Exactly. I'm old enough that my first active crossover build used 12AX7s. The last analog crossover I built used OPA2134s. The two crossovers I've used most recent were "black boxes" built using DSP chips of which I don't even know the part number, "programmed" from a laptop, and they have adjustable delay and all sorts of other useful things that used to be difficult or impossible to implement. I doubt I'll ever build another analog one . . . so a book I would have drooled over 20 years ago has less appeal today (not enough for me to even contemplate buying it).
I had the same criticism of "Small Signal Audio Design" (which I did buy, to some modest benefit) . . . the only analog circuits which remain interesting or useful these days are low noise microphone pre-amps . . . and the closest the book gets to them (in any serious way) is the discussion of . . . moving coil phono preamps. Like anyone cares about those anymore. At least the concepts translate . . .
As for "DIY" . . . if we let that limit us to obsolete and historically unsatisfactory technologies what's the point?
As pointed out elsewhere :
you hardly read cutting edge ideas in books first - its those boards like here where "the development" happens nowadays - or at least - starts out...
Michael
Perhaps I'm failing to get my point across here.
As I said before, at least two-thirds of the book as currently envisaged is applicable to either analogue or DSP. When you know what filters and/or EQ you want, it is then and only then you must decide to be analogue with 5532s or to start cutting code for a DSP.
Suppose I added a chapter on DSP-ing to match the chapter on using opamps, well, what do you think it should include? Is it appropriate to start getting into the details of implementation on different processors? If so, how many different processors should be included? Clearly that could get out of hand in terms of length. A tutorial on DSP from scratch is obviously out of the question, so where do you start? How to code a basic bi-quad? The best ways to implement noise-shaping in the calculations? (a fascinating topic in its own right, I might add) I just don't see I have the space to do all this.
Finally, as someone wisely remarked up-thread, DSPs don't come in DIL packages. I have done no DSP work for a while now, so perhaps people could enlighten me on just how practical it is today to build a DSP crossover on the kitchen table. I don't want to write a book that is for professionals only.
As I said before, at least two-thirds of the book as currently envisaged is applicable to either analogue or DSP. When you know what filters and/or EQ you want, it is then and only then you must decide to be analogue with 5532s or to start cutting code for a DSP.
Suppose I added a chapter on DSP-ing to match the chapter on using opamps, well, what do you think it should include? Is it appropriate to start getting into the details of implementation on different processors? If so, how many different processors should be included? Clearly that could get out of hand in terms of length. A tutorial on DSP from scratch is obviously out of the question, so where do you start? How to code a basic bi-quad? The best ways to implement noise-shaping in the calculations? (a fascinating topic in its own right, I might add) I just don't see I have the space to do all this.
Finally, as someone wisely remarked up-thread, DSPs don't come in DIL packages. I have done no DSP work for a while now, so perhaps people could enlighten me on just how practical it is today to build a DSP crossover on the kitchen table. I don't want to write a book that is for professionals only.
I also believe that DSP is the future. On the other hand I have come across DSP S/W that made it obvious that the programmer had no idea how analog active filters work, what they really can do and how to translate that into the digital world. 2 R and a C could do more than a 250 bucks DSP. So for such guys the book should become a mandatory lecture. Hence it would cause any "analogi" less grief when using new technology.
P.S. I would probably buy the book.
P.S. I would probably buy the book.
Last edited:
As a German native, you may be interested in an article from the January 2011 issue of Stereoplay. It describes how we got to the current state where virtually all large-scale PA XOs are DSP-based, to correctly steer the many driver arrays that are commonly used. It's a nice little summary of history, of which I find especially the beginning very interesting (although developments then were much slower than they are now).
--
Greetz,
MatchASM
Depends on what you mean by "build". I certainly don't fab DSP chips in my kitchen. I didn't fab op-amps or roll my own capacitors, either (although I did make a vacuum tube once, back in High School). But I *can* take something like a miniDSP board (and I expect that there will be more like it), plug in a USB cable, and program it using my laptop . . . on the kitchen table. Could do that with a DCX2496 too (except with a RS232 cable, and different software).
