Opima Project - Aims and scope

Aims and scope

State of the art

In recent times, switched-mode actuation has become extremely popular, due to growing consciousness in energy efficiency and to a boom of portable electronic apparatuses, where costs and weights are dominated by batteries, so reduced dissipation is a must. However, a parallel advance of modulations has not been witnessed. In most cases, current applications still rely on traditional pulse modulation schemes such as PWM, at most trying to correct their inherent defects by smart feeback techniques.

This approach is not always fully successful. For instance switching audio amplifiers are generally largely inferior to their analogue counterpart, switching regulators tend to have significant problem in complying to some electromagnetic compatibility (EMC) regulations, and so on.

Unfortunately, these issues are traditionally takled by the addition of new hardware: more parts are introduced into the system to compensate for the non-idealities. For instance: feedback mechanisms are introduced to compensate for the loss in signal to noise ratio (SNR); shields and filters are introduced to compensate for the inferior EMC; etc. Very little effort has so far been put in seeing whether these problems can be tackled in the modulations themselves. Consequently, today very modern hardware is often paired with information representation schemes born many years ago, often from intuition and the need to solve contingent applicative issues rather than from a deep understanding of implications. The result is paradoxical: in some extreme case the modulations perform better on yesterday hardware that in the most modern one. For instance this is the case with PWM, that operates decently on purely analog designs, but that may deliver very poor SNRs when adopted in new digital designs where the PWM switching events are compelled to align with some external clock.

Recently, some signs of change are apparent, though. Most of them are related to the extreme success of Sigma-Delta (ΣΔ) data converters, that have raised the attention on PDM, being that ΣΔ-codes are in the end a rather sophisticated case of it. ΣΔ codes are not truly ideal for actuation, since they tend to exhibit a too high rate of switching, however they have encountered a good deal of success, particularly in the audio field. Here, there are now the first attempts at using this modulation not only for power amplification, but also as a means for general information management and storage, as proved by the Direct Stream Digital (DSD) format and the recently standardized Super Audio Compact Disc (SACD) based on it. Recent attempts at lowering the ΣΔ switching rates and enhancing the perceived audio quality are also reported, in the form of patented solutions such as DDX and Tri-path. Also, some techniques pursuing an EMC enhancement of more traditional modulations such as PWM are currently being investigated. In no current solution, however, there is any attempt at at devising impulsive modulations that are optimized by simultaneously taking into account a full mixture of constraints and merit factors.

To summarize, it can be concluded the state of the art in the field of impulsive modulations has been rather still up to very recent times and that there are now strong signs of an imminent revitalization of the research area. One of the current project goals is to anticipate such a wave of change.

Expected advance of knowledge

As already mentioned, pulse modulations and switching power control techniques can actually be exploited in many different tasks. In each of them, deepening the study of modulations and introducing more options can have strong, positive repercussions. In this project a few representative areas where significant advances are expected are identified as follows:

Sinthesis of harmonic and periodic waveforms with high spectral purity. Here, one aims at synthesizing test signals and excitation signals for a variety of apparatuses. The project outcomes shall help developing fully digital approaches. Rather that playing back through a data converter a pre-digitalized and stored sequence of signal samples, the proposed idea is to algorithmically synthesise a pulse code such that its continuous time filtering can directly result in the target waveform. In this task, the pulse modulations need to be optimized principally for the final spectral purity.
Synthesis of pulse codes for power actuation. Here, one aims at synthesising pulse codes specifically well suited for driving the switches of power converters. In this case, one has possibly to deal with many channels at once (such as in drivers for multiphase electrical machines). Also, operation with more than two levels in the pulse codes can be an appealing option. The project outcomes shall help in developing next-generation codes capable of enhancing power efficiency and EMC.

Representation of audio signals for data storage and switching audio amplification. Here, SNR and perceived quality are of primary importance. Furthermore, optimization is complicated by the absence of time constraints, being the target signals fully aperiodic. Techniques such as signal segmentation and windowing shall help dealing with this case. In this task the project outcomes shall help develop coders that can be dropped in existing applications, enhancing the acoustic experience.

Apart from the rather applicative points above, the project also aims at an

Extension of the theoretical and fundamental knowledge in the area of pulse modulations. Here, the interdisciplinary group of the research team shall take advantage at best of its resources in the mathematical area. Expected outcomes include the ability to relate pulse modulators to traditional optimization problems (for instance hints exist that ΣΔ modulators can be formally related to sub-optimal solvers of binary quadratic programming). Also fundamental knowledge about the size and geometry of the mathematical spaces where pulse codes are generated and about information-theoretic bounds shall definitively be sought.

Scientific program, milestones, results

The project aims at combining some of the different research activities already individually practised by its research units and to do so creating the opportunity for long lasting interdisciplinary cooperation.

The start-up phase shall consequently involve integration tasks and shall develop along the following lines:

► background search in the area of traditional pulse modulations and characterization of their performance in specific application, to create sets of benchmark results; ► design of simulation platforms to test the above, co-simulating signal coders, actuators and plants; ► analysis of mathematical features of traditional pulse modulations; ► application of optimization methods to the synthesis of periodic waveforms by bipolar and tripolar pulse codes in the unconstrained case, with characterization by real world instrumentation; ► evaluation of the behaviour of performance-optimized codes in terms of EMC on real-world power apparatuses; ► extension of traditional modulation to host additional degrees of freedom to be used as optimization parameters and characterization; ► mathematical and information-theory characterization of the optimization spaces in the above cases; ► extension of the above techniques to aperiodic signals, such as audio;► development of software based coders exploiting the above techniques.

The program completion phase shall involve:

► extension and formalization of the analysis of mathematical features of optimization spaces, with particular regard to the application of real world constraints; ► application of optimization methods to the synthesis of periodic waveforms by bipolar and tripolar pulse codes in representative constrained case, with characterization by real world instrumentation; ► design of EMC optimized codes for power actuation; ► development of full hardware solution and drop-in coders for existing applications in possible partnership with industrial entities.

More...

Impact

[ Back ]