You only need to have one component: a pre-amp with balanced outputs!
Anyway, I think that this is a good capability to offer, but if there is no demand and adds to the cost, then there isn't much motivation for me to make a PCB just for balanced I/O. So if people want that, I need to hear from them.
-Charlie
Or a balanced line driver, like I will probably use for my active surrounds. I'll try it single ended first, but it seems like 50' runs will want to be balanced.
Time Delay / Soft-Start / Relay Board
Thinking about what other kind of circuitry might be useful, I thought that something that could prevent turn-on and turn-off thumps might be useful. After playing around with a few ideas, I realized that a general circuit that can implement turn-on and turn-off time delays via LM555 timers would be very useful for:
This would be a very useful circuit for a number of DIY applications where power amp and line level circuits operate together.
-Charlie
Thinking about what other kind of circuitry might be useful, I thought that something that could prevent turn-on and turn-off thumps might be useful. After playing around with a few ideas, I realized that a general circuit that can implement turn-on and turn-off time delays via LM555 timers would be very useful for:
* delaying the start-up of the power amps(s) while the line-level crossover circuitry PS stabilized
* delaying the turn-off of the line-level crossover circuitry while the power amp(s) shut down
* implementing soft start using a series power R that is switched out of the AC line after a few tenths of a second
* power amp mute
The circuit is activated (turned on and off) via a single pole AC switch. All switching is done via 12VDC coil "ice-cube" relays that are external to the circuit board, to keep AC away from the DC power as much as possible. Both the delay-off and delay-on times are adjustable for each timer via potentiometers. The present design has three independent timers, however the number of timers could be increased.* delaying the turn-off of the line-level crossover circuitry while the power amp(s) shut down
* implementing soft start using a series power R that is switched out of the AC line after a few tenths of a second
* power amp mute
This would be a very useful circuit for a number of DIY applications where power amp and line level circuits operate together.
-Charlie
update on modular (active) crossover boards...
Well, a month has passed me by and unfortunately not much has progressed with the boards. 😡 I sent off the PCB design for the modular crossover board with the HP/BP/LP filters before the Xmax holiday to a company that makes cheap prototypes in Bulgaria. Well that was a mistake, since I still haven't gotten any thing from them and I am out 60 Euros.
That's the last time I use them for anything. My next choice fab is shut down until next week, but the turn around time should be much faster, and I am hopeful that these will get off the ground soon. Here is an update on what it coming down the pipeline...
UNIVERSAL ACTIVE FILTER BOARD
In the mean time, I have come up with a "universal" active circuit. I call it universal because the same circuit footprint can be configured as any of the following:
The "universal" circuit can do all that, but it is not adjustable after you build it. The user would need to determine and source all the components for the design. As a result, these will be sold as a bare board only.
MODULAR ACTIVE FILTER BOARD
For something more user friendly, the modular crossover board has a fixed design that is adjustable for frequency and Q after it is built. So between these two products, I think that the bases will be covered nicely. See the overview pdf file at this link for more information.
OTHER CIRCUIT BOARDS IN DEVELOPMENT
In addition, the new plan is to offer the following:
All PCBs will have an identical footprint, so multiple boards can be stacked up using standoffs to make a compact package.
I will update this thread with new details as they come up. Please post any questions or comments and I will be happy to answer them. Thanks
-Charlie
Well, a month has passed me by and unfortunately not much has progressed with the boards. 😡 I sent off the PCB design for the modular crossover board with the HP/BP/LP filters before the Xmax holiday to a company that makes cheap prototypes in Bulgaria. Well that was a mistake, since I still haven't gotten any thing from them and I am out 60 Euros.
That's the last time I use them for anything. My next choice fab is shut down until next week, but the turn around time should be much faster, and I am hopeful that these will get off the ground soon. Here is an update on what it coming down the pipeline...UNIVERSAL ACTIVE FILTER BOARD
In the mean time, I have come up with a "universal" active circuit. I call it universal because the same circuit footprint can be configured as any of the following:
inverting and non-inverting gain stages
first order HP or LP filter stages
second order HP or LP filter stages (Sallen-Key)
second order filter stage plus notch, HP or LP
notch filter
first or second order all-pass filters
This means you can create just about any crossover function that exists with these boards, including elliptical filters, Chebyshev type II filters, Hardman filters, etc. first order HP or LP filter stages
second order HP or LP filter stages (Sallen-Key)
second order filter stage plus notch, HP or LP
notch filter
first or second order all-pass filters
The "universal" circuit can do all that, but it is not adjustable after you build it. The user would need to determine and source all the components for the design. As a result, these will be sold as a bare board only.