Does that constitute "building" a crossover any less than populating a breadboard with some (pre-built) op-amps and "passive" components?
It does get easier and easier for a simple consumer to submit custom SMD multilayer PCB designs to a manufacturer. As long as the PCB footprint is ok, it is surprisingly easy to solder on DSPs and DACs. Nonetheless, if you get the wiring wrong (quite likely for a DIY designer) it takes ages to fix.
Then the DIY DSP programmer would have to learn about the differences between IIR and FIR filters (number of multipliers required, achievable filter steepness, differences in phase behavior), frequency warping, finite wordlength effects and their solutions, limit cycles, probably some interpolation/decimation/polyphase techniques, synthesis tools and their limitations, and any number of advanced topics.
It's realistic to be able to achieve something state-of-the-art as a DIY DSP designer, but it takes a lot of time and effort. And I think a comprehensive book would be approx. 2-4 times the size of your ~500-page book.
It's a lot of skills to be mastered, which will probably pose a barrier for a lot of DIY audio enthusiasts who only plan to build a speaker once in their life, but still would like to understand what goes on inside every part they touch with their soldering iron.
--
Greetz,
MatchASM
Note of encouragement to go with the current toc.
I would buy it, it interests me to see the filter topologies and their general application when taken the speakers (and room?) into consideration. I would especially be interested in the tips and tricks on selection of cutoff frequencies, how to measure the filters and speakers as a system and so on.
of course more complex filters with more advanced compensations can be done in the digital world. A lot more can go wrong as well and a coder who isn't decent at it wont have much chance at figuring it out unless they get some nice application on top of the processing unit which abstracts the implementation to a level where even novices can interact with it. I like the trend with self assembling user interfaces that i've begun to see in linux audio stackable drives for audio functions but that is so far off the mainstream and really in its infancy... Having said that, i for sure would like to see a next book get into the digital real of filtering, both true digital filters (digital in, digital out) as well as digital filters implemented on analog in/out signals. I think that would be a very interesting topic for a different book.
here is what i would love to see in your current book mr self.
DIY'ers are tinkerers. I love to tinker with stuff. And filters lends themselves to a lot of tinkering. What are the good practical advices you think one should think about when doing implementation of analog filters to ensure that they stay "tinkerable". just pragmatic stuff like, consider using caps of this decades because this allows you to alter resisters in these locations to change the frequencies, use this style circuit for a bit of EQ'ing because you only need to modify this value of component to change the behavior. These are the lessons that are gained from hands on expertise which i think are invaluable in print. True "this makes things easier and here is why" type of experiences is so much more interesting than the basic science and formulas to the general masses.
then consider doing a followup book which references this book on DSP audio processing followed by a book on pc based audio processing. the pc based audio one will be quickly obsolete, or rather will require frequent revisions but i think it would be useful to you and to the community.
i've also written a few books on implementations (my books was in the area of computer and IT security and disaster recovery) and the consistent feedback i got is that the pragmatic "look out for this" or "consider this in case you want to change something later on" is what the readers really enjoyed.
good luck and please let us know when its going to print!
I would buy it, it interests me to see the filter topologies and their general application when taken the speakers (and room?) into consideration. I would especially be interested in the tips and tricks on selection of cutoff frequencies, how to measure the filters and speakers as a system and so on.
of course more complex filters with more advanced compensations can be done in the digital world. A lot more can go wrong as well and a coder who isn't decent at it wont have much chance at figuring it out unless they get some nice application on top of the processing unit which abstracts the implementation to a level where even novices can interact with it. I like the trend with self assembling user interfaces that i've begun to see in linux audio stackable drives for audio functions but that is so far off the mainstream and really in its infancy... Having said that, i for sure would like to see a next book get into the digital real of filtering, both true digital filters (digital in, digital out) as well as digital filters implemented on analog in/out signals. I think that would be a very interesting topic for a different book.
here is what i would love to see in your current book mr self.
