Hi Shaun,
Thanks. I did not know XoverWizard, I played some minutes with it. Sure, nice program. I'll test it more.
I am now very accustomed to my TINA simulator and I think I can do a little more, maybe a bit slower, with it than with the wizard. An RLC circuit filter in simulation is also more visualy speaking than knowing its frequency response of a second order filter with a Q value. And I use Tina for other purposes like amp and servoed drivers circuits.
Currently, I am looking a way for summing only the modules of the frequency responses.
Regards.
Thanks. I did not know XoverWizard, I played some minutes with it. Sure, nice program. I'll test it more.
I am now very accustomed to my TINA simulator and I think I can do a little more, maybe a bit slower, with it than with the wizard. An RLC circuit filter in simulation is also more visualy speaking than knowing its frequency response of a second order filter with a Q value. And I use Tina for other purposes like amp and servoed drivers circuits.
Currently, I am looking a way for summing only the modules of the frequency responses.
Regards.
What discourages the use of active crossovers, is the need for additional power amplifiers and the prospect of their additional costs. If such amplifiers are purpose designed to meet the needs only of the drivers they power, total system cost can be reduced below that of a conventional system of equal quality.
Book sections dealing with Power amplifier design and driver interfacing I believe are the key ingredient missing from most, if not all, treatments of the subject of active crossover design. Maybe content of the book you are working on should be divided into several volumes rather than contained in one very large tome.
Three volumes come to mind in this regard, they are:
1) Analog Signal Processing,
2) Digital Signal Processing, and
3) Driver Powering, Interfacing and Control.
Regards,
WHG
Last edited:
An "Off Topic" question:
I am looking for a CAD software for mixed crossover design, i. e. both active design include equalizing and passive design at the same time.
For the passive design include equalizing I use since a long time the CAD programm "BASSYST" and I get in combination with "DAAS3NT" excellent results, cause the posibility of "REAL TIME" design. The URL therefore is
Beschreibung BASSYST
(unfortunately only in German, but I think, about language tools you get not too much losses by engl. translate)
For the active design I use until now "circuitmaker" (that I also use for amp design), to use considerably more complicated than "BASSYST". "Circuitmaker 2000" is similar to
Linear Technology - Design Simulation and Device Models
or
Cadence OrCAD Solutions
but much more older and no longer supported.
I want a CAD programm like bassyst, where it is possible, to work with active crossover networks but also with this comfortable features of bassyst.
What are currently on the marked ??
Thank you for any hints.
I am looking for a CAD software for mixed crossover design, i. e. both active design include equalizing and passive design at the same time.
For the passive design include equalizing I use since a long time the CAD programm "BASSYST" and I get in combination with "DAAS3NT" excellent results, cause the posibility of "REAL TIME" design. The URL therefore is
Beschreibung BASSYST
(unfortunately only in German, but I think, about language tools you get not too much losses by engl. translate)
For the active design I use until now "circuitmaker" (that I also use for amp design), to use considerably more complicated than "BASSYST". "Circuitmaker 2000" is similar to
Linear Technology - Design Simulation and Device Models
or
Cadence OrCAD Solutions
but much more older and no longer supported.
I want a CAD programm like bassyst, where it is possible, to work with active crossover networks but also with this comfortable features of bassyst.
What are currently on the marked ??
Thank you for any hints.
Last edited:
I'd like to thank everybody who has contributed to this discussion.
I am afraid there will be no DSP-specific content, for as several of you have pointed out, to do it justice would require a book of twice the size. As I have said, a large part of the book will still be useful to those planning a DSP approach.
I have made quite a number of changes to the contents as a result of suggestions posted here, and a revised list can be seen at:
The Design of Active Crossovers by Douglas Self
The writing is finished and the book is now with my publishers; it should hit the shelves in May 2011.
Once again, many thanks to you all.
I am afraid there will be no DSP-specific content, for as several of you have pointed out, to do it justice would require a book of twice the size. As I have said, a large part of the book will still be useful to those planning a DSP approach.
I have made quite a number of changes to the contents as a result of suggestions posted here, and a revised list can be seen at:
The Design of Active Crossovers by Douglas Self
The writing is finished and the book is now with my publishers; it should hit the shelves in May 2011.
Once again, many thanks to you all.
I haven't read the complete thread very carefully - i. e. the risk, that the follow is a repeating from an other member, is very large:
I want to know the right way for finding the active crossover topology with exactly the same acoustical funktion than the (already present) passive transmission function (transfer function) at certainly speakers.
I. e., how looks an active crossover than the same results than the passive crossover network like follow:
http://www.kef.com/Resources/KEFTopics/KEFTOPICS_vol3no2_the calinda and cantata loudspeakers.pdf
That is the question for me.
With respect to copper prices for air inductors (for passive crossover networks in loudspeakers), this would be particularly interesting to know.
Last edited:
Will we see some of your designs from the book available as PCBs from signal transfer soon.
The final chapter contains a high-performance generic Linkwitz-Riley design which could well be made available as a PCB. However, it does not currently include any provision for equalisation stages. I would like such a PCB to be as versatile as possible so I would be glad to hear all suggestions as to what facilities should be included. How many EQ stages? What sort? Metering? Level controls?
