Why are you taking entries in :Category:Filter theory out of alphabetical order? I cannot fathom what you are trying to achieve. I can kind of see the point of highlighting one or two key articles (but even then, it would be better to link them in the lede of the category page) taking out a large tranche of them is not helpful. Certainly, the one I just reverted, Mechanical filter is not a key article for filter theory. If you are working to some kind of formatting master plan, you really ought to get some consensus for the plan before making widespread changes. SpinningSpark 11:33, 1 August 2010 (UTC)
:I also have problems with a lot of the other recategorisation you have done to these articles, but I won't revert any more until I hear from you. Taking m-derived filter as an example: why have you removed it from :category:linear filters when it clearly is a linear filter; why remove it from :category:filter theory when it forms part of image parameter theory; why does an article which discusses electronic designs not belong in :category:electronic design; why remove it from :category:electronics terms when, obviously, it is a term? SpinningSpark 12:11, 1 August 2010 (UTC)
::Hello - thanks for your note. I'll answer your questions:
::1) Why are you taking entries in :Category:Filter theory out of alphabetical order? I can kind of see the point of highlighting one or two key articles
:::It seems to me there are around a dozen, some of which overlap severely: at least one major article seems a POV split duplication. I'd like to discuss these cases soon; anything unnecessary can be put back, I will not leave the town trashed, but at the moment it is hard to navigate the subject. I began over at :Category:Sound technology and found there was no category there for the various audio filters used in amplification - some of which are name products, some generic terms. Especially sound synthesis articles must be able to access info on filter transfer function classes (band forms), cut-off freq, impedance and famous topological applications like the Moog capacitor ladder.
:::So I created the Cat :Category:Tone, EQ and filter - you will see that Filter theory has been a member of that for months, which is illogical but I could find no obvious ways to link audio with filter electronics, because, although electronics filters are commonly classified by topology, linear function class, technology (eg analogue, digital) and application (eg audio domain), there is no such categorisation at present and so no way to link through to these relevant subjects or to review the articles available while avoiding wrongly categorising articles on, say, radio-freq antenna filters or digital bitstream filters under "Sound". Rather, many formal logical inconsistencies are visible, such as that Linear filter appeared as Filter theory but Nonlinear filter did not. Similarly, articles on elementary topology are not consistently classified. Internal links and refs are few. Please bear in mind these evident facts and the desirability of some work.
:::I'd like to class articles according to topology, application, technology and math function in order to give meaningful entry and overview to readers entering from those contiguous domains. These categories will, obviously, themselves be members of higher classes such as Filter design. Otherwise, there is no way for me to say "I have looked at everything on wiki to do with audio filters in sound technology" without reading absolutely bloody everything on all filters whatsoever - which I am prepared to do, but only so that nobody else has to.
::::I cannot follow this, despite rereading several times. The reason for non-alpha sort still escapes me. The creation of :Category:Tone, EQ and filter is fine, but the relationship to the sorting issue is not obvious. I am suspicious of the idea that all filter categories must be a tree with "filter design" at the top, this may not work out in every case. Categorisation is designed so that it does not have to be a tree structure. There is overlap with "history" and "applications" for instance. SpinningSpark 19:14, 1 August 2010 (UTC)
::Certainly, the one I just reverted, Mechanical filter is not a key article for filter theory.
:::I find it important in the short term to be able to see all filter technologies articles, though this should really be highlighted under "design". This means I can ensure all such technologies are linked/mentioned in the appropriate articles and can even consider a "filter technology" category. As a distinct technology, mechanical is important - and it is also of some historical significance, which is another important vector of entry to the subject. But the real place things seem to have come unstuck in the past is "is digital electronic?" To me, analogue, digital and electromechanical are subsets of electronic, which are subsets of "signal" filtering, which is a subset of filtering per se. You ask;
::::I am particularly concerned with the Mechanical filter article because it is being prepared for FAC. This is a sister article of Distributed element filter which had an enormously difficult FAC but eventually got on the front page. The last thing that is needed is an ongoing dispute about categories, but also, a categorisation scheme that is not intelligble to reviewers is not acceptable if it leads to questions that cannot be easily answered. I agree with you that mechanical is an important technology and there is a problem with the articles based on technology. The article Electronic filter was originally intended (according to some of its authors) to be the "discrete component" technology. However, it slowly accumalated lots of other stuff including filter theory transfer functions. I have started to address this with the creation of mechanical filter and distributed element filter but you are right to point to digital filter and, I think, there is also an issue with crystal and acoustic wave filters. To my mind, it was better to create infrastructure with badly covered areas first before attacking existing articles.