DIY'ers are tinkerers. I love to tinker with stuff. And filters lends themselves to a lot of tinkering. What are the good practical advices you think one should think about when doing implementation of analog filters to ensure that they stay "tinkerable". just pragmatic stuff like, consider using caps of this decades because this allows you to alter resisters in these locations to change the frequencies, use this style circuit for a bit of EQ'ing because you only need to modify this value of component to change the behavior. These are the lessons that are gained from hands on expertise which i think are invaluable in print. True "this makes things easier and here is why" type of experiences is so much more interesting than the basic science and formulas to the general masses.
then consider doing a followup book which references this book on DSP audio processing followed by a book on pc based audio processing. the pc based audio one will be quickly obsolete, or rather will require frequent revisions but i think it would be useful to you and to the community.
i've also written a few books on implementations (my books was in the area of computer and IT security and disaster recovery) and the consistent feedback i got is that the pragmatic "look out for this" or "consider this in case you want to change something later on" is what the readers really enjoyed.
good luck and please let us know when its going to print!
Given how quickly DSP is advancing, i think a chapter that deals with it on the level of what it is, how it works, what kind of capabilities it brings, but with little specific detail would be appropropriate. If any specifics are added, they will be out-of-date so quickly as to be not much use (like someone's Pentium today).
Missing out on some of the more exotic analog subjects -- such as gyrators -- i think is a mistake. Also are you going to touch on buffered & unbuffered PLLXO?
dave
Yes, and I'd hate to see the book veer off into DSP implementation details. What might be worth a mention are the things that can be done with DSP that are not easy with analogue electronics. For example, digital delay may be useful to compensate for path length differences in a horn system.
OTOH, DSP also allows things like filters with zero phase shift and/or extremely steep slopes, which might seem appealing at face value, but are likely to lead to grief. As anatech's sig says: "Just because you can, doesn't mean you should".
So I'd say DSP is worth perhaps 1 or 2 pages, just to point out what can be done, but not how to do it.
As an aside, I stopped buying electronics magazines when the "construction projects" started being little more than 5 pages of code for a microcontroller. To me that's not DIY electronics, it's programming.
I like this suggestion. I would also like to see a bit about considerations in implementing a DSP crossover, such as gain structure, accumulator bit depth and a feel for how much is enough... dithering, what noise shaping to use and when. Also, some discussion on how to specify/select a DSP system or chip would be useful.
While there are fundamentals of filter design that hold for either DSP or analogue systems, the two approaches are sufficiently different to warrant individual attention.
An overview of available software design packages for modelling and simulating (even emulating) active crossovers would be the cherry on top.
Ditto, Dave has summed it up nicely. We are at the beginning of DSP and it will change fast. But knowing what biquads are and what they do is important. There is already a huge amount of information on filters that can be applied to analog or digital, but I'd like to know how that actually translates into digital. I have a good idea already how it works in analog, but not digital - at all.
Otherwise, with out knowing how digital filters work and are implemented - I'm stuck using someone else's filter functions with the DSP crossover I've build on my kitchen table.
Otherwise, with out knowing how digital filters work and are implemented - I'm stuck using someone else's filter functions with the DSP crossover I've build on my kitchen table.
Which I think is where half of the problem lies.
I have the technical ability to physically create the PCB + soldering of SMD devices, such that I could make my own DSP board. However what I'd want is a digital input accepting I2S that went into a DSP and split it into 8 channels, then would go into multiple PCM1792/4s for D/A conversion.
The trouble is the coding side of things. To operate it I'd want USB functionality, that would easily allow me to alter the xover parameters, like the DCX, whilst offering much more flexibility. DSP stuff, as you said, isn't about DIY, but more about your ability as a programmer. And from what I understand you need to know exactly what your doing so as to preserve as much of the original signal as you can.
This is a major headache, which is why I've always built my own analogue active crossovers, because it does feel like true DIY and because I can. Maybe I feel this way because I cannot code. The irony, is that my brother, who has a degree in programming, but knows zero about making electronics, would stand a better chance of creating a working prototype. Maybe we should work together
I will however most likely be buying the proposed book, simply because it is in every which way relevant to my interests. Maybe a lot of it contains stuff that I already know, but this I do not mind. Looking at something from someone elses perspective never fails to help illustrate the things that you might already know, perhaps a little better. And on top of that it might get me/you thinking about something in an entirely different way, thus encouraging creativity.
