Talk:Lattice phase equaliser

This topic is probably only readable for electrical engineers

This is a topic I myself have a fondness for, but it isn't a topic that anyone other than an electrical engineer with a background in filter synthesis will be able to understand. Its been written by someone who has had only an academic encounter with lattice filters, and that's enough to write an overview. The discussion about compensating for the parasitic resistance of inductors made it clear that the authors and editors have never made even a single prototype lattice section. Please remove the discussion of any realization considerations, its uninformed, and no citable references exist. The references etc. are at least one generation after the actual original references, but if people are crediting Zobel for his work on the math for the synthesis of improved lattice phase compensators he's as good as anyone, why not.

The references are crediting authors a generation after the Bell System Technical Journal articles discussing compensating long distance and undersea telephony and telegraphy cables cataloged all of the topologies and transfer functions, and those may have been a generation too late to credit as inventors. The post WWI generation articles are concerned with methods for synthesizing particular filter response functions, i.e. refining existing compensation network designs. The some of the earlier generation articles delve into some of the practical problems they had realizing them, but they refer to the capacitors as electrolytic tanks (NB: Before 1914 a tank was a municipal water reservoir at the top of a hill), the ideal inverting transformer for the Tee realization was then called a gyrator, any reactors should be understood to be inductors, and many other familiar concepts are also described using other names. The vocabulary changed sufficiently to frustrate word searches for article titles and abstracts using the current names. A bound set of the BSTJ can be flipped through to find relevant articles because figures that depict lattice sections are otherwise rare and they put relevant articles in the same month for many years in a row. It would be nice if AI image recognition works well enough to find them otherwise it will take around 12 hours, or at least that's how long it took in 1980. Some of the early BSTJ articles refer to still earlier references that I wasn't able to locate in any of the Boston area engineering libraries.

All pass lattice sections haven't been a topic for which any public references have been created since WWI by anyone who was actually making them or using them. Proprietary company references are known to have existed in at least 4 companies, but by definition can't be cited for a Wikipedia article. There would not be any citable references for a competent author to give, so with no chance of improving the practicum, I suggest abandoning any further attempts to suggest it.

FWIW: Here's a taste of what building a first prototype will reveal. The phase shift section goes from 3°-90°-177° over the span of 2 decades, i.e. from Wc/10 to 10Wc. 0° to 180° (rounded left of the decimal point) is 4 decades. There aren't any components that have a constant inductance or capacitance over 2 decades around the frequency where |Zx| = Zo for any useful Zo. Furthermore, components with a tight enough tolerance to realized a phase shift section as described don't exist. Anything realized without tuning variable components will display an anomalous resonant suck-out and a group delay anomaly at roughly w=Wc. Any realizable design has to incorporate at least the first order parasitic effects of its components, and a good design will have to go much further.

I'm not aware of any orderable (cataloged) product made using lattice section in the last 50 years. A few have been made using Tee sections. PolychromePlatypus (talk) 22:27, 27 December 2024 (UTC)

Applications section is imaginative but incorrect.

The applications section may be more imaginative than factual. The original application was to phase correcting the amplitude compensation applied to long cables used for baseband communications. Those applications ceased a century ago. Applications not connected to the original application exist, but are not suitable for a Wikipedia article.

First, with respect to the original applications, the article's commentary about microphone lines and monaural audio in general is correct for studio audio but is not true more generally. Historically, the reason for creating phase equalizers was because after amplitude equalizing long distance terrestrial baseband telephone signals, there was so much phase distortion that the voice was unintelligible. This topic disappears from the literature after the 1920's because long distance telephony switched to using single sideband carrier modulation to frequency domain multiplex multiple conversations on a single coaxial cable or point-to-point radio link. The circular waveguide never did overcome its propagation anomalies to make low loss guided waves, radio through a pipe, a reality until graded index fiber made it practical at optical frequencies - but the point is that long distance baseband audio became as extinct as the passenger pigeon.

Second, I think a somewhat suggestive name, there phase matched audio 'radio loop' led to an erroneous conclusion that broadcasters used phase compensated landlines for baseband audio to deliver stereo from remote studios to transmitters. The 'radio loop' was a service of the digital telephone network. Digitizing the audio and delivering it over digital channels of a synchronous digital telephone network built around 1 megabit/sec T1 links. That network used fixed multiplexed bandwidth allocation, synchronous sampling, sigma-delta A/D converters. The radio loop used multiple standard POTS (plain old telephone system) digital channels. The phase matching for a stereo link was the sampling clock for the two channels, ordinarily a sample clock was synchronized to minimum delay with respect to its interleaved bitstream. Aligned samples helps FM stereo by minimizing the L-R stereo difference which is transmitted on an amplitude modulated carrier offset from the nominal FM center frequency. The stereo difference signal doesn't enjoy the 'FM advantage' suppressing noise at reduced signal levels at the expense of being unable to demodulate the channel at all below a low threshold SNR. [If there was any other version in 1961 when the FCC approved FM stereo it would have been multiple modulated carriers in a single coax cable. The SSB modulation used by telcos was inherently phase coherent and multiple channels on a single coax have identical group delay. No lattice section phase equalizers would have been needed]

Third, the part about landlines being inherently balanced signaling is only generally true during the 19th century and early 20th century for the telegraph and telephone system. It remains true for 'local loop' 'last mile' 600 ohm characteristic impedance copper pairs, but those weren't compensated enough to use lattice sections in the days of electromagnetic switchgear and the later electronic Telco switch uses digital filtering to compensate the lines. PolychromePlatypus (talk) 17:02, 30 December 2024 (UTC)