MODULAR ACTIVE FILTER BOARD
For something more user friendly, the modular crossover board has a fixed design that is adjustable for frequency and Q after it is built. So between these two products, I think that the bases will be covered nicely. See the overview pdf file at this link for more information.
OTHER CIRCUIT BOARDS IN DEVELOPMENT
In addition, the new plan is to offer the following:
regulated power supply kit
Equalizer board (use with BP output from the modular crossover board)
6dB/oct shelving filter board for baffle-step / open-baffle compensation
Biquadratic filter to implement a "Linkwitz transform" style equalizer
Some of these were previously lumped together on what I called the "main board", but I got many requests to have these available separately, so I decided to break each part out on to its own PCB.Equalizer board (use with BP output from the modular crossover board)
6dB/oct shelving filter board for baffle-step / open-baffle compensation
Biquadratic filter to implement a "Linkwitz transform" style equalizer
All PCBs will have an identical footprint, so multiple boards can be stacked up using standoffs to make a compact package.
I will update this thread with new details as they come up. Please post any questions or comments and I will be happy to answer them. Thanks
-Charlie
active crossover design
Check this link for proper way to implement active crossover :
http://www.auratronsystems.com
Then click "3PX8" "Design Philosophy".
Check this link for proper way to implement active crossover :
http://www.auratronsystems.com
Then click "3PX8" "Design Philosophy".
That's one philosophy. It seems designed to promote a specific product. Call me jaded, but it seems a case of choose a design to differentiate my product from others, then specify the performance characteristics that mine meets and theirs doesn't as essential.
I tend to follow the Linkwitz Lab - Loudspeaker Design philosophy. You may be surprised that Linkwitz considers most implementations of the LR filter flawed as well. That's why he includes several phase correction stages in his crossovers. Do they count as gain stages? The Orions wouldn't work nearly as well without them.
In and of itself, the electrical slope of a crossover matters not at all. Choose the right crossover point and drivers and you can meet a LR2 acoustic goal with first order electrical filters. I usually go with at least second order filters to reduce the tweeter excursion on the low end, so I can get a LR4 response with a third order electrical filter. Add a bit of phase correction and I am happy.
I've found physically aligning the drivers as suggested by auratron tends to cause as many problems as it solves. Reflections and diffraction of the tweeter output caused by the stepped or kinked baffle cause response irregularities. No matter what your method, perfect phase alignment only happens at one point unless the acoustic centers of the drivers are coincident. I'll take the larger delays and phase adjustment and a flat baffle thank you. I have yet to try a design with waveguides that should eliminate the relections, though.
I tend to follow the Linkwitz Lab - Loudspeaker Design philosophy. You may be surprised that Linkwitz considers most implementations of the LR filter flawed as well. That's why he includes several phase correction stages in his crossovers. Do they count as gain stages? The Orions wouldn't work nearly as well without them.
In and of itself, the electrical slope of a crossover matters not at all. Choose the right crossover point and drivers and you can meet a LR2 acoustic goal with first order electrical filters. I usually go with at least second order filters to reduce the tweeter excursion on the low end, so I can get a LR4 response with a third order electrical filter. Add a bit of phase correction and I am happy.
I've found physically aligning the drivers as suggested by auratron tends to cause as many problems as it solves. Reflections and diffraction of the tweeter output caused by the stepped or kinked baffle cause response irregularities. No matter what your method, perfect phase alignment only happens at one point unless the acoustic centers of the drivers are coincident. I'll take the larger delays and phase adjustment and a flat baffle thank you. I have yet to try a design with waveguides that should eliminate the relections, though.
I purchased an early version of their xover in the early '80s. A few years later, Linkwitz-Riley introduced their product and I had the chance to speak with the guy who designed the 3PBP to question their design philosophy. I spent several weekends in the engineering library studying all the technical papers they referenced and came to the same conclusion they did about proper xover design. (It helps to discard all the chaff if you are an electronics engineer.) When I conveyed to Linkwitz-Riley by correspondence the results of my research, they did admit the flaws you referenced above.
It appears Auratron did their research first, then designed the product.
I sold my first 3BPB in favor of the newer 3PX8 which had more features (LFEQ, finer gain steps). I have also implemented a time-aligned 4-way system and, after comparing it to all the other configurations, I am still amazed by the illusion of depth of soundstage and imagery. Over the years I have experimented with all of the various xover configurations / driver alignments and was never able to tolerate any of the lesser designs.
Phase-correction stages DO COUNT as additional active gain stages. Also, time-alignment means making the acoustic centers of the drivers coincident. I have never observed any baffle-induced response anomolies that were greater than the inherent ripples in even the best drivers.
It appears Auratron did their research first, then designed the product.