Even if DSP is the way forward, analogue filters will still have their place in circuit design. Heck even D/A converters require low pass filters on their outputs, which is somewhat relevant to this book.
My 2c, some of which has been stated above:
However it looks like the book has the bulk of what I'd want as is.
- If you mention specific DSP chips you'll be nearly instantly obsolete
- If you get into some basic DSP theory, avoid the pretentious scholarly sigma notation and explain things in plain English and graphics. It can be done, but usually isn't. The DSP Handbook by Bateman and Paterson-Stephens is the only text I've seen explain DSP well.
- Mention some of the items mentioned here that you won't include, why they might be useful and point us to references.
- I presume in chapter 12 or 13 there will be some guidance on ranges of what R and C values to pick for a given filter Q so as not to under or overload a 'typical' op-amp or otherwise cause impedance problems.
- I'd like to see you throw a simple discrete op-amp into chapter 13, you can then tell us how horrible it is and what's wrong with it (as I'm sure you would) for filters. Or even simpler, the applicability of a discrete follower/buffer (like a JFET follower with a JFET current source load (Pass B1)) for filters that have op-amps that act as followers, and what the limitations of that may or may not be. That'll offset this DSP nonsense for the analog folks.
However it looks like the book has the bulk of what I'd want as is.
I agree with Dave's statement:
One wouldn't need much programming knowledge with nowaday's specialised audio DSP solutions like Analog Devices' familiy of chips that can be programmend with their software called "Sigma Studio". Here you just link functional blocks on a GUI and then the software does the rest. These chips also have their converters "on board". And they come at a very interesting price as well.
BUT you'd still have to know what the analog variant of the crossover that you are building would have to look like. There's no way around that !
Such things could be covered in a short chapter. If you wanted to go deeper it would fill a whole book and it would certainly be beyond the average DIYer's skills.
Apart from that: Analog design is still much more interesting than digital !
Regards
Charles
One wouldn't need much programming knowledge with nowaday's specialised audio DSP solutions like Analog Devices' familiy of chips that can be programmend with their software called "Sigma Studio". Here you just link functional blocks on a GUI and then the software does the rest. These chips also have their converters "on board". And they come at a very interesting price as well.
BUT you'd still have to know what the analog variant of the crossover that you are building would have to look like. There's no way around that !
Such things could be covered in a short chapter. If you wanted to go deeper it would fill a whole book and it would certainly be beyond the average DIYer's skills.
Apart from that: Analog design is still much more interesting than digital !
Regards
Charles
True. However, there'd be no harm in using one as an example for purposes of discussion, especially since this will not be a cookbook.
Which I think is where half of the problem lies.
I have the technical ability to physically create the PCB + soldering of SMD devices, such that I could make my own DSP board. However what I'd want is a digital input accepting I2S that went into a DSP and split it into 8 channels, then would go into multiple PCM1792/4s for D/A conversion.
The trouble is the coding side of things. To operate it I'd want USB functionality, that would easily allow me to alter the xover parameters, like the DCX, whilst offering much more flexibility. DSP stuff, as you said, isn't about DIY, but more about your ability as a programmer. And from what I understand you need to know exactly what your doing so as to preserve as much of the original signal as you can.
This is a major headache, which is why I've always built my own analogue active crossovers, because it does feel like true DIY and because I can. Maybe I feel this way because I cannot code. The irony, is that my brother, who has a degree in programming, but knows zero about making electronics, would stand a better chance of creating a working prototype. Maybe we should work together
I will however most likely be buying the proposed book, simply because it is in every which way relevant to my interests. Maybe a lot of it contains stuff that I already know, but this I do not mind. Looking at something from someone elses perspective never fails to help illustrate the things that you might already know, perhaps a little better. And on top of that it might get me/you thinking about something in an entirely different way, thus encouraging creativity.
Even if DSP is the way forward, analogue filters will still have their place in circuit design. Heck even D/A converters require low pass filters on their outputs, which is somewhat relevant to this book.
