Dear Douglas,
Hmmm, I do remember a certain state variable derived topology that had symmetrical slopes. I need to dig through my notes and references. I have not really pursued active filter design much since the mid 90's so I am not totally up to speed.
Well, yes. Kind of.
Obviously, as we will be fitting all the resistors and capacitors each gyrator block is just a basic "virtual LC Trap" which can be set to almost any desired frequency, Q and boost/cut level.
My point of having 8 is so that you can assign each gyrator freely to any of the three ways, simply by fitting the resistor that sets boost/cut into the right position, so it connects to the right way.
So, in theory you could have all 8 Gyrators on the bass and use them for room correction, or you can have one in the HF channel to EQ a resonance in the driver, two or three in the midrange to deal with the various and annoying resonances present there and the rest in the bass...
Not at all. If your level control uses a similar structure to that in your pre-amp from WW then it is easy to add the cut/boost.
Note, I see NON of the settings (including level) adjustable by potentiometer, but all using fixed resistors. This way we can "design in" everything.
I would not use either one of these IC's. Take my advise, give people singles, so they have the complete choice and you will sell more units.
Adding an SMD footprint below a DIP IC is trivial, I do it all the time, again, it gives more flexibility to your customers and thus you will sell more.
Well, two ways. The so called processor PA's of the 1980's usually used sense-lines.
My own application used instead simple "predictive mode", iy required setting the power amplifiers to a fixed gain and using power amplifiers with known clipping behaviour.
The thermal limiter used a true RMS chip with a very long hold time and was set so that output of the poweramp would not exceed the RMS power handling of the attached driver.
The clipping limiter was a simple and fast peak limiter, set to limit at the rated power of the Amp.
The excursion limiter for the bass used a filter to boost the low frequencies in a pattern analogue to the excursion in the (vented) box (makes for interesting filter design) and used again fast peak limiting.
The way I had it implemented I had one optocoupler with optoresistor in the peaking 2nd order highpass which would first remove the boost and eventually silde the LF cutoff upwards dynamically and after that range was "used up" it would start applying the peak limiter which also used an optocoupler. The Thermal limiter would adjust the peak limiter treshold if the RMS power was exceeded.
That was all done using discretes and op-amp's in the 1980's. It gave us the ability to run much bigger Amp's and get a lot more SPL out of our touring stacks, while at the same time the number of driver repairs we had fell extremely...
Coincidentally the same kind of behaviour is programmable into all modern Pro-Sound digital crossovers.
As someone who has used active PA systems (and other speakers) for quite a few years I would NOT use any active system without some form of protection for the individual drivers. It is too easy to blow up drivers otherwise.
Ciao T
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.
Hmmm, I do remember a certain state variable derived topology that had symmetrical slopes. I need to dig through my notes and references. I have not really pursued active filter design much since the mid 90's so I am not totally up to speed.
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.
Well, yes. Kind of.
Obviously, as we will be fitting all the resistors and capacitors each gyrator block is just a basic "virtual LC Trap" which can be set to almost any desired frequency, Q and boost/cut level.
My point of having 8 is so that you can assign each gyrator freely to any of the three ways, simply by fitting the resistor that sets boost/cut into the right position, so it connects to the right way.
So, in theory you could have all 8 Gyrators on the bass and use them for room correction, or you can have one in the HF channel to EQ a resonance in the driver, two or three in the midrange to deal with the various and annoying resonances present there and the rest in the bass...
Certainly, but hard to combine with level control.
Not at all. If your level control uses a similar structure to that in your pre-amp from WW then it is easy to add the cut/boost.
Note, I see NON of the settings (including level) adjustable by potentiometer, but all using fixed resistors. This way we can "design in" everything.
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.
I would not use either one of these IC's. Take my advise, give people singles, so they have the complete choice and you will sell more units.
Adding an SMD footprint below a DIP IC is trivial, I do it all the time, again, it gives more flexibility to your customers and thus you will sell more.
But how would the crossover know what was going on in the power amp?
Well, two ways. The so called processor PA's of the 1980's usually used sense-lines.
My own application used instead simple "predictive mode", iy required setting the power amplifiers to a fixed gain and using power amplifiers with known clipping behaviour.
The thermal limiter used a true RMS chip with a very long hold time and was set so that output of the poweramp would not exceed the RMS power handling of the attached driver.
The clipping limiter was a simple and fast peak limiter, set to limit at the rated power of the Amp.
