Talk:Constant k filter

GA Review

This review is transcluded from Talk:Constant k filter/GA1. The edit link for this section can be used to add comments to the review.

This article deserves GA status. It is well written, neutral, stable and well referenced with in-line citations. The author has done a great job trying to explain this very specific topic to non-specialists. A number of minor problems has been fixed during the review, as documented below. Materialscientist (talk) 00:17, 26 August 2009 (UTC)

Really? I just read the whole thing and still have no idea what "k" is. 206.205.33.194 (talk) 14:06, 28 August 2015 (UTC)

Extended content

• Suggest to rewrite the article into a traditional GA structure: lead is summary without refs, lead expanded in the body with refs.
• Again a formal GA comment, more references needed to support individual statements
• Book page needed for ref. 4
• Why L in "for the low-pass L section example" If this is because L and C can be exchanged then the "L section" needs explanation?
• File:Constant-k Prototype Lowpass Image Impedance.svg what is Ω? (if you decide to redraw it, change "/rad s^-1" to "(rad s^-1)" at the X-axis)

• This is not my drawing, my original, admittedly not so good File:Constant k 3.png, had "ohms" spelled out. Ω means the units are in ohms. But I can change the new drawing to spell this out also. SpinningSpark 18:37, 21 August 2009 (UTC)

• "Band-stop and multiple band filters are also possible" sounds incomplete.
• I would rewrite "For a constant k section in general the transmission parameters for a half-section are given by;" - there is an unclear mix of "section" and "half-section"

• Clarified

• (BTW, why semicolon and not colon ?)

• Beats me, I've never really appreciated the difference between colon and semi-colon. To my mind, a colon should have clauses each side that could stand as complete sentences. In this case a colon is ruled out and the choice is therefore semi-colon or comma.

• Why L in "for the low-pass L section example" If this is because L and C can be exchanged then the "L section" needs explanation?

• The L in "L-section" has nothing to do with the L's and C's of inductance and capacitance. It purely refers to the topology of the circuit which is made clear in the opening sentence of the derivation section - "The building block of constant k filters is the "L" network, called a half-section, composed of a series impedance Z, and a shunt admittance Y". It is also diagrammed in the filter sections template at the bottom (although another editor has chosen to roll these up under a hide button). Not sure what more can be done to clarify that point but I am open to suggestions.
• SpinningSpark 23:07, 20 August 2009 (UTC)

  Off course you're right, and I missed that, but I was confused too and so will be another reader. Something constructive should be done here to clarify. "The building block of constant k filters is the "L" network, called a half-section" needs more explanation without sparring words, because indeed, L can be understood as inductance, L-shape or L-number (L-section and L-network might be treated as different notions too) and few readers would distinguish. Is it possible to keep L only for inductance, ignoring possible rules and conventions? Materialscientist (talk) 23:20, 20 August 2009 (UTC)

• I would rather not start inventing terminology, L-network or L-section is very widespread terminology, just as much as T-section is. It might help if some words were added that Z can be any impedance whatsoever - which would make it clear it is not necessarily inductance. However I feel the real problem is that the filter sections diagram is not prominent enough or early enough in the article since "a picture is worth a thousand words". The trouble is, if it is unrolled and put in the middle of an article, it interrupts the flow of the article quite severely. A possible solution is to redo it as a sidebox like in this article. Another approach is in mm'-type filter#Background which explains the topology and image impedance terms all in one diagram. Your views? SpinningSpark 23:49, 20 August 2009 (UTC)

I hope you admit that the situation is difficult - too many people will confuse inductance for L-topology - and need a sacrifice. I suggest avoiding "L" in L-topology. Maybe adding a note that one may encounter L-topology in electronics literature, but it has nothing to do with presence or arrangement of inductances in the circuit. Practically this would mean replacing

"The building block of constant k filters is the "L" network, called a half-section" by, e.g., "The building block of constant k filters is a unit, which is called half-section and is shown in pictures to the right"

"Again, for the low-pass L section example," by "Again, for the low-pass half-section," "Several L half-sections" by "Several half-sections"

File:Image filter terms.svg or its more appropriate form is definitely worth including. You might wish to put 2-3 images in a "horizontal-line" arrangement across the page (e.g., using a table) instead of stacking them all vertically. Materialscientist (talk) 04:17, 21 August 2009 (UTC)