Matt,
You'll find that most DSP vendors like TI and ADI have turn-key software packages that reduce the programming effort to graphically interconnect functional blocks on the screen and downloading to the DSP. Some or free, some aren't, and you still need to know what you're doing, but there's no need to get down into the weeds anymore.
jan didden
Hi,
You may not have noticed, but DS writes his books essentially as "References" in the Academic sense, aimed more at students and industry professionals than at DIY'ers. For most DIY'ers his works are rather dry and not really that praxis focused, you will not find extensive compendiums of circuits ready to be build from the page, but rather the fundamentals to design your own or to modify your designs simply for improved performance.
A DIY work on active crossovers would probably be best off with a bunch of active crossover circuits (tested, with dcumented performance and PCB CD in 20 different formats) using OP-Amp's, Discrete Solid State and Tubes (to be all things for all DIY'ers) with tables for crossover Values by Frequency and some advanced chapters how to calculate your own and how to turn the textbook filters into transtitional ones.
And?
Very, very few DIY'ers make their own Speaker Drivers or Build CD-Mechanisms. In many areas DIY simply takes existing Commodity items from the industry and applies them. For example, most DIY'ers CD-Players will be generic commercial "button pushing" items that have been modified to suit the individual DIY'ers taste, preferences and prejudices, be it with the latest "magical Op-Amp" or Tube outputs.
For example the Behringer DCX2496 is a very inexpensive digital crossover that is hampered by poor analogue stages (I nearly lost my breakfast when I first saw the schematic and it was almsot lunchtime) and poor compatibility with HiFi Systems.
This has given rise to a huge range of DIY Modifications to correct these flaws and preserve what is good about it. That to me certainly is valid DIY.
BTW, if fitted with decent analogue stages the Behringer DCX's DAC Chip can offer a dynamic range of 120dBA, which is not far off the analogue system DS references.
Even better, we now can handle the whole crossover, timealignment and room EQ in a PC (that is also media player) with even more flexibility. Good external sound cards with 8 Channels worth of 120dB SND DAC Chips crippled by bad analogue stages can be had for very little money.
Hence a few chapters on making analogue stages that do not cost 7dB SNR or more and offer good performance with modern DAC Chips (including measures to deal with the high levels of ultrasonic noise without compromising performance) combined with chapters how to synthesise the needed filters and equalisations from measured driver responses, how to ensure the system provides good summation on and off axis and so on makes a lot of sense and will be relevant in decades to come.
The book as proposed right now is 75% strictly historical interest, if not now then a few years hence. As DS invited comments on what is missing, we are telling him, that's all.
Ciao T
PS, I have realised several active systems based around digital crossovers for others (my own personal penchant for tubes and vintage speakers means I normally do not use such myself), since the mid 90's arrival of the BSS digital crossovers (the first available at sensible budgets than had all you needed).
There is ABSOLUTELY NO WAY I would ever again attempt to make an active system aimed at really high quality and using digital sources exclusively using analogue active crossovers, it is a complete waste of time and can never match the results of the correctly set digital system, despite taking hugely more time to make it work. Even in a system intended to be used strictly with LP it would have to be a VERY, VERY fat cheque to convince me to do it analogue (if at all).
Life is simply to short for analogue crossovers and was too short 15 Years ago.
You may not have noticed, but DS writes his books essentially as "References" in the Academic sense, aimed more at students and industry professionals than at DIY'ers. For most DIY'ers his works are rather dry and not really that praxis focused, you will not find extensive compendiums of circuits ready to be build from the page, but rather the fundamentals to design your own or to modify your designs simply for improved performance.
A DIY work on active crossovers would probably be best off with a bunch of active crossover circuits (tested, with dcumented performance and PCB CD in 20 different formats) using OP-Amp's, Discrete Solid State and Tubes (to be all things for all DIY'ers) with tables for crossover Values by Frequency and some advanced chapters how to calculate your own and how to turn the textbook filters into transtitional ones.
And?
Very, very few DIY'ers make their own Speaker Drivers or Build CD-Mechanisms. In many areas DIY simply takes existing Commodity items from the industry and applies them. For example, most DIY'ers CD-Players will be generic commercial "button pushing" items that have been modified to suit the individual DIY'ers taste, preferences and prejudices, be it with the latest "magical Op-Amp" or Tube outputs.