The excursion limiter for the bass used a filter to boost the low frequencies in a pattern analogue to the excursion in the (vented) box (makes for interesting filter design) and used again fast peak limiting.
The way I had it implemented I had one optocoupler with optoresistor in the peaking 2nd order highpass which would first remove the boost and eventually silde the LF cutoff upwards dynamically and after that range was "used up" it would start applying the peak limiter which also used an optocoupler. The Thermal limiter would adjust the peak limiter treshold if the RMS power was exceeded.
That was all done using discretes and op-amp's in the 1980's. It gave us the ability to run much bigger Amp's and get a lot more SPL out of our touring stacks, while at the same time the number of driver repairs we had fell extremely...
Coincidentally the same kind of behaviour is programmable into all modern Pro-Sound digital crossovers.
As someone who has used active PA systems (and other speakers) for quite a few years I would NOT use any active system without some form of protection for the individual drivers. It is too easy to blow up drivers otherwise.
Ciao T
Hi Douglas
The following answer to a question by thorten made me scratch my head:
Regards
Charles
The following answer to a question by thorten made me scratch my head:
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.
Regards
Charles
Hi,
Simple subtractive filters derived from a sallen key high- or lowpass and subtractor will ove only a 6dB slope for the derived slope. See here:
Subtractive Crossover Networks
This BTW is not necessarily a disaster. Such systems can work quite well. For example if using a wideband midrange and ribbon treble unit having a 4th order HPF on the ribbon and a subtractively derived LPF on the midrange may work in a most excellent manner and fashion.
Ciao T
The following answer to a question by thorten made me scratch my head:
Simple subtractive filters derived from a sallen key high- or lowpass and subtractor will ove only a 6dB slope for the derived slope. See here:
Subtractive Crossover Networks
This BTW is not necessarily a disaster. Such systems can work quite well. For example if using a wideband midrange and ribbon treble unit having a 4th order HPF on the ribbon and a subtractively derived LPF on the midrange may work in a most excellent manner and fashion.
Ciao T
Hi Douglas,
I found the reference I was looking for...
A FOURTH-ORDER STATE VARIABLE FILTER FOR LINKWITZ-RILEY ACTIVE CROSSOVER DESIGNS - DENNIS A. BOHN AES 10/83
The key advantage of this circuit over more traditional ones (sallen/key) is the relative robustness to component tolerances (capacitors) and te use of only four identical value capacitors for the whole crossover. It seems to have no appreciable drawbacks over sallen-key filters either, at least on cursory inspection.
This topology is now essentially 'bog standard" in pro-audio.
Ciao T
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.
I found the reference I was looking for...
A FOURTH-ORDER STATE VARIABLE FILTER FOR LINKWITZ-RILEY ACTIVE CROSSOVER DESIGNS - DENNIS A. BOHN AES 10/83
The key advantage of this circuit over more traditional ones (sallen/key) is the relative robustness to component tolerances (capacitors) and te use of only four identical value capacitors for the whole crossover. It seems to have no appreciable drawbacks over sallen-key filters either, at least on cursory inspection.
This topology is now essentially 'bog standard" in pro-audio.
Ciao T
Hi Thorsten
I am aware of the fact with the simple subtractive crossovers - but I am also aware of the paper that you pointed to.
Furthermore it is possible to build subtractive crossovers that are at least second order on both outputs. Basically they can be made with any desired order, symmetrical or asymmetrical. Though very high orders don't have advantages only.
I hope Douglas is aware of these thingsas well !!!!
I have used an asymmetrical one (2nd 0oder acoustical on the MSW, 1st order on the woofer) with great success on my Manger. OTOH I will go for a 2nd/2nd order one as soon as time permits.
Regards
Charles
I am aware of the fact with the simple subtractive crossovers - but I am also aware of the paper that you pointed to.
Furthermore it is possible to build subtractive crossovers that are at least second order on both outputs. Basically they can be made with any desired order, symmetrical or asymmetrical. Though very high orders don't have advantages only.
I hope Douglas is aware of these thingsas well !!!!
I have used an asymmetrical one (2nd 0oder acoustical on the MSW, 1st order on the woofer) with great success on my Manger. OTOH I will go for a 2nd/2nd order one as soon as time permits.
Regards
Charles
Hi,
Well, one would expect Douglas to know, after all, he has just written "THE BOOK on analogue active crossovers".
As for very steep crossovers, I personally have found I prefer either extremely low order crossovers (zero or first), or extremely steep (elliptical/NTM) ones.