I don't want to be argumentative at this early stage, but I really have to take issue with that. I am fine with adding a few words of clarification, but it would be completely wrong to expunge the term from the body of the article and just leave a note. It is a principle on Wikipedia to use the most common term for something, and L-network or L-section is certainly universally used for this topology. This Google books search lists more than 400 text books using the term (which is rather a lot for this relatively obscure topic - "Harry Potter" gets only 1,183), including this one which uses L in both senses in the heading without the slightest hint that this might be ambiguous. The author does not even feel the term needs defining at all. Like the author of that book, it had not even occured to me that it was possible to misread this in this way. A books search on that term without the quotes yields 17,000 hits, still all on the topic of filters, at least for the first few pages I checked, and a general search of the web gets an enormous number of hits. There are plenty of other ways this topology could be described: it might be argued that Gamma-section would be more logical since it is invariable an upside-down L being described. But the Wikipedia way is to use the most common, not the most logical, term.

Another small issue here is that although half-section and L half-section are synonymous in the context of this article, this is only so when taking half-sections of a T or Pi-section. Taking half-sections of some other network will not, in general, result in L-sections. The term "L half section" therefore fully defines the object being discussed.

SpinningSpark 09:10, 21 August 2009 (UTC)

No problem, just describe term "L-section" in ample detail, explaining that it might be confused with inductance, but so are tradition in the field, perhaps not as a note, but in the body text. I think merely adding dashes (between L and "section" and other possible words) would already reduce the confusion. Materialscientist (talk) 09:32, 21 August 2009 (UTC)

• It is unclear what are the two impedances $Z_{iT}$ and $Z_{i\Pi }$

  Usage preceeds definition. I would also include a figure, such as File:Image filter terms.svg Materialscientist (talk) 08:39, 22 August 2009 (UTC)

• "Where $Z_{iT}$ faces $Z_{iT}$" sounds strange (how impedances can face each other?)

  Same issue, and same picture will cover this. Materialscientist (talk) 08:39, 22 August 2009 (UTC)

• That is now in the article along with a lengthy description of the terms - hopefully that resolves both those issues. SpinningSpark 11:48, 22 August 2009 (UTC)

• File:Lowpass_Filter_LC.svg what is v?

• I find it hard to believe that any reader would not find the meaning of v_in and v_out to be self-evidently voltage in and voltage out. A reader who does not have that basic level of understanding is unlikely to get much out of this article whether or not the symbol is explained - in any event they are doing nothing more than marking the input and output ports, which in any case are in the conventional diagram positions, and are not used for anything in the text. SpinningSpark 18:37, 21 August 2009 (UTC)

  Think again. Those symbols are not nice big capital V but small ν (frequency). One trick might be your literature doesn't uses ν for frequency (f), but many other fields do; another is WP text interpreter poorly reflects Greek "nu" making it same as "v". Materialscientist (talk) 08:39, 22 August 2009 (UTC)

• Lowercase is the usual convention for instantaneously varying signals. In combination with the arrow, the meaning really cannot be mistaken, but even if it is, it does not matter since the only purpose is to mark the input and output of the circuit. This would still be correctly marked no matter what the symbols are read as (although it would be quite a stretch to imagine that this circuit was capable of transforming frequency). Frankly, I would rather replace them with just text "in" and "out" rather than give the reader the distraction of an explanation of unused symbols. SpinningSpark 11:48, 22 August 2009 (UTC)

  Sure, just erase "v" from input and output. I could have done this myself, but can't edit svg; can convert to PNG and then edit though. Materialscientist (talk) 12:01, 22 August 2009 (UTC)

• File:Constant k 4.png What is α, β and neper? (again, if you decide to redraw, please enlarge all symbols)

• added to caption

  Please enlarge all symbols or give me ascii data and I'll redraw. Materialscientist (talk) 08:39, 22 August 2009 (UTC)

• The original was generated by an Excel spreadsheet. If you send me an e-mail I will send you the original spreadheet (or you can use the formulae in the article, they are very simple). I do not see any advantage in updating my version, these png graphs from Excel don't seem to be liked on Wikipedia and a lot of them have been redone by other editors (not sure why this one was missed) so amending it would probably be a wasted effort. SpinningSpark 11:48, 22 August 2009 (UTC)

