Save power, improve performance, and program for a wide range of products.


SiliconIntervention applies an architectural approach to solving the real-world problem of maximizing efficiency in Class-D drivers over a wide range of power levels. This architectural approach is the basis for products with a self-determining ability to maximize efficiency over a wide range of operations.

FRACTAL-D based silicon products can leverage the advances already established using SiliconIntervention’s half-buck techniques and extend those gains automatically over a wide operating spectrum.

FRACTAL-D based products utilize conventional CMOS processes, do not call for non-volatile memory, and do not demand CPU/DSP processing or unique control signals/pins.

FRACTAL-D is smart enough to figure out how to apply maximum efficiency to the task at hand—in real time.

FRACTAL-D IP is another example of a growing IP product portfolio based on the “New AnalogTM”: a groundbreaking approach to sensor domain signal processing using next generation advanced analog and mixed signal design techniques.

Audio Fractal-D: high efficiencies for audio applications.

A large increase in driver efficiency with the advantage that it:

  • Needs no regulator or associated components.
  • Removes the need to change the driving digital signal gain
  • Removes the need for any envelope prediction
  • Offers optional feedback to create high quality output
  • Achieves more than the efficiency gain that a 100% efficient external buck regulator would provide.

Save power, save space, save complexity in audio products.

Based on advanced sigma-delta methods that perform a precise PCM to PWM conversion, the FRACTAL-D core is a scriptable HDL design achieving good performance prior to feedback. Digital feedback with included ADC meets the most demanding applications.

An agile solution for high efficiency over a range of applications.

  • Has <10 µV A-weighted noise
  • Has improved power supply rejection
  • Has digital feedback so PSRR is good
  • Utilizes adaptive FET drivers, saving power


Typical Class-D amplifiers are generally tuned to give the highest efficiency when outputting large power levels.  For example, a typical amplifier may have a prime tuning point at or around 1 W, primarily driven by the delivery of a low distortion output into a 4-ohm speaker load at a typical Li-Po battery supply (3.3 to 4.2 V).  Efficiencies of 80 to 90% at 1 W are achievable with standard ‘H-bridge’ Class-D amplifiers. 

We have seen that these amplifiers have a rather low efficiency at lower power levels (typically 30-40% at 10 mW).  A new type of bridge configuration included in FRACTAL-D improves the efficiency at these lower power levels to ≈60% at 10 mW while maintaining the ≈90% efficiency at 1 W. 

BridgeSupply5 V or 8 V22 nm with 5 V devices -or- 55 nm with HV devices
Switching Rate350-400kHz
THD> 90 db < 0.004%

4/8/32 Ω 15-30 µH Analog Closed Loop or Digital Open Loop
Noise Level Vrms (A-wt)< 3[1] or < 83 µV A-wt 1 V RMS 5 V or lower, 8 µV A-wt 5 V RMS, 8 V
Noise Level dBR (A-wt)< -112Using SI advanced analog control loop -114 dB @ 1 Vrms FS
Efficiency at 1% Max Power62%Exceeds typical industry standards of 45%
Figure 1: Efficiency Comparison

Figure 1 shows the benefit of FRACTAL-D over some typical 1 W Class-D amplifiers on the market today. Note the >60% efficiency at 1% power compared to currently available solutions at <40%.


FRACTAL-D is available as IP together with expert services for mapping into CMOS process nodes of choice.