A speaker I designed with 10th order passive filters was just reviewed by Paul Messenger in HiFI+. He noticed just how seamless the transition between woofer and tweeter was.
He was moderately surprised, given that I used a 1.3KHz crossover (to most "smack bang in the middle of the most sensitive range of the ear") with very steep slopes (supposedly a "bad thing") AND a large diameter cone woofer married to a waveguide in a 2-way.
Originally the speakjer had a 2KHz crossover with 4th order LR slopes. It sounded very similar in the nearfield, but was horribly "phasey" in certain rooms. It was the super steep filter that made the "seamless" transition possible... 😛
I find in reality most ACOUSTIC slopes are asymmetrical anyway, only the electrical filter functions appear symmetrical.
Ciao T
I am aware of the fact with the simple subtractive crossovers - but I am also aware of the paper that you pointed to. Furthermore it is possible to build subtractive crossovers that are at least second order on both outputs. Basically they can be made with any desired order, symmetrical or asymmetrical. Though very high orders don't have advantages only.
I hope Douglas is aware of these thingsas well !!!!
Well, one would expect Douglas to know, after all, he has just written "THE BOOK on analogue active crossovers".
As for very steep crossovers, I personally have found I prefer either extremely low order crossovers (zero or first), or extremely steep (elliptical/NTM) ones.
A speaker I designed with 10th order passive filters was just reviewed by Paul Messenger in HiFI+. He noticed just how seamless the transition between woofer and tweeter was.
He was moderately surprised, given that I used a 1.3KHz crossover (to most "smack bang in the middle of the most sensitive range of the ear") with very steep slopes (supposedly a "bad thing") AND a large diameter cone woofer married to a waveguide in a 2-way.
Originally the speakjer had a 2KHz crossover with 4th order LR slopes. It sounded very similar in the nearfield, but was horribly "phasey" in certain rooms. It was the super steep filter that made the "seamless" transition possible... 😛
I have used an asymmetrical one (2nd 0oder acoustical on the MSW, 1st order on the woofer) with great success on my Manger. OTOH I will go for a 2nd/2nd order one as soon as time permits.
I find in reality most ACOUSTIC slopes are asymmetrical anyway, only the electrical filter functions appear symmetrical.
Ciao T
There is one advantage of state-variable crossovers that is talked about very little: By using more than just the high- and lowpass outputs one can include the driver's natural response in a very elegant way (i.e. they have an "intrinsic Linkitz Transform").
Regards
Charles
Regards
Charles
Note, I see NON of the settings (including level) adjustable by potentiometer, but all using fixed resistors. This way we can "design in" everything.
I can understand the thoughts behind this attitude, but I think it is a bit short sighted. I can see a lot of people not having a perfect design made and actually want some adjust-ability, especially with regards to the level. Also if all that's available is fixed resistors you'd probably end up destroying or damaging the PCB with all the changing in and out of different values as you experiment.
I can also see the board being more popular the more adjustable it is. Of course you can replace the pots with fixed resistors if you want to.
My boards use nothing but trim pots and purist you may not call it, but the reason it's adjustable is what's made it useful.
I can understand the thoughts behind this attitude, but I think it is a bit short sighted. I can see a lot of people not having a perfect design made and actually want some adjust-ability, especially with regards to the level. Also if all that's available is fixed resistors you'd probably end up destroying or damaging the PCB with all the changing in and out of different values as you experiment.
I can also see the board being more popular the more adjustable it is. Of course you can replace the pots with fixed resistors if you want to.
My boards use nothing but trim pots and purist you may not call it, but the reason it's adjustable is what's made it useful.
I'll second this - the above reasons are why I'm getting ready to switch to DSP. Nothing slows down the desire to try new things more than the thought of having to extricate and replace caps and resistors from a tight layout.
And the design should accept balanced and unbalanced input also
I think probably just a balanced input, as it can still be used to cancel out ground noise when fed from an unbalanced source. The problem is to design a low-noise balanced input stage- a standard 4 x 10K differential amplifier is very noisy. That's covered in the book.
Dear Douglas,
My own application used instead simple "predictive mode", iy required setting the power amplifiers to a fixed gain and using power amplifiers with known clipping behaviour.
The thermal limiter used a true RMS chip with a very long hold time and was set so that output of the poweramp would not exceed the RMS power handling of the attached driver.
The clipping limiter was a simple and fast peak limiter, set to limit at the rated power of the Amp.
The excursion limiter for the bass used a filter to boost the low frequencies in a pattern analogue to the excursion in the (vented) box (makes for interesting filter design) and used again fast peak limiting.
