Showing papers by "Blaise Ravelo published in 2010"

A New Technique of Interconnect Effects Equalization by Using Negative Group Delay Active Circuits

Abstract: A novel and innovative technique of interconnect effect equalization in electronic systems was developed through theoretical studies and experimentally validated. It relies on the use of active circuits able to simultaneously generate gain and negative group delay in baseband over broad bandwidth. The properties of these circuits were used to propose a new compensation approach consisting in an equalization of both the positive group delay and attenuation induced by interconnects by an equivalent negative group delay and gain. The theory on commonly used circuits to model interconnect effects were briefly recalled. Then, a circuit composed of a first-order interconnect model (i.e. an RC-circuit) cascaded with an NGD cell was theoretically studied in order to determine the conditions required to compensate for both the degraded propagation delay and the attenuation and to express the synthesis relations to be used in the determination of the values of the NGD cell components. This NGD cell simply consists in a FET fedback by an RL series network. Then, for this first modelling of interconnect line, a proof-of-concept circuit implemented in hybrid planar technology was fabricated to demonstrate the efficiency of this technique. The experimental results in frequency- and time-domains were in very good agreement with simulations and validated the compensation technique in the case of an input signal with a 25 Mbit/s data rate. Indeed, the reductions of the rise time and the 50% propagation delay were 71 and 86%, respectively. In many VLSI systems and particularly in long wires and/or for high data rates or clocks, the inductive spurious effects can no longer be neglected. So, a more elaborated system composed of an RLC interconnect model compensated with NGD cells was also studied analytically in order to check for the validity and efficacy of the equalization technique and to determine the synthesis relations to be used in further applications. To validate the approach, a first series of simulations was run with an RLC lumped model for an input signal at 1 Gbits/s-rate; then , the model used in the second set of simulations was an RLC distributed line for an input signal at a rate of 200 Mbit/s. These simulations under realistic conditions confirmed the compensation approach with reduction of the propagation delay of the same order as previously. Moreover, as observed with the RC-model, the front and trailing edges both showed great enhancements indicative of a good recovery of the signal integrity. Finally, to be able to compensate for interconnect effects in VLSI systems, the proposed circuits must be compatible with a VLSI integration process. This requirement drove us to propose improvements of the proposed topology in order to cope with inductance integration and manufacturing prerequisites. So, a topology with no inductance, but with the same behaviour and performances as previously was proposed. A theoretical study provided evidence of its ability to exhibit a negative group delay in baseband together with gain. Then, time-domain simulations of a two-stage NGD device excited by a 1 Gbits/s-rate input signal were run to validate the expected compensation approach and check for the signal recovery. The implementation of this equalization technique in the case of a VLSI integration process is expected to allow compensation for spurious effects such as delay and attenuation introduced by long inter-chip interconnects in SiP and SoC equipments or by long wires and buses. A preliminary step would be the design and implementation of such a circuit in MMIC technology and especially by using distributed elements. At this stage, even if experimentally the NGD cells were not particularly sensitive to noise contribution, it would be worth comparing this approach and repeater insertion under rough conditions, i.e. long wires with a significant attenuation, in order to gain key information on their respective behaviour under conditions of significant noise. As identified in ITRS roadmap, the power consumption is now one of the major constraints in chip design and has been identified as one of the top three overall challenges over the last 5 years. Faced to these constraints, further investigations are needed to accurately evaluate the consumption of the presented NGD active circuits.

E-Field Extraction from h-Near-Field in Time-Domain by Using Pws Method

Abstract: A novel technique of the electric- or E-fleld extraction from the magnetic- or H-near-fleld in time-domain is reported. This technique is based on the use of the Maxwell-Ampere relation associated to the plane wave spectrum (PWS) transform. It is useful for the E-near-fleld computations and measurements which are practically complicated in time-domain in particular, for the EMC applications. The considered EM-fleld radiation is generated by a set of electric dipoles excited by an ultra-short duration current having frequency bandwidth of about 10-GHz. The presented EM- fleld calculation technique is carried out by taking into account the evanescent wave efiects. In the flrst step, the time-dependent H-fleld data mapped in 2-D plan placed at the height z0 above the radiating devices are transposed in frequency-dependent data through the fast Fourier transform. In order to respect the near-fleld approach, the arbitrary distance z0 between the EM-fleld mapping plan and radiating source plan should be below one-sixth of the excitation signal minimal wavelength. In the second step, one applies the PWS transform to the obtained frequency-data. Then, through the Maxwell-Ampere relation, one can extract the E-fleld from the calculated PWS of the H-fleld. In the last step, the inverse fast Fourier transform of the obtained E- fleld gives the expected time-dependent results. The relevance of the proposed technique was conflrmed by considering a set of flve dipole sources placed arbitrarily in the horizontal plan equated by z = 0 and excited by a pulse current having amplitude of 50mA and half-width of about 0.6ns. As expected, by using the Hx, Hy and Hz 2-D data calculated with Matlab in the rectangular plan placed at z0 = 3mm and z0 = 5mm above the radiating source, it was demonstrated that

