The challenges of distributed modulation
Making the control network invisible
A TECHNOLOGICAL CHALLENGE WITH TWO PARTS
There are essentially two technical challenges to that:
Ultra-low-latency data transfers between devices
Distributed modulators must receive their parameters (e.g. duty cycle, phase, etc.) with no further delay compared to the centralized approach.
Perfect synchronization of all time-related actions
The distributed modulators must provide the same time resolution and accuracy as modulators belonging to a unique hardware (e.g. DSP or FPGA).
PWM synchronization accuracy
The total synchronization mismatch between PWM signals is an extremely important parameter, notably as it constraints the minimum dead-time between complementary – high and low – PWM signals.
In the B-Box RCP, special attention was paid to minimize any unbalance also at the hardware level, up to the electrical and optical fiber outputs.
|PWM specification||Same device||Between devices|
|Propagation delay asymetry, electrical PWM||± 3.8ns||± 5.8ns|
|Propagation delay asymetry, optical PWM||± 13ns||± 15ns|
Eventually, the minimum dead-time also requires to account for other sources of propagation unbalances, after the modulators themselves, such as within the gate drivers and other similar circuits. This topic is addressed by a dedicated article on the imperix knowledge base.
PWM synchronization accuracy for N=6 devices
Histogramms of the synchronization error between multiple devices hosting distributed modulators. The maximum error is bounded to ±2ns.