The design of active crossovers- Douglas Self wants your opinions

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hello all:

I am in the process of putting together a book, on active crossover design. This is planned as a weighty tome of 480 pages or so, and I am trying to make it as comprehensive as possible. To this end I have put the provisional Table of Contents on my website at The Douglas Self Site

I would be very glad to know about it if anyone thinks that anything is is missing. Even in 480 pages I cannot hope to cover the whole field, but I want to make sure that nothing important is left out. I would also be glad to hear any opinions that anyone may have on this project. In particular I would like to know if anyone is aware of any other book on the topic- I am not.

I would stress that this post is not an advertisment- publication is many months away. This is going to take time!
 
Looks like a great book! I assume from the table of contents that you'll be dealing with the excess distortion issues of Sallen-Key (e.g., the Billam paper in JAES)? Will you be talking about the use of GICs/gyrators and the like in quasi-passive filter implementations?

I'll be diplomatic and not ask about showing vacuum tube versions...:D
 
Will you be talking about the use of GICs/gyrators and the like in quasi-passive filter implementations?

I have to say it seems unlikely. I have sketched out the chapter on highpass & lowpass filters, dealing just with S&K and MFB types, and it looks like it is already the longest in the book. But thanks for the input- if I can squeeze it in I will.
 
Sounds like a great book! Looking forward to see it published!
The other book like that (that I know of) is Active Filter Cookbook by Don Lancaster. ©1975
There's another one, Moshits, G., Horn, P. Active Filters Design ©1981 - I only have it in russian translation, though it seems to be originally from the same US publisher as the previous one.

P.S. - Now that I think of it, these books aren't about Active Crossovers per se, but they discuss very similar topics anyway...
 
Last edited:
Sounds like a great book! Looking forward to see it published!
The other book like that (that I know of) is Active Filter Cookbook by Don Lancaster. ©1975
There's another one, Moshits, G., Horn, P. Active Filters Design but I only have it in russian translation, but it seems to be from the same US publisher as the previous one.

Ah yes, but they are books on general active filter design, not specifically active crossover design.
 
Any chance of getting some discussion on digital crossovers? Maybe just highlighting what's different when doing crossovers in the digital domain?

Much of the book will be on the basic crossover and filter concepts, which can be applied in both the analogue & digital domains. I'm not however planning to get into any details of digital implementation.

Many thanks for your input.
 
Hi Douglas,
I have read the TOC, it's very dense and seems cover the essential about analog active filter. This book is promising :).

For a first impression, you don't talk about amplification with chips amps, often used in this kind of design.
The second thing is about numerical XO ? It will be an other book :D
But can you talk a little about this ? For me it is very logical to talk about numerical processing because the source is numerical, now amplifiers become numeric (class D) and the process between the two (source and amplifier) should be ?

Third, when we design loudspeakers we use measurement tools, simulation tools ... A chapter on these ? I think you cannot avoid this. How can you evaluate time delays ? could be a very instructive chapter
For a more general design of active filter how to trace transfer response.

Some more precises impressions :
Chapter 2 : Can you talk about open baffle ? Very useful to use active XO. I have seen this in chapter 11

Chapter 13 : The AOP survey, i will curious to see the list ;) You have in the TOC at least five generation of AOPs ;)

Hope this helps if I notice other things, I will post later.

Rodolphe.
:)
 
Hi,

The contents look very comprehensive and I'm sure will cover lots of filter theory.

However loudspeakers are not electrical circuits, and I've read too many articles
by electrical filter designers about designing loudspeakers that are fairly clueless.

I'm not saying that is applicable here, that is presumptious, but its the acoustic
responses that matter, and the required electrical filters, just like passive c/o's,
usually have nothing much to do with any classical electrical filter parameters.
That may seem obvious, but far too many want to assume perfect drivers,
and/or ignore the fact that that might be only true into infinite halfspace.