Hi Douglas,
I assume the "generic LR" filter you refer to is a modern option with a subtractive highpass, not sallen key?
For EQ stages, I would suggest a fairly large bank of EQ stages that can be assigned to each way for either cut or boost to suit. As simple gyrators can be made quite flexible it would be easy to make these work for either boost or cut with suitable resistor and capacitor values with a wide range of frequencies and Q's, maybe 6 to 8 gyrators per channel freely assignable should do. In addition the LF at least needs a linkwitz transform and a highpass.
I would also look at allowing additional "shelf" type EQ's with minimal added parts so that large "trending" non flatness (be it of the smoothly rising response of many mid/woofers or the CD Horn rolloff present with waveguide loaded tweeters). This could be combined with a level adjustment stage.
I would suggest arranging this as much as possible that additional EQ does not add many more avoidable stages.
I would suggest to use only single op-amp's with the PCB pattern for both SMD and TH Op-Amp's, all else TH parts, this way people can use whatever kind of op-amp's they like (instead of being limited to NE5532's and TL082's).
Ideally one would also include options to have both over-level (clipping) and thermal limiters for each way and dynamic adjustment of the Bass EQ and Bass Highpass to allow the drivers to be protected from excessive power (thermal overload), power amplifier clipping and over-excursion.
Plus, it would make sure the PCB has equal abilities to any cheap digital crossover such as a Behringer DEQ2496, a dBX Driverack or an Alto Maxidrive include routinely.
Ciao T
I assume the "generic LR" filter you refer to is a modern option with a subtractive highpass, not sallen key?
For EQ stages, I would suggest a fairly large bank of EQ stages that can be assigned to each way for either cut or boost to suit. As simple gyrators can be made quite flexible it would be easy to make these work for either boost or cut with suitable resistor and capacitor values with a wide range of frequencies and Q's, maybe 6 to 8 gyrators per channel freely assignable should do. In addition the LF at least needs a linkwitz transform and a highpass.
I would also look at allowing additional "shelf" type EQ's with minimal added parts so that large "trending" non flatness (be it of the smoothly rising response of many mid/woofers or the CD Horn rolloff present with waveguide loaded tweeters). This could be combined with a level adjustment stage.
I would suggest arranging this as much as possible that additional EQ does not add many more avoidable stages.
I would suggest to use only single op-amp's with the PCB pattern for both SMD and TH Op-Amp's, all else TH parts, this way people can use whatever kind of op-amp's they like (instead of being limited to NE5532's and TL082's).
Ideally one would also include options to have both over-level (clipping) and thermal limiters for each way and dynamic adjustment of the Bass EQ and Bass Highpass to allow the drivers to be protected from excessive power (thermal overload), power amplifier clipping and over-excursion.
Plus, it would make sure the PCB has equal abilities to any cheap digital crossover such as a Behringer DEQ2496, a dBX Driverack or an Alto Maxidrive include routinely.
Ciao T
From experience you'd probably want to include:
Tweeter network;
Ability to go up to 4th order electrical highpass.
At least one notch filter.
Variable gain.
All pass time delay, three networks would need to be cascaded for time alignment in high xover frequency xovers.
Midrange;
Ability to go to 4th order electrical on both the high and low pass.
At least one notch filter.
A shelving filter to compensate for bafflestep roll off. If another notch filter is included here then the shelving filter + notch can compensate for open baffle roll off + OB peak before roll off.
Variable gain.
Bass;
2nd-4th order electrical highpass to protect the driver from low frequencies that it cannot really reproduce.
Ability to go up to 4th order electrical on the low pass.
At least one notch filter.
A shelving filter to compensate for bafflestep roll off. If another notch filter is included here then the shelving filter + notch can compensate for open baffle roll off + OB peak before roll off.
A Linkwitz transform circuit.
Variable gain.
You might want give the flexibility to add in a peaking filter too.
If a three way board had all of that built in you'd have the flexibility to design a great number of three ways around it. The trouble is that it's quite complex, would require a decent sized PCB and a large number of components, it would be quite expensive too. Using surface mount components would be a fantastic way to keep the real estate requirements to a minimum, especially for the caps. And I'd wager that anyone half decent with a soldering iron could work with 1206/1210 sized components.
Of course the reason for the complexity is that it contains all the stuff you'd need to design around less then ideal drivers, such as a wave-guide loaded tweeter, a metal cone mid or metal cone bass. Plus you might want to add a couple of notch filters that'd affect the entire spectrum, to add a presence 'dip' or notches for flattening room modes.
Maybe it'd be better if you didn't produce a PCB as all of that sounds like a real headache (I know it's a headache I've built a 4 way version of it and with trimpots in place of all the resistors to give flexibilty). It's easy to see why people buy a DCX.
Tweeter network;
Ability to go up to 4th order electrical highpass.
At least one notch filter.
Variable gain.
All pass time delay, three networks would need to be cascaded for time alignment in high xover frequency xovers.
Midrange;
Ability to go to 4th order electrical on both the high and low pass.