The way I had it implemented I had one optocoupler with optoresistor in the peaking 2nd order highpass which would first remove the boost and eventually slide the LF cutoff upwards dynamically and after that range was "used up" it would start applying the peak limiter which also used an optocoupler. The Thermal limiter would adjust the peak limiter treshold if the RMS power was exceeded.
That sounds like a thoroughly ingenious system, and I take my hat off to it. The only problem is that it is closely tailored to just one loudspeaker system. Changing the parameters for a wholly different loudspeaker setup must be quite a business?
I'll second this - the above reasons are why I'm getting ready to switch to DSP. Nothing slows down the desire to try new things more than the thought of having to extricate and replace caps and resistors from a tight layout.
You could always try Soundeasy or LspCAD; they have crossover emulators.
Hi,
This is DIY Audio, not BYO (buy yur own) Audio.
Level adjust ability with a generic textbook crossover is completely useless. Given that slopes, equalisation etc. need to be carefully tailored to a given set of drivers level is your least worry.
If you want easily and infinity adjustable, get a digital crossover (I said that before, didn't I?).
First, you could temporary wire in variables if you need to, secondly, a well manufactured PCB combined with adequate soldering skills (mine are barely adequate) go a long way. Of course, if you buy those ultra-cheap chinese PCB's with halve ounce copper (actually, I'd estimate quarter ounce) and crappy through plating, you may be in trouble changing anything.
That is why all my PCB's specify 2 ounce copper and I only use honest and capable vendors. Of course, it does cost money.
Maybe, but it will never be 1/10th as adjustable as a cheap commodity level pro-audio digital crossover. So if you are going to terminally cripple adjustability anyway, you might as well minimise performance bottlenecks. Otherwise what is the point doing it analogue?
Ciao T
I can understand the thoughts behind this attitude, but I think it is a bit short sighted. I can see a lot of people not having a perfect design made and actually want some adjust-ability, especially with regards to the level.
This is DIY Audio, not BYO (buy yur own) Audio.
Level adjust ability with a generic textbook crossover is completely useless. Given that slopes, equalisation etc. need to be carefully tailored to a given set of drivers level is your least worry.
If you want easily and infinity adjustable, get a digital crossover (I said that before, didn't I?).
Also if all that's available is fixed resistors you'd probably end up destroying or damaging the PCB with all the changing in and out of different values as you experiment.
First, you could temporary wire in variables if you need to, secondly, a well manufactured PCB combined with adequate soldering skills (mine are barely adequate) go a long way. Of course, if you buy those ultra-cheap chinese PCB's with halve ounce copper (actually, I'd estimate quarter ounce) and crappy through plating, you may be in trouble changing anything.
That is why all my PCB's specify 2 ounce copper and I only use honest and capable vendors. Of course, it does cost money.
I can also see the board being more popular the more adjustable it is. Of course you can replace the pots with fixed resistors if you want to.
Maybe, but it will never be 1/10th as adjustable as a cheap commodity level pro-audio digital crossover. So if you are going to terminally cripple adjustability anyway, you might as well minimise performance bottlenecks. Otherwise what is the point doing it analogue?
Ciao T
AFAIK you can also do this with some SPICE derived demo software but I can't remember which one (Switcher CAD ??).
Regards
Charles
Edit: Was waaaaay too slow in typing - I am refering to post 172 of course !!
Regards
Charles
Edit: Was waaaaay too slow in typing - I am refering to post 172 of course !!
Last edited:
Hi,
Nothing original or unique by me I'm afraid, jsut adapted from what I could find out about the state of the art in the western world appx. 1987 and realise with the parts I could get in eastern europe. Really ancient history.
In a completely analogue system it is quite a chore, yes.
But to be honest, if we are not prepared to apply heavy duty equalisation specifically tailored to a given speaker system and room any adjustable textbook 4th order LR crossover with adjustable levels is completely useless.
Setting the parameters is not that complex actually. Programming them into any appx. mid 90's digital crossover was not anywhere near that chore and you could save multiple settings for multiple settings.
Of course, that was a decade and a halve ago and things have advanced substantially since.
Ciao T
That sounds like a thoroughly ingenious system, and I take my hat off to it.
Nothing original or unique by me I'm afraid, jsut adapted from what I could find out about the state of the art in the western world appx. 1987 and realise with the parts I could get in eastern europe. Really ancient history.
The only problem is that it is closely tailored to just one loudspeaker system. Changing the parameters for a wholly different loudspeaker setup must be quite a business?