::::I do not consider the top level article to be filter design, but rather filter (signal processing) where the various ways of categorising filters are explained and lead the reader to the articles they may be interested in. SpinningSpark 19:14, 1 August 2010 (UTC)
::If you are working to some kind of formatting master plan, you really ought to get some consensus for the plan before making widespread changes.
:::Well, this is the outline of the plan, above, and, since you have obviously done lots of work and do not want it to turn out wrong, you can say how you think it should be and bring in anybody you think you ought to - I have already created "Category:Electronic filter topology" and so may as well populate it - it can link all over the place. And I need to create "Category: Electronic filter applications" so that I can make my little "Tone EQ and Filter" category a member of it and make the page Audio filter (see, I finally found it!) its main article. Finally, general linear function terms such as "high pass", "order", "cut-off freq" should perhaps be collected separately from those polynomial mathematical functions, such as Butterworth's, that already have, not their own category but at least their own sidebar. Perhaps that's the way to go with the more general articles I mentioned - the ones that most sound engineers will look up first - though subcats can still be used too. I'd appreciate your thoughts on that - subcats and templates to give ease of access to filter theory for someone who wants to understand the virtual parametric EQs and sweeping 24Db per Oct resonant bandpasses on their virtual instruments, guitars, preamps....***((SEE BELOW!) Meanwhile, I hope it will be unexeptionable to populate these two subcats, for the moment, of the "filter theory" cat while awaiting consensus on my audacious stunt.
::::You want to make "Category:Electronic filter topology" and "Category: Electronic filter applications" both members of "filter theory"? Maybe "topology" but "applications" does not really fit. Why not have a top level category of "signal processing filter" to fit with the top level article? As I said above, I think you ought to be opening a discussion on this at a Wikiproject (Electronics, or Telecommunications, or somewhere and advertise it on appropriate pages) as what you are doing is quite far reaching. A relational diagram of what you are proposing would really help. SpinningSpark 19:14, 1 August 2010 (UTC)
::Taking m-derived filter as an example: why have you removed it from :category:linear filters when it clearly is a linear filter;
:::It's a member of a subset of linear filters, and categorisation policy says avoid nesting - which really helps, when you're trying to find out what is actually there, but the same articles are everywhere. Plus, it's a medium-complexity linear filter topology and, now that category of article has its own category, it can be listed in any convenient way you like while making more general articles such as "image parameter theory" even easier to find in a higher-level category. In the same way the new category is already a subset of "filter design", etc. so that individual topologies do not need to be - this is policy but it makes sense, it helps, I support it. So, unless you want to go to "categories for deletion", I'd better populate those I started, and you tell me what you think and what you do not think, and I'll do the work. Regards Redheylin (talk) 15:11, 1 August 2010 (UTC)
::::Ok, I had not realised there was a nesting issue. Thanks for pointing it out. An appropriate edit summary would help avoid problems in the future. SpinningSpark 19:14, 1 August 2010 (UTC)
:::***I found your "Bandform" template, so I hope I can take it you agree this is a good idea and shall add the template to the bandform pages. I also created a category "wireless tuning and filtering" uner "applications", since there's also a load under telecommunications that did not link to general theory. Redheylin (talk) 17:17, 1 August 2010 (UTC)
::::"Wireless" is a bit obsolete. Simpler and more succint would be "radio filtering". SpinningSpark 19:14, 1 August 2010 (UTC)
:::::I found that they use the term on wiki to comprehend, radio, TV, microwave etc, whereas "radio" tends to be the braodcasting medium, so I went along with it. It's like "Tone, EQ and Filter" - it looks a bit comic-book from this side of the curtain, but it tells the person at the other end what to expect, and stops me getting grief that "TV is not radio"! I am creating categories that span two linguistic domains. Otherwise I'd have done what you'd have done. Redheylin (talk) 19:17, 1 August 2010 (UTC)
::::::Point taken of "TV", but "wireless" is still a horribly bad choice. "RF filters" is better. SpinningSpark 20:16, 1 August 2010 (UTC)
:::::"The reason for non-alpha sort still escapes me." - if it still bothers you in 7 days, I'll put everything where you say it should be, OK? I am just trying to understand the issue of technologies, where the divisions are.