Frequency-independent active phase shifters for UWB applications

Abstract: This paper describes an active phase shifter, which exhibits a frequency-independent phase shift over the ultra-wideband (UWB) frequency range. The design of this pure phase shifter relies on the use of a negative group delay (NGD) circuit. The topology of the active NGD cell is based on a Field Effect Transistor in cascade with a series resonant network. A 90°±5° phase shifter with a single NGD stage was designed, at first, and implemented in hybrid microstrip technology; its measurements showed a 75% relative bandwidth and validated the proposed approach. Then, a compact pure phase shifter with a multi-stage NGD cell was designed. It was aimed at meeting specifications in the UWB frequency range, i.e. 3.1–10.6 GHz (110% relative band). Over this band, the corresponding simulation results showed a constant transmission phase of −45°±3°, an insertion gain from 2 to 5.5 dB and insertion losses better than 9 dB.

Characterization of the Regular Polygonal Waveguide for the RF EM Shielding Application

Abstract: This article presents a theoretical characterization of the regular polygonal waveguide (RPW) having n-sides. Based on the symmetrical circular symmetry of the RPW and the circular waveguide (CW), the analogy between the electromagnetic (EM) behaviors of these to waveguide (WG) is established. After a brief recall about the state of the art concerning the WG engineering and its application, we introduce a basic theory of the WG presenting a regular polygonal cross-section with n-sides. By considering, the fundamental mode TE11, we develop the main mathematical formulas summarizing the difierent characteristics (cut-ofi frequency, fc, propagating constant, k11 and S-parameters) appropriated to any RPW in function of its physical parameters (number of sides, n, diameter, D and height, h). In order to verify the validation of the developed analytical expressions, comparisons between the HFSS simulated and theoretical dispersion diagrams of regular pentagonal (n = 5), hexagonal (n = 6), heptagonal (n = 5) metallic (copper) WG with for example, 50mm outer diameter are presented. So, it was demonstrated that very good correlation between the theoretical predictions (fc(n), k11(n)) is found with a relative error less than 1%. As application of the present study in terms of EM wave shielding, simulation of metallic wall with hexagonal aperture is also performed. Finally, discussion about the future work is drawn in conclusion.

Study and Application of Microwave Active Circuits with Negative Group Delay

Abstract: A simple topology of an NGD active circuit consisting of a FET terminated by a shunt RLC-resonant network and dedicated to the microwave signals was proposed and extensively studied. To our knowledge, in this chapter, the first experimental time-domain demonstration of a circuit able to exhibit simultaneously gain and NGD in microwave domain is proposed. By injecting in the NGD circuit a sufficiently smoothed input short-pulse modulating a sine carrier, one gets an output having an envelop peak in advance compared with the input one. Of course, this phenomenon does not contradict the causality principle. It is also worth emphasizing that the tested circuit respects all required criteria of classical active microwave devices such as gain, matching and stability. As predicted in theory (Ravelo et al., 2007a, 2007b, 2007c and 2008a), for a prototype implemented in planar technology, we have measured in time-domain a pulse peak advance of about -2 ns or 24% of the 1/e-input pulse half-width without attenuation. It is also interesting to note that through this experimentation, the input noise contribution did not destroy the occurrence of time-domain advance induced by the NGD active circuit. Moreover, in this chapter, thanks to the S21-magnitude form, the understudied NGD circuit is able to exhibit a pulse compression phenomenon with a possibility of amplification. Then, it should be worth using the presented NGD active topology to compensate for dispersion effects and especially to reduce the intersymbol interference in certain telecommunication channels. As a potential application of this NGD circuit, a new principle of frequency independent phase shifter is proposed. By cascading a classical transmission line with this NGD circuit, a constant phase value is obtained. The efficiency of this principle was demonstrated by measurement. Indeed, a constant phase value of 90°±5° was obtained within a 76% relative frequency band centred at about 1.5 GHz. The impacts of the PS parameter variations and sensitivity analysis are stated. The main benefits of this NGD active PS concerns its compactness and also the facility to generate very low group delay, and the broad band characteristics. Besides, a two-stage NGD PS was also designed; the simulation results showed a bandwidth enhancement of the constant phase up to 125%. Some fields of applications such as the design of broadband active balun for RF front end architectures are discussed. As ongoing research, design of reconfigurable devices dedicated to telecommunication applications is envisaged. Future investigations will be devoted to the design of NGD devices able to operate at higher frequencies through the use of distributed elements. In this optic, the implementation of MMIC devices based on distributed elements is envisaged.