At least one notch filter.
A shelving filter to compensate for bafflestep roll off. If another notch filter is included here then the shelving filter + notch can compensate for open baffle roll off + OB peak before roll off.
Variable gain.
Bass;
2nd-4th order electrical highpass to protect the driver from low frequencies that it cannot really reproduce.
Ability to go up to 4th order electrical on the low pass.
At least one notch filter.
A shelving filter to compensate for bafflestep roll off. If another notch filter is included here then the shelving filter + notch can compensate for open baffle roll off + OB peak before roll off.
A Linkwitz transform circuit.
Variable gain.
You might want give the flexibility to add in a peaking filter too.
If a three way board had all of that built in you'd have the flexibility to design a great number of three ways around it. The trouble is that it's quite complex, would require a decent sized PCB and a large number of components, it would be quite expensive too. Using surface mount components would be a fantastic way to keep the real estate requirements to a minimum, especially for the caps. And I'd wager that anyone half decent with a soldering iron could work with 1206/1210 sized components.
Of course the reason for the complexity is that it contains all the stuff you'd need to design around less then ideal drivers, such as a wave-guide loaded tweeter, a metal cone mid or metal cone bass. Plus you might want to add a couple of notch filters that'd affect the entire spectrum, to add a presence 'dip' or notches for flattening room modes.
Maybe it'd be better if you didn't produce a PCB as all of that sounds like a real headache (I know it's a headache I've built a 4 way version of it and with trimpots in place of all the resistors to give flexibilty). It's easy to see why people buy a DCX.
But that would mean that the derived slope was only 6 dB/octave. I know about the L&V paper but adding delay filters is not as simple as some people think. Sallen & Key for me, I think.
Sort of a graphic equaliser with fixed cut/boost settings? Interesting idea. The book does cover the use of gyrators for this sort of thing, but I hadn't considered using 8 at once.
Certainly, but hard to combine with level control.
That's going to make for a big PCB. I think it will be 8-pin DIL only, allowing 5532s or LM4562 if you're feeling prosperous.
But how would the crossover know what was going on in the power amp? I think this is getting a bit over-complicated.
Many thanks for your input, which I will carefully think over.
Any thoughts on frequency range and Q of the notch?
I have something a bit more sophisticated in mind for the delay, such as a third-order allpass.
Not sure I quite follow about the notch & peak
Two bafflestep compensators?
Surely a job for a room equaliser rather than a crossover
A challenge! Many thanks for your input.
How about breaking up such a complicated crossover into 2 parts, a design that handles the EQ and one that handles the basic crossover with level control. This might be easier to handle.
Steevo
Steevo
Usually tweeters don't have any large steep peaks that require attenuating, if they do they are a bad tweeter!. As a waveguide SEAS DXT is a good example of a small one. The nominal output for this tweeter without the horn would be around 87-88dB as indicated by the flattening of the response between 15-20khz. The hill west of this is the additional output created by the waveguide.
Now the highpass can do quite a lot to shape the lower end, but as you can see you'd really need a low Q notch centred around about 4khz, with about a 6-8dB cut. The larger the wave-guide the lower this hump goes.
Sounds interesting already 🙂
Usually one doesn't go about adding a peak (the opposite of a notch filter) via a peaking filter, because lots of on axis dips fill in as you go off axis. And any attempt at filling then will correct the on axis performance but create a bump in the power response. Nevertheless there are a few drivers that genuinely do have a dip in their response that requires EQ. I would hasten to add that these are rare, so leaving the peaking filter off the board wouldn't be a bad way to reduce complexity.
I apologise I expressed that poorly. I meant include another notch with the shelving network, where you'd bypass the notch if you're building a closed box. But include it if you were going open baffle. These peaks tend to be in the range of say 200-800hz, require 3-6dB of attentuation with a Q of around 5.
I agree, Linkwitz however includes a couple of notches on his boards I think to remove severe room modes if you should have them. This is an area I think you could leave off to reduce complexity, as they'd be quite difficult to implement correctly and could do more harm then good.
You're welcome, challenges are always fun I find, that's if they end up being productive.😉
I did wonder about that. Have a xover board that handles all the high/low pass, gain, delay and shelving functions. That would allow a three way with well behaved drivers to be constructed.
The notches start becoming necessary when you want to make an open baffle or use metal cone drivers. You'd then have to offer single boards that add in notches etc. It might just be easier to add in the notches to the overall design, but it might be worth considering.
If you want true balanced operation you'd have to double up the number of boards. One handles the + signal the other handles the - signal. Buy twice as many boards!
It isn't hard though to swap the input buffer from single ended to balanced operation, although this does put a limit on the input impedance if you want to keep the noise low.
Indeed, but I don't think anyone has produced a full featured board that can handle a three way. Which is imo where going active really takes off.
The trouble here is balancing the cost, especially when things like the minidsp exist for so little. Of course the minidsp won't match the analogue version for sound quality, nor specification. I would never buy one as a result. But if the analogue board ended up too expensive it would significantly reduce the number sold - that however might not be a driving goal.
- Status
- Not open for further replies.