• Could you explain what is the "size of the section" ? BTW, if a person like myself doesn't understand something upon careful reading, then probably clarification is needed in the text. I guess this means for figure 1 that L is distributed, and becomes zero, but C remains same, but I am really not sure. Most people (say I :), even if familiar with electronics or/mathematics, are not used to distributed electronic elements. Materialscientist (talk) 08:39, 22 August 2009 (UTC)

• added an example to that which should make it clear. SpinningSpark 12:40, 22 August 2009 (UTC)

• "Like impedance must always face like in these combinations" - I understood, but suggest clarifying this sentence. Materialscientist (talk) 08:39, 22 August 2009 (UTC)

• I'm hoping the terminology section now illuminates this idea. SpinningSpark 12:40, 22 August 2009 (UTC)

• I am going to rewrite your edits to the bit about Zobel coining the term. While "coined" might be an idiomatic use of the word, it is nevertheless widely understood in English and has a specific meaning. This is not the Simple English wiki, and idioms are allowed, even helpful in this case as it immediately makes clear that the term was invented by Zobel, and who invented the terms is always of encyclopedic interest. I also do not like the construction in Zobel started distinguishing wave filters into constant k and m-derived filters. This gives the impression that all wave filters, or at least all image filters, can be so divided - which is not true. SpinningSpark 12:04, 23 August 2009 (UTC)

• We need to expand (the lead probably) on what filters are covered in the article: analogue/digital, passive/active. Also, if possible, mention somewhere which electronic elements are part of constant k filters (I guess R,L,C only). Materialscientist (talk) 05:49, 24 August 2009 (UTC)
• "ideal filter frequency response" should be clarified in the introduction. I guess it means perfect square in the frequency domain, but I shouldn't guess. Materialscientist (talk) 05:49, 24 August 2009 (UTC)
• "However, they are rarely considered for a modern design" - asks for why that, maybe adding "more efficient" before "other methodologies", or mentioning something on impractical number of components could help here. Materialscientist (talk) 05:52, 24 August 2009 (UTC)
• "the principles behind them having been superseded by other methodologies" - please mention a few, with wikilinks. Materialscientist (talk) 05:52, 24 August 2009 (UTC)

• I have expanded the lede, mostly by providing wikilinking to explanatory articles which covers most of those points. Diagrams are added to the "cascading" section showing the improvement in response with increasing numbers of sections, which is the most appropriate place to do this. SpinningSpark 12:23, 24 August 2009 (UTC)

• It would be nice to have an example how fast, with increasing number of sections, the transfer function approaches "the ideal response" (a step). Materialscientist (talk) 05:49, 24 August 2009 (UTC)

  Good updates, but I would be happier to see 3 curves in one graph, perhaps 1,4 and more (say 10 ;-) sections - scales are linear, and it is not clear yet that one can get to the square easily. Again, feel free to email me ascii, it is a matter of a minute to plot them. I am not satisfied with the graphing in the last subsection. Materialscientist (talk) 01:11, 25 August 2009 (UTC)

• I have sent you the spreadsheet. Of course, it does not ever go completely square until n = ∞ SpinningSpark 09:03, 25 August 2009 (UTC)

Image Method

The phrase image method is not sufficiently developed. It's used in the first and last paragraphs. It is an important concept.

In its first use the word image is linked to image impedance. Image impedance is not an image method.

What caught my attention is the need to "bare a concept in mind" closely related to it. Quoting section 4

"It should be borne in mind that the characteristics of the filter predicted by the image method are only accurate if the section is terminated with its image impedance."

Be There Do That (talk) 20:02, 29 August 2009 (UTC)cpiral

There is also composite image filter, an article that did not exist at the time this one was written, which is more aimed at the practicalities of using the method, and could be linked instead. Image impedance might not be equal to the image method but the image impedance article certainly explains the foundations of the method. The particular point you quote is covered in the image impedance article but there is now also image filter end terminations which explores this problem in more detail. Again, it is an article that could be linked here but did not exist at the time of writing. SpinningSpark 20:27, 29 August 2009 (UTC)