For example the Behringer DCX2496 is a very inexpensive digital crossover that is hampered by poor analogue stages (I nearly lost my breakfast when I first saw the schematic and it was almsot lunchtime) and poor compatibility with HiFi Systems.
This has given rise to a huge range of DIY Modifications to correct these flaws and preserve what is good about it. That to me certainly is valid DIY.
BTW, if fitted with decent analogue stages the Behringer DCX's DAC Chip can offer a dynamic range of 120dBA, which is not far off the analogue system DS references.
Even better, we now can handle the whole crossover, timealignment and room EQ in a PC (that is also media player) with even more flexibility. Good external sound cards with 8 Channels worth of 120dB SND DAC Chips crippled by bad analogue stages can be had for very little money.
Hence a few chapters on making analogue stages that do not cost 7dB SNR or more and offer good performance with modern DAC Chips (including measures to deal with the high levels of ultrasonic noise without compromising performance) combined with chapters how to synthesise the needed filters and equalisations from measured driver responses, how to ensure the system provides good summation on and off axis and so on makes a lot of sense and will be relevant in decades to come.
The book as proposed right now is 75% strictly historical interest, if not now then a few years hence. As DS invited comments on what is missing, we are telling him, that's all.
Ciao T
PS, I have realised several active systems based around digital crossovers for others (my own personal penchant for tubes and vintage speakers means I normally do not use such myself), since the mid 90's arrival of the BSS digital crossovers (the first available at sensible budgets than had all you needed).
There is ABSOLUTELY NO WAY I would ever again attempt to make an active system aimed at really high quality and using digital sources exclusively using analogue active crossovers, it is a complete waste of time and can never match the results of the correctly set digital system, despite taking hugely more time to make it work. Even in a system intended to be used strictly with LP it would have to be a VERY, VERY fat cheque to convince me to do it analogue (if at all).
Life is simply to short for analogue crossovers and was too short 15 Years ago.
Hi Douglas,
That would certainly be useful. But as someone who has done it both ways more than a few ways I would suggest that the approaches I would use when using a digital crossover would differ greatly from those I'd take with analogue ones.
Analogue systems are poor chocies for adjustability and flexibility and to correct large amounts of time-shift due to mechanical arrangements is a fools errand.
On the other hand digital systems are extremely flexible and even the correction of massive mechanical time delays is trivial.
While not available in most commercial systems, digital systems allow us to divorce the frequency domain and time domain (eg. they are not required to be minimum phase systems) in ways that are extremely difficult and complex, if at all realisable in analogue systems. This has implications for system design as we can overcome some of the fundamental limitations in analogue (active or passive) multi-way speaker systems this way.
So to simply blanket comment "it is applicable no matter whether you go DSP or NE5534" would not get remotely close to getting the best from DSP based systems, it merely allows us to use DSP to emulate 1980's analogue technology, which I am unsure is of much if any use.
First, it should cover all the areas where DSP based systems are different and arguably dramatically superior to analogue approaches and illustrate the ideas, ideally with measurements (polar plots etc) to illustrate the crushing superiority of well applied digital technology to traditional analogue systems. You may need a co-author on this subject, as it seems you are not necessarily the expert on that (neither am I, I hasten to add).
Secondly, you would need to cover the issues of interfacing the digital world with the analogue one, an area where in much of the available gear appalling atrocities are committed. This is probably one area where your unique abilities and skills can add much and when I last looked you had not really done much work there (DAC and ADC interfacing has some unique challenges)...
Not really. DSP Chip's are like Op-Amp's, except more so. There is always another new "magical" one every other week and the variety is massive. I mean your book will not cover each and every Op-Amp made by every Vendor in the world, will it?
However, just as you when covering Op-Amp's you teach people what to look for when implementing them and why a Fet Input Op-Amp may be better suited in one are of a circuit and bipolar in another you could cover the way these things work and what one needs to look to select the right one in a given application.
Most modern DSP's come with software support packages that already take care of the optimised low level implementation.
So rather a tutorial using one such solution how to implement a given Speaker, with given drivers and mechanical geometry using advanced DSP techniques (that is not just replicating OP-Amp Filters and EQ) is what I would suggest.
