of delta-sigma modulation. This process uses a one bit quantizer that produces either a 1 or 0 depending on the amplitude of the analog signal. A 1 or 0 corresponds to a signal that is all the way up or all the way down, respectively. Because in the real world, analog signals are rarely all the way in one direction, there is a quantization error, the difference between the 1 or 0 and the actual amplitude it represents. This error is fed back negatively in the ΔΣ process loop. In this way, every error successively influences every other quantization measurement and its error. This has the effect of averaging out the quantization error.
The process of decoding a PDM signal into an analog one is simple: one only has to pass the PDM signal through a low-pass filter. This works because the function of a low-pass filter is essentially to average the signal. The average amplitude of pulses is measured by the density of those pulses over time, thus a low pass filter is the only step required in the decoding process.
Relationship to biology
Notably, one of the ways animal nervous systems represent sensory and other information is through rate coding whereby the magnitude of the signal is related to the rate of firing of the sensory neuron. In direct analogy, each neural event – called an action potential – represents one bit (pulse), with the rate of firing of the neuron representing the pulse density.
A digital model of pulse-density modulation can be obtained from a digital model of the delta-sigma modulator. Consider a signal in the discrete time domain as the input to a first-order delta-sigma modulator, with the output. In the discrete frequency domain, the delta-sigma modulator's operation is represented by
Rearranging terms, we obtain
Here, is the frequency-domain quantization error of the delta-sigma modulator. The factor represents a high-pass filter, so it is clear that contributes less to the output at low frequencies, and more at high frequencies. This demonstrates the noise shaping effect of the delta-sigma modulator: the quantization noise is "pushed" out of the low frequencies up into the high-frequency range.
There are two additional constraints to consider: first, at each step the output sample is chosen so as to minimize the "running" quantization error . Second, is represented as a single bit, meaning it can take on only two values. We choose for convenience, allowing us to write
This, finally, gives a formula for the output sample in terms of the input sample . The quantization error of each sample is fed back into the input for the following sample.
The following pseudo-code implements this algorithm to convert a pulse-code modulation signal into a PDM signal:
// Encode samples into pulse-density modulation // using a first-order sigma-delta modulator function pdm(real[0..s] x) var int[0..s] y var real[-1..s] qe qe[-1] := 0 // initial running error is zero for n from 0 to s if x[n] >= qe[n-1] y[n] := 1 else y[n] := -1 qe[n] := y[n] - x[n] + qe[n-1] return y, qe // return output and running error