Inside a 7th-Order SDM

A seventh-order sigma-delta modulator sounds intimidating. In practice, the difficult part is not the number seven. The difficult part is controlling where quantization noise goes without creating a system that becomes unstable under real musical signals.

Why Higher Order?

The basic purpose of an SDM is to push quantization noise away from the audible band.

Increasing modulator order allows a steeper noise-shaping characteristic, but every increase in aggressiveness comes with additional constraints.

The Important Variables

Among the variables that matter are:

• loop topology;
• coefficient set;
• maximum stable input;
• target DSD rate;
• input scaling;
• and the desired out-of-band noise profile.

CRFB and CIFF

CRFB and CIFF are two useful loop structures, but their behavior is not identical.

At very high DSD rates, the enormous oversampling ratio provides substantial room for shaping noise outside the audible band. At DSD64, the problem is considerably less forgiving.

This means a configuration that looks excellent at DSD512 may not be the configuration we prefer at DSD64.

Stability Is Part of the Design

A modulator that produces spectacular simulated noise performance but becomes unstable near realistic signal levels is not useful.

Input scaling therefore becomes part of the system rather than an inconvenient afterthought.

The best noise floor is irrelevant if the modulator cannot safely reproduce the signal presented to it.

DSD512 Changes the Problem

At DSD512, the available oversampling ratio allows extremely low in-band noise.

Our measurements can reach levels where the practical question is no longer simply “how low is the noise?” but whether additional optimization produces any meaningful benefit elsewhere in the playback chain.

Conclusion

A seventh-order SDM is not one algorithm with one universal setting.

The useful design is a family of operating points chosen according to output rate, stability requirements and the behavior of the complete playback system.