I sold my first 3BPB in favor of the newer 3PX8 which had more features (LFEQ, finer gain steps). I have also implemented a time-aligned 4-way system and, after comparing it to all the other configurations, I am still amazed by the illusion of depth of soundstage and imagery. Over the years I have experimented with all of the various xover configurations / driver alignments and was never able to tolerate any of the lesser designs.
Phase-correction stages DO COUNT as additional active gain stages. Also, time-alignment means making the acoustic centers of the drivers coincident. I have never observed any baffle-induced response anomolies that were greater than the inherent ripples in even the best drivers.
Time-alignment follow-up --
When I first implemented my time-aligned system, there were no PC's or computer-based analysis equipment, so I had to do it all just by listening. When MLSSA (DRA Laboratories - MLSSA Acoustical Measurement System) was introduced, I arranged to have my system analyzed, and, to our astonishment, I had correctly aligned all the drivers to within +/- 0.5 cm ! ! !
I believe that validates the importance of and audibility of the time-alignment concept.
When I first implemented my time-aligned system, there were no PC's or computer-based analysis equipment, so I had to do it all just by listening. When MLSSA (DRA Laboratories - MLSSA Acoustical Measurement System) was introduced, I arranged to have my system analyzed, and, to our astonishment, I had correctly aligned all the drivers to within +/- 0.5 cm ! ! !
I believe that validates the importance of and audibility of the time-alignment concept.
I'd had my fill of marketing hype and other propoganda today when I read their page, sorry.
Coincident to me means acoustic centers in the same place, which doesn't happen except sometimes in a coaxial driver. You can align drivers so that the acoustic centers are equidistant from a designated listening point. As you move up and down you lose the alignment, we just do our best to make it work in a normal listening position.
With a Focal Micro Utopia inspired design (tweeter straight ahead, woofer on a ~15 degree sloped baffle) I got fairly broad 2-3 dB peaks centered at frequencies whose wavelengths correspond to the distance from tweeter to baffle intersection and to the woofer surround with a bit of suckout in between. The tweeters measure flat on that region on a flat baffle.
I'm not so worried about an opamp in the signal path if its function is necessary. Enough to do the job and no more.
I guess we can agree to disagree on some aspects of "proper" implementation of an active crossover. I choose a different philosophy but we are both happy with our choices. Both philosophies can be realized with Charlie's intended design. Enjoy the music.
Coincident to me means acoustic centers in the same place, which doesn't happen except sometimes in a coaxial driver. You can align drivers so that the acoustic centers are equidistant from a designated listening point. As you move up and down you lose the alignment, we just do our best to make it work in a normal listening position.
With a Focal Micro Utopia inspired design (tweeter straight ahead, woofer on a ~15 degree sloped baffle) I got fairly broad 2-3 dB peaks centered at frequencies whose wavelengths correspond to the distance from tweeter to baffle intersection and to the woofer surround with a bit of suckout in between. The tweeters measure flat on that region on a flat baffle.
I'm not so worried about an opamp in the signal path if its function is necessary. Enough to do the job and no more.
I guess we can agree to disagree on some aspects of "proper" implementation of an active crossover. I choose a different philosophy but we are both happy with our choices. Both philosophies can be realized with Charlie's intended design. Enjoy the music.
I drew this up about 2 years ago but never got around to building it:
Also a matching high pass (which I didn't finish the layout for).
Uses dip switches and, as can be seen, a lot of resistors.
Also a matching high pass (which I didn't finish the layout for).
Uses dip switches and, as can be seen, a lot of resistors.
Ahh the good old switched selection.
I've never tried to do something with that many options but a cheaper way of doing selection is to use 'split circle' solder pads which you can bridge with the touch of a soldering iron. Works nicely in situations where you might want discrete fixed levels of attenuation, but still want in-situ adjustment.
I've never tried to do something with that many options but a cheaper way of doing selection is to use 'split circle' solder pads which you can bridge with the touch of a soldering iron. Works nicely in situations where you might want discrete fixed levels of attenuation, but still want in-situ adjustment.
The purpose for my original post to this thread was to offer to those who plan to experiment with the multi xover board a heads up about which configuration would most likely yield favorable results.
I find it strange and disheartening how you and others so readily accept flawed "propaganda and marketing hype" from one manufacturer while summarily dismissing a thoughtful and accurate analysis from another.
Have you actually built a multi-way system based on 1st or 3rd order Butterworth xover with time-aligned drivers? If so, did you fail to notice that as you move through the sound field, the soundstage remains most coherent compared to other implementations? If not, you (and anyone else reading this) owe it to yourselves to at least do the experiment. You will not be disappointed.
To CharlieLaub, my suggestions for options
Do you plan to offer 3rd order Butterworth HP & LP implemented in single stage option? An accurate analysis will show this is the optimum configuration, and it should be included as a point of reference for other less-than-optimum designs.