:::::"Categorisation is designed so that it does not have to be a tree structure. There is overlap with "history" and "applications" for instance." Agreed - the subject in itself has roots both in experiment and in maths. Parts of the theoretical content branch off rapidly into the former, discussing coils and crystals, while others, like the concept of roll-off and filter order, go immediately to differential equations. Experiment was different in different domains. I am keen on history of science - did some work on developmental biologists - and would like to see a smooth transition from that direction too. I'm looking from bottom up, outside in and do not have any G.U.T. on the basis of which to impose a heirarchy. However, clearly, some aspects of filter theory, like the pure maths, have a certain general validity across technologies and applications, whereas others, like topology, only inhabit the subset of analogue electronics.
:::::Why not have a top level category of "signal processing filter" to fit with the top level article? I completely agree - as I said, the "theory" category was simply the only place available. It seems a no-brainer to me. Redheylin (talk) 20:00, 1 August 2010 (UTC)
::::::My concern is that you are proceeding at such a rate of knots that it going to be difficult to later unpick and I can see some disagreement on this. Why not seek a consensus? SpinningSpark 20:16, 1 August 2010 (UTC)
:::::::Because nobody has sorted out these problems for years, thousands of category edits have made me bold, nobody seems to care but you, you obviously know your stuff and I am thinking what you are thinking - so! Redheylin (talk) 20:20, 1 August 2010 (UTC)
::::::::Maybe others haven't noticed - or are too busy. If you are thinking what I'm thinking, why am I reverting so many of your edits? SpinningSpark 21:24, 1 August 2010 (UTC)
:::::::"I am particularly concerned with the Mechanical filter article because it is being prepared for FAC....The last thing that is needed is an ongoing dispute about categories" - Sorry to have caused you anxiety. "I think, there is also an issue with crystal and acoustic wave filters." - Yes - there's no classification by technology. But for that matter, should all analogue and electromechanical devices be classed linear? Take the case of a spring reverb in which an analogue signal is passed through an electromechanical device - this imposes a mechanical formant filtering upon the signal, which seems non-linear to me. Synthesiser filters are made to self-oscillate, creating nonlinearity. Actually I came back to this because I am seeking a circuit to filter low odd harmonics from a signal, which made me consider a mechanical device in inverted phase - can you point me anywhere more, erm, conventional? Redheylin (talk) 20:34, 1 August 2010 (UTC)
::::::::No, the spring reverb is linear. In what sense do you think it is non-linear? But you are right, not all analog is linear, all passive analog is though. The phase cancelling filter is called a transversal filter, this is usually digital, but an analog delay line works just fine. The idea is that the signal is tapped off the delay line at the point where the frequency to be rejected is in antiphase and then added back in to the output. Several different taps can be added to obtain the desired response. This is ideal if all you want to do is kill specific harmonics. Of course, analog delay lines are expensive and bulky which is why DSPs are so popular. SpinningSpark 21:24, 1 August 2010 (UTC)
:::::::::In what sense do you think it is non-linear? Because the formant is not a linear function of the input signal - you have modulation that cannot be expressed in a simple transfer function. But anyway - this delay line would have to adjust its delay time to the incoming frequency to achieve phase cancellation, no? Redheylin (talk) 21:30, 1 August 2010 (UTC)
:::::::::: It is not a rational function of the input signal (because of the transmission-line-like nature of the innards), so the transfer function cannot be expressed in polynomials, but it is still a mathematically linear function. It must be, because the spring reverb is entirely passive linear components, basically Hooke's law. You are right, that a changing input frequency will require changing the tap-offs of the delay if you want the harmonic relationship to remain constant. This requires detecting the value of the fundamental, which is still possible, but again a lot easier to do in DSPs than in analogue. You also need to consider what you mean by "third harmonic" in a complex signal. This is straightforward for a monotonic instrument but for harmonies needs some thought. A two part harmony singing an interval of a fifth apart, for instance, will have the higher voice second harmonic right on top of the lower voice third harmonic. If all you are trying to do is remove some soft-clipping (which is what odd harmonics suggests to me), it would probably be more productive to try improving the linearity of your source. SpinningSpark 19:44, 2 August 2010 (UTC)
I really appreciate your explanation - I had assumed that the input amp of the spring (which needs a big signal) was driving the imposition of the "foreign" harmonics but what you say makes sense - most likely my maths is not good enough to understand the linearity. As for the rest, I am working with a single, monophonic fundamental over about two octaves. I have considered introducing soft clipping into the negative side but I am afraid it will increase with frequency, whereas what I need is control over the 3,5,7,9 in that order of amplitude and let the upper partials pass or, at least, for the effect to roll off. So I am currently looking at harmonic synthesis, though I hear great things have been achieved with finely-tuned diode ladders. Redheylin (talk) 00:50, 3 August 2010 (UTC)