An area of interest worth exploring in such a tome IMO :

Incorporating the the power amplifiers as active devices in the schema,
not many people know how to go about doing this, and what can be done.
e.g. its easily possible to build the treble c/o into the treble amplifier.
(i.e. high pass is easy, but not low pass due to stability issues.)

Its really about combining flexible gain with filter functions.

That might lead on to another interesting practical issue :
Varying the amplifiers output impedance via feedback. That leads on to
the Stahl ACE bass or similar principles using complex output impedances.

An extension of your previous work regarding the statistical distribution
of the dynamics of a musical signal might be to extend this to the
seperate band limited sections of an active speaker, unlike the
former I'd expect different distributions with different music types.

Say you ~ knew this for each typical 1/3 octave. What is ~ the
relative powered required for each section ? How to derive it ?

There is so much nonsense about the active amplifier capabilities
required on the internet, a sensible analysis would be really good.

rgds, sreten.

A 3rd order elliptical lowpass filter and a 2nd order highpass
used in reality to produce 4th order L/R lowpass and highpass.
http://www.zaphaudio.com/audio-speaker17.html
 
Last edited:
Hello,
480 that is a lot of pages, perhaps a few more regarding the application of low noise discrete JFET’s.
Zero feedback JFET circuits would be of interest to me. Shame on me I bought 2K 2SK170’s
Thanks to you, your website and books I have built many circuits with 5532 Op-Amps.
DT
All just for fun!
 
Hi,

The book looks very promising, and will be on my shopping list when available. What I miss in a lot in many of the current audio books available, is measurement examples actual of actual loudspeakers and a written description on how these measurements need to be interpreted. Nowadays it easy to make high resolution measurements with low cost gear (pc + soundcard), but it as before to interpret these correctly.

The ideal book for me would place the focus on how to select between different types of cross-over/filters (based on measurements), rather then on the implementation details of active cross-overs.
 
Doug,

The TOC of your book looks promising. I wish you all the best for the writing stage.

As filter design is a complex topic (one need only take a look at the "handbook" by Wai-Kai Chen to understand this), you will need to cut out relevant material somewhere.

Important things that are often, IMHO, wrongly omitted:

1. Related to section 3.1 of your book, the definition of "adequate flatness" often does not take into account that at the crossover point, two drivers are connected to the room, meaning that the effective total driver efficiency is actually slightly higher, so we would often like an amplitude dip in the summed filter outputs. I'm not sure you planned to cover this; anyhow, I think it's important.

2. From the TOC, I cannot see a reference to baffle step diffraction. I think it's relevant.

3. Multiple-feedback or "Rauch" filters are convenient especially in integrated active crossover designs, as the resulting capacitor values are often small (=convenient in integrated filter design). This is often omitted; I think it's important.

4. Antoniou's General Impedance Converter is IMHO a very relevant synthesis tool. I think the difference between series-L and shunt-C ladder prototypes w.r.t. dynamic range is relevant.

5. Perhaps Voltage-Controlled Filters (VCFs) are relevant to mention, as they have special properties. They are mainly (and very scarcely) mentioned in older books, I can dig in my library to provide you with at least one reference, if you like. I find it remarkable that modern equipment such as Allen&Heath mixing consoles still use such filters.

6. I see you plan to cover time delay filters. I hope LC time-delay filter prototypes deserve a mention... they are especially hard to wrap your head around, due to their nonplanar topologies.

7. Perhaps series-filter prototype simulation is relevant. I love passive series-filters and I think under some circumstances it could make sense to have an actively simulated version.

8. An important topic in industry is to have good PSRR of the filter. Perhaps this is also relevant for the book (tricks such as referencing a dummy AC coupling cap on your filter input differential pair to a supply rail to make PS ripple turn into common-mode).

--
Greetz,
MatchASM
 
2. From the TOC, I cannot see a reference to baffle step diffraction. I think it's relevant.


Greetz,
MatchASM

Hi, "Diffraction compensation equalisation" is in there, rgds, sreten.