Do you plan to offer 3rd order Butterworth HP & LP implemented in single stage option? An accurate analysis will show this is the optimum configuration, and it should be included as a point of reference for other less-than-optimum designs.
Alex, everyone has different reactions to sound and different priorities. I apologized for reading the page you linked to with such a jaundiced eye. However, people can interpret the same data differently based on their own experiences and preferences.
The point of building the Focal inspired speakers I described earlier was to attempt physical time alignment of the drivers. Using Thuneau's Frequency Allocator I can fairly quickly switch between a phase corrected LR4 and a Butterworth alignments by loading saved presets. I don't find much difference in the sound stage. There is a difference, but I am not sure either is clearly better.
That could be due to the programs's automatic phase adjustment that attempts to make every filter transient perfect, that the notch required to kill the woofer's breakup mode pulls the response off the target for an octave or so, or something else completely. Phase alignment at crossover or lack thereof and smooth power response make more difference than about anything else to me.
Yes, this hobby is about experimenting and finding what works for your preferences. As you do, I encourage everyone to experiment. I was not in the best mood yesterday and I overreacted a bit to your stating flatly that your link's is the only valid way to design a crossover. I apologize for the tone.
Some people prefer a 5 watt no feedback Single Ended Triode amp, others prefer a 500W solid state amp with loads of feedback. Neither one is right for all and there are plenty of options in between. You seem to prefer lower order filters and their lower phase shift. I tend to prefer higher orders to keep breakup modes well attenuated and Tweeter excursion down with a low crossover point. If there was only one right way to do things, the world would be a dull place.
Now back to our regularly scheduled program.
Charlie - any progress to report?
As noted earlier, Charlie has proposed a board that allows odd order filters and adjustable Q second order blocks, so you can get Butterworth or any other slope that the user prefers. However, the state variable second order filters are comprised of three op amps. Alex, your Butterworth preference is starting to sound like a religion. Please no Holy wars here.
The point of building the Focal inspired speakers I described earlier was to attempt physical time alignment of the drivers. Using Thuneau's Frequency Allocator I can fairly quickly switch between a phase corrected LR4 and a Butterworth alignments by loading saved presets. I don't find much difference in the sound stage. There is a difference, but I am not sure either is clearly better.
That could be due to the programs's automatic phase adjustment that attempts to make every filter transient perfect, that the notch required to kill the woofer's breakup mode pulls the response off the target for an octave or so, or something else completely. Phase alignment at crossover or lack thereof and smooth power response make more difference than about anything else to me.
Yes, this hobby is about experimenting and finding what works for your preferences. As you do, I encourage everyone to experiment. I was not in the best mood yesterday and I overreacted a bit to your stating flatly that your link's is the only valid way to design a crossover. I apologize for the tone.
Some people prefer a 5 watt no feedback Single Ended Triode amp, others prefer a 500W solid state amp with loads of feedback. Neither one is right for all and there are plenty of options in between. You seem to prefer lower order filters and their lower phase shift. I tend to prefer higher orders to keep breakup modes well attenuated and Tweeter excursion down with a low crossover point. If there was only one right way to do things, the world would be a dull place.
Now back to our regularly scheduled program.
Charlie - any progress to report?
As noted earlier, Charlie has proposed a board that allows odd order filters and adjustable Q second order blocks, so you can get Butterworth or any other slope that the user prefers. However, the state variable second order filters are comprised of three op amps. Alex, your Butterworth preference is starting to sound like a religion. Please no Holy wars here.
Last edited:
Optimum crossover preference
If experimenting to find one's own preference is the goal, then "the other way to do it" shoud also be included as an option. I am amazed that almost everyone has solely embraced Linkwitz-Rily as the bible. Maybe my enthusiasm over "the other way" will inspire enough curiosity that others might discover what I have found. (Or at least be aware that there is another viable alternative.)
If experimenting to find one's own preference is the goal, then "the other way to do it" shoud also be included as an option. I am amazed that almost everyone has solely embraced Linkwitz-Rily as the bible. Maybe my enthusiasm over "the other way" will inspire enough curiosity that others might discover what I have found. (Or at least be aware that there is another viable alternative.)
Last edited:
RE: Third order Butterworth circuit board
Hi Alex,
Let me answer your question in a bit of a round-about fashion...
I prefer to conceptually envision filters for crossovers as being one of HP, LP, or all-pass. Building blocks made up from first and second order sections that have variable/adjustable corner frequency and Q can be cascaded to make any crossover order of Butterworth and Linkwitz-Riley, which are the most common used for loudspeakers. Also Bessel and Chