Jump to content

Talk:Bridged T delay equaliser

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

Superconductor planar implementation

[edit]

The section presently describing a 2.8GHz filter network using distributed transmission line elements may be a topology inspired by the low frequency Bridged-T network, but it doesn't incorporate its defining feature, the reversing transformer. Its simply a second order distributed element tee network.

A Bridged-T network converts a balanced 2nd order lattice section into an unbalanced network by using a reversing transformer to achieve a negative component. The advantage of the 2nd order network is that the parasitic inductance of the large capacitor can be absorbed by the small inductor and the parasitic capacitance of the large inductor can be absorbed in the small capacitor. (absorbed meaning the component's value is chosen as the ideal circuit value less the parasitic value) The bandwidth is limited only by the transformer's parasitic uncoupled inductance and interwinding capacitance. Since it isn't required to provide safety insulation the transformer can be wound bifilar, confining the fields advantageously, effectively a twin lead transmission line in a region of high magnetic permeability. 3 or more decades of bandwidth are readily achieved. The odd-mode impedance is controlled by the per-unit-length inductance and capacitance of the bifilar winding and the even mode impedance by the permeability of the transformer's magnetic core material. The minimum operating frequency is determined by the even mode inductance, and the maximum frequency by the bifilar transmission line's phase length.

The microwave realization is using the time delay of a long transmission line to achieve a similar negative reactance but the bandwidth of the technique is inherently limited to less than an octave. The topology doesn't use magnetic field coupling, or electric field coupling, to achieve impedance inversion. The use of high temperature superconducting material is largely irrelevant, it just reduces the losses in the long high impedance line section and the interdigitated (meshed fingers) series capacitance. This is also a solution in search of a problem. Seymour Cohn described additions to the basic branch line coupler that yield an equivalent performance over multiple octaves using common materials in the 1970's.

In summary for those unfamiliar with the technical details, apart from being a resonant delay with a partially similar schematic, the microwave circuit is unrelated to a bridged-t. The described innovation is also a solution to a problem that already had a better solution decades earlier that didn't need high temperature superconductors.PolychromePlatypus 17:25, 4 August 2018 (UTC)

I'll not take a position on whether such an experimental design belongs in this article. Perhaps it doesn't, and I wouldn't challenge its removal. However, it is certainly a bridge T topology. Whether or not there is coupling between the two series branches of L1 is irrelevant; that doesn't stop it being bridged-T. The referenced source says it is an LC allpass. The originial paper is "High-temperature superconductive lumped-element microwave allpass sections" from Electronic Letters, which also offers it as an allpass design. Thus, reliable sources have classified it as allpass. The semantics of whether a bandwidth of less than an octave can count as allpass is not something for Wikipedia to resolve. We go with what the sources say. SpinningSpark 17:55, 4 August 2018 (UTC)[reply]
I have now read the original paper. Their plotted results show an insertion loss of less than 1 dB in the range 0–6 GHz at 77K using yttrium barium copper oxide. The paper says the insertion loss is <0.7 dB 0–4 GHz and <0.3 dB after subtracting connector losses. By any measure, that is way beyond an octave passband. Perhaps you meant the band over which maximum delay is achieved? That is indeed just over an octave according to their results. SpinningSpark 09:36, 5 August 2018 (UTC)[reply]

Error In The Design Section

[edit]

I think the formula for the resonance frequency is incorrect. It should be wo = 1/sqrt(4*L*C) = 1/sqrt(L'*C')

Secondly, the entire article is extremely sketchy to the extent that it has taken me many hours of hard work to understand it. It's entirely unclear how to construct L' and C' - and whether one needs to allow for the -M from the transformer model in the leg of the tee.

Thirdly, there is no mention that in order for the circuit to function as an all-pass filter it must be terminated in Ro - but this is NOT the same as Zo as described in the linked to article on Zobel networks. Indeed, the description there of the bridged T coil network is also skimpy, but that's a separate matter.

Finally, I would like to see explicit formulae for Z', namely C'=4*L/Ro^2 and L'=C*Ro^2. I have run comprehensive simulations with these substitutions and it is indeed an all-pass filter in this case. Otherwise it looks like a mess, because the minimum phase zero is not cancelled by one of the poles.

I would like to edit the article accordingly but I don't know how to typeset the equations to do so. Could somebody please help?Frank van Kann 03:35, 28 April 2023 (UTC) Frank van Kann 00:33, 28 April 2023 (UTC) — Preceding unsigned comment added by Frank van kann (talkcontribs) I have now made the above suggested changes.Frank van Kann 03:35, 28 April 2023 (UTC) — Preceding unsigned comment added by Frank van kann (talkcontribs) 01:46, 28 April 2023 (UTC)[reply]