If you approached the major manufacturers for help, who knows, one or two may be quite willing to support you substantially in doing so, as a way to promote their wares.
A lot of prepacked "DSP on a board" systems are already available widely, in differing configurations. Among them the DIY focused "MiniDSP" system that was referenced several times already in this thread.
As computers are becoming more prevalent choices as Playback Systems all the DSP one could wish for can be done using the CPU and in more modern configurations using the Graphics Card processors.
There are now DSP Packages for PC that can make use of the "CUDA" interface to harness the power build into graphics cards (such as the NVIDIA one in my cheep and cheerful Asus Laptop) for gaming to do audio DSP.
This area is and will remain in flux, but the first principles and fundamentals to handle the processing for a multiway speaker do not change if I am running on Motorola 56K DSP, an AD Blackfin or a TI TMS32XXXX or via a software package on an Intel CPU with help fropm a CUDA linked NVIDA Graphics card.
Ciao T
That would certainly be useful. But as someone who has done it both ways more than a few ways I would suggest that the approaches I would use when using a digital crossover would differ greatly from those I'd take with analogue ones.
Analogue systems are poor chocies for adjustability and flexibility and to correct large amounts of time-shift due to mechanical arrangements is a fools errand.
On the other hand digital systems are extremely flexible and even the correction of massive mechanical time delays is trivial.
While not available in most commercial systems, digital systems allow us to divorce the frequency domain and time domain (eg. they are not required to be minimum phase systems) in ways that are extremely difficult and complex, if at all realisable in analogue systems. This has implications for system design as we can overcome some of the fundamental limitations in analogue (active or passive) multi-way speaker systems this way.
So to simply blanket comment "it is applicable no matter whether you go DSP or NE5534" would not get remotely close to getting the best from DSP based systems, it merely allows us to use DSP to emulate 1980's analogue technology, which I am unsure is of much if any use.
First, it should cover all the areas where DSP based systems are different and arguably dramatically superior to analogue approaches and illustrate the ideas, ideally with measurements (polar plots etc) to illustrate the crushing superiority of well applied digital technology to traditional analogue systems. You may need a co-author on this subject, as it seems you are not necessarily the expert on that (neither am I, I hasten to add).
Secondly, you would need to cover the issues of interfacing the digital world with the analogue one, an area where in much of the available gear appalling atrocities are committed. This is probably one area where your unique abilities and skills can add much and when I last looked you had not really done much work there (DAC and ADC interfacing has some unique challenges)...
Not really. DSP Chip's are like Op-Amp's, except more so. There is always another new "magical" one every other week and the variety is massive. I mean your book will not cover each and every Op-Amp made by every Vendor in the world, will it?
However, just as you when covering Op-Amp's you teach people what to look for when implementing them and why a Fet Input Op-Amp may be better suited in one are of a circuit and bipolar in another you could cover the way these things work and what one needs to look to select the right one in a given application.
Most modern DSP's come with software support packages that already take care of the optimised low level implementation.
So rather a tutorial using one such solution how to implement a given Speaker, with given drivers and mechanical geometry using advanced DSP techniques (that is not just replicating OP-Amp Filters and EQ) is what I would suggest.
If you approached the major manufacturers for help, who knows, one or two may be quite willing to support you substantially in doing so, as a way to promote their wares.
A lot of prepacked "DSP on a board" systems are already available widely, in differing configurations. Among them the DIY focused "MiniDSP" system that was referenced several times already in this thread.
As computers are becoming more prevalent choices as Playback Systems all the DSP one could wish for can be done using the CPU and in more modern configurations using the Graphics Card processors.
There are now DSP Packages for PC that can make use of the "CUDA" interface to harness the power build into graphics cards (such as the NVIDIA one in my cheep and cheerful Asus Laptop) for gaming to do audio DSP.
This area is and will remain in flux, but the first principles and fundamentals to handle the processing for a multiway speaker do not change if I am running on Motorola 56K DSP, an AD Blackfin or a TI TMS32XXXX or via a software package on an Intel CPU with help fropm a CUDA linked NVIDA Graphics card.
Ciao T
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.