What I don't like is a complex chain of EQ dealing with this and that separately.
Very often the whole thing could be far simpler, but not as easy to adjust.
 
Last edited:
Administrator
Joined 2004
Paid Member
This looks like it will be a good a thorough book. I especially like chapters 14, 16 & 17. Interfacing it with the real world and keeping an eye on gain structure is a vast area that not many people seem to understand, and it can make a huge difference.

However, to be a real party pooper, I don't see why a crossover book to be published in the 2011 would be so opamp centric. Who would use them for an active crossover in the 21st century, except as input and output buffers? In this day of ever cheaper and better DSP, who would build an active crossover with opamps? It seems so 1980s. Not that it can't be done well, but who would? Some hobbyists or students, perhaps, but that's all.

DSP and biquads are the flavor of the future. Learning how to use those to achieve the filter slopes laid out in your major chapters should be the interest of the forward looking crossover designer - IMO. Opamps make good examples, but I can't see them making very many crossovers in the future.

Sorry for the downer. :(
 
DSP and biquads are the flavor of the future. Learning how to use those to achieve the filter slopes laid out in your major chapters should be the interest of the forward looking crossover designer - IMO. Opamps make good examples, but I can't see them making very many crossovers in the future.

Sure DSPs are more interesting, especially for high-end use.

But consider: Price NE5532 ~$0.10, Price TMS320VC5501 ~$5.00. Op-amps will be around for a long time.

--
Regardz,
MatchASM
 
I agree with Pano - DSP is the way of the future, and to leave things in the analog domain would encourage early obsolescence of the book. I've been designing active crossovers for a couple of years now, and would never consider doing them analog anymore, primarily because of the cost, and secondarily because of the lack of flexibility. I've designed/built complex analog crossovers that cost over $300, and are totally fixed in their design, and take many many hours from start to finish, and are prone to circuit errors. Nowadays, I can spend $200, and get an extremely flexible DSP board, and within minutes be listening to different crossover variations. That is a huge difference.

Not only that, but DSP can do things that analog can't, like time and phase corrections. I guarantee that the future of loudspeakers will be heavily influenced by DSP - it is the only new technology that is making a significant differences to what is currently possible.
 
I'm not however planning to get into any details of digital implementation.

In my view a wise decision. Digital XOs would require a whole book.

MatchASM said:
But consider: Price NE5532 ~$0.10, Price TMS320VC5501 ~$5.00. Op-amps will be around for a long time.

That's hardly an apples-apples comparison - where can you get a 5532 for 10 cents incidentally? :) The opamp will certainly remain for simple XOs in cheap active speakers. For even something like the Linkwitz Pluto ASP it takes many opamps - the DSP solution might well be more cost-effective.

A common or garden DSP nowadays has 100's of MMAC/s performance - that's a hell of a lot of bi-quads at 44k1. A dual opamp can do a pair of 2nd order stages. If the DSP needs 5 MACs per bi-quad then it'll do roughly 5 stages per MMAC of performance. At 100 MMACs and $5 that makes just 1c per stage - five times cheaper than the opamp.
 
The opamp will certainly remain for simple XOs in cheap active speakers.

= many X-overs. This is the only point I was trying to make (please read the post I quoted previously). BTW, if you order a couple tens of thousand 5532s, I'll bet you Fairchild will cut you a deal in the order of $0.10 a piece.

Regarding apples-apples comparison, it's not meant to be. The DSP still needs an ADC, a DAC + reconstruction filter for every output, more board space, more test time (also lower yield due to higher transistor count and pin count), more supply rails, additional clock circuitry, etc. I omitted those for simplicity. The point is: simple 2nd order filter, go opamp!

@doug: I'm sorry for taking up so much space in your thread defending the validity of the book. I'll try to come up with a simpler example next time.

--
Greetz,
MatchASM
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.