In a completely analogue system it is quite a chore, yes.
But to be honest, if we are not prepared to apply heavy duty equalisation specifically tailored to a given speaker system and room any adjustable textbook 4th order LR crossover with adjustable levels is completely useless.
Setting the parameters is not that complex actually. Programming them into any appx. mid 90's digital crossover was not anywhere near that chore and you could save multiple settings for multiple settings.
Of course, that was a decade and a halve ago and things have advanced substantially since.
Ciao T
You could always try Soundeasy or LspCAD; they have crossover emulators.
I sim everything in LTspice, but that doesn't tell me what a stereo pair of multi-way dipoles are going to sound like in a real room.
I sim everything in LTspice, but that doesn't tell me what a stereo pair of multi-way dipoles are going to sound like in a real room.
In contrast to what a simulator does, an emulator implements the design, functionally. IOW, it allows you to hear how the crossover will sound. Uses your PC platform plus multi-channel sound card. I'm only familiar with the SoundEasy product, but I'm sure LspCAD works the same way: passive and active crossovers alike can be emulated. SE allows the user to play .wav files or otherwise it processes the 2 channel analogue input of the sound card. You can read about it (SE calls it their Digital Filter) on the Bodzio site Home Page .
Last edited:
Hi,
This is DIY Audio, not BYO (buy yur own) Audio.
Ciao T
I agree with everything you've said and it's the way it should be done. Trouble is I don't see most people doing it that way.
Also a highly variable analogue crossover that can handle the vast majority of three ways is a niche that I don't think has currently been filled?
Are we looking for a useful product that could possibly please a lot of people, or one that pleases only a few?
I mean we're the ones that do it the right way, yet funnily enough we're also the ones that don't need to buy the board ourselves, we've done it all before. But think back to the time where you didn't know all the answers and we're muddling through loudspeaker design and experimenting with different things, would what you're suggesting have appealed to you then?
I see those who buy this, as being those who don't want a DCX or a minidsp, that isn't the answer for them, otherwise they'd have one and not be buying this. Hence this would be the analogue equivalent, but maybe that isn't the design goal.
If the design goal is state of the art, with fixed value components that you arrive at, only having simulated heavily before hand and/or possibly emulated digitally first, then by all means, it won't be as appealing to the DIY crowd though.
So I guess the question is, what is it that Douglas wants to design? The possibility if doing both would be a great feature. Just like having pad placement for accepting both surface mount and through hole opamps. One could have part placement for fixed value resistors as well as trimpots - this would use more board space and as a result cost more though, but it would make for an more flexible product.
In contrast to what a simulator does, an emulator implements the design, functionally. IOW, it allows you to hear how the crossover will sound. Uses your PC platform plus multi-channel sound card. I'm only familiar with the SoundEasy product, but I'm sure LspCAD works the same way: passive and active crossovers alike can be emulated. SE allows the user to play .wav files or otherwise it processes the 2 channel analogue input of the sound card. You can read about it (SE calls it their Digital Filter) on the Bodzio site Home Page .
You are quite correct, that's how LspCAD does it too.
Hi,
You mean except for tens if not hundreds of commercial pro audio products, like Behringers Super X Pro CX3400 which retails around 140 USD?
What would be the point of making another one like that.
In the light of units like it and such as digital crossovers ANY product aimed at DIY has to be either way cheaper to be made (if you add the parts up even without the work it's a tall order) than such a commercial "fully adjustable crossover", or it has to aim at the highest percieved quality with maximum flexibility, in my view at least.
Now maximum quality is non too compatible with freely adjustable, especially as the slopes increase in steepness, unless we go digital.
Ciao T
Also a highly variable analogue crossover that can handle the vast majority of three ways is a niche that I don't think has currently been filled?
You mean except for tens if not hundreds of commercial pro audio products, like Behringers Super X Pro CX3400 which retails around 140 USD?
What would be the point of making another one like that.
In the light of units like it and such as digital crossovers ANY product aimed at DIY has to be either way cheaper to be made (if you add the parts up even without the work it's a tall order) than such a commercial "fully adjustable crossover", or it has to aim at the highest percieved quality with maximum flexibility, in my view at least.
Now maximum quality is non too compatible with freely adjustable, especially as the slopes increase in steepness, unless we go digital.
Ciao T
- Status
- Not open for further replies.
- Home
- Source & Line
- Analog Line Level
- The design of active crossovers- Douglas Self wants your opinions