Excitation Conditions: On the Importance of Couplers

In this section, we investigate suitable excitation structures for proper plastic waveguide propagation. Following the discussion in 2.2.1, relevant figures of merit can be listed below:

 Low insertion loss,

 Fields matching,

 Impedance matching,

 Phase velocity matching,

 Mode selectivity,

 Mechanical robustness,

 Coupling reproducibility,

 Low cost.

It has been shown in 2.1.2.2 that satisfying all these conditions is very difficult so that optimizing couplers for a given plastic waveguide may still be considered as an open research topic. Based on the literature, we can mainly identify four different approaches summarized in Table 2-5 with identified advantages and drawbacks.

Although on-chip or in-package coupling using near-field antenna radiation is appealing for the implementation of a compact and cost-effective solution, the poor control of the excited mode(s) and associated polarization, in addition to bandwidth limitation caused by the resonant behavior of the antenna, has led the community to consider the use of intermediate waveguide sections, particularly metallic waveguides. Introducing an intermediate section is not benign and this approach may be not be favored in a long-term perspective because the direct coupling at chip or package level has a higher potential…once reported issues will be solved.

Advantages  Drawbacks 

 Need a custom transition from chip to metallic waveguide

 Higher Cost

 Bulky

Table 2-5: Summary of excitation schemes reported in the literature along with their respective advantages and drawbacks.

For short-term concerns, and in order to ease characterization, the third approach is then utilized in this work and is presented in Figure 2-39. Once connected to a suitable WR to PCB transition, this solution could enable a conceptual connector. It is actually the purpose of Chapter 3 to present the necessary WR to PCB transition. For the moment, only the WR to plastic waveguide coupler is described here. To do so, a WR-12 standard is considered to provide a suitable reference plane in the 60 – 90 GHz range. In practice, it was even possible to extend the measurement range to 55 – 95 GHz in agreement with EM simulations of this standard. To match the largest dielectric waveguide to the metallic waveguide standard, this latter is enlarged with a standard taper mode converter or pyramidal horn antenna so that the tapered dielectric waveguide can be inserted inside. The waveguide can be mechanically maintained in the metallic aperture using low-density plastic foams introducing negligible perturbation compared to pure air. Such a simple transition has been reported in many references, further justifying the relevance of the proposed approach. In [Hofmann, 2003], characterization frequencies up to 600 GHz are demonstrated and positioning enhancement is proposed by adding a dielectric (foam) centering disc in the metallic aperture. [Zhou, 2013]

and [Zhou, 2017] carried out a dielectric taper shape analysis and showed the superiority of a pyramidal taper placed at the center of the metallic waveguide.

Figure 2-39: Description of the coupler to excite a dielectric rod waveguide (grey) from a WR-12 metallic waveguide using a conically tapered section placed at the center of a large pyramidal horn antenna (not represented for visibility). An isometric view is presented in (a), a top view in (b) and a

lateral view in (c). Orange lines indicate metallic conditions imposed by the horn while green lines indicate “radiation boundaries” around the “open” dielectric waveguide.

Figure 2-40: Simulated propagation constants at WR-12 port (left) and dielectric rod port (right) of the coupler.

(a)

(b)

(c)

Propagation constant β(rad/m)

1000 1200 1400 1600 1800 2000 2200 2400 2600

55 65 75 85 95

400 600 800 1000 1200 1400 1600 1800 2000

55 65 75 85 95

Propagation constant β(rad/m)

Frequency (GHz) Frequency (GHz)

TE₁₀

TM₀₁ HE₁₁ HE₁₁

WR-12 PORT Dielectric rod PORT

Figure 2-41: Field description of possible modes at input and output ports of the coupler. At input, (a) shows that only the fundamental 𝑇𝐸₁₀^𝑉 mode can propagate. At output, (b) and (c) correspond to the degenerated 𝐻𝐸₁₁^𝐻 and 𝐻𝐸₁₁^𝑉 modes respectively. The following higher order mode (d) is the centrally

symmetric TM01 mode.

In spite of an existing background related to the metallic / dielectric waveguides transition, it appears that no comprehensive analysis is available describing both mode excitation selectivity and cross-polarization isolation in the case of symmetrical waveguides.

Consequently, a simplified simulation model of the proposed coupling scheme is proposed above to investigate these aspects. Due to computing limitations, it was not possible to simulate the whole structure. Instead, the horn aperture has been divided by four in order to reduce computational complexity. To maintain the flaring angle, the horn length (in the propagation direction) is also reduced accordingly. The simulation model is presented in Figure 2-39. Moreover, a canonical dielectric rod waveguide is considered with a 90 GHz maximum single-mode frequency. A modal analysis is presented in Figure 2-40 confirming single-mode operation at both WR-12 and dielectric rod ports. In this later, the fundamental mode is degenerated due to rotational isotropy with the presence of two orthogonally polarized modes having similar propagation constants. Note that no symmetry simplifications can be applied here because cross-polarization isolation is of primary interest. For that particular reason, no assumption is made on the nature of excited modes in the simulation.

These possible modes are detailed in Figure 2-41.

(a)

(b)

(c) (d)

Figure 2-42: E-field complex magnitude (phase-independent) investigation in the coupler presented in Figure 2-39. Fields are plotted at 90 GHz assuming an input excitation from the WR-12 TE10 mode:

(a) isometric view, (b) top view, (c) lateral view. Note that all E-field scales are logarithmic.

Figure 2-43: S-parameters simulation of the coupler. (a) Reflection parameters and (b) transmission parameters. Mode indices convention is the same as in Figure 2-41.

Simulated fields and S-parameters are presented in Figure 2-42 and Figure 2-43 respectively. Excellent return losses are obtained for the TE10 and HE₁₁ modes over the entire considered bandwidth. On the contrary, the HE₁₁ mode exhibits strong reflections (Sbb ≈ - 1 dB). Regarding insertion losses, an obvious transfer between TE10 and HE₁₁ modes is observed with very low loss (< 1 dB) while transmission from TE10 to HE₁₁ is negligible (cross-polarization > 50 dB). Finally, excitation of the higher order TM01 mode also appear insignificant. Even though propagation constants are closer (Figure 2-40), the noticeable fields discrepancies between TE10 and TM01 modes (Figure 2-41) induce very low coupling.

To sum up, the proposed coupler achieves an excellent transition between rectangular metallic waveguides and dielectric waveguides. Insertion loss lower than 1 dB is simulated with negligible return loss and very good mode selectivity (assuming a single-mode design).

(a)

However, as discussed earlier, this performance is obtained at the expense of a higher volume footprint, higher cost and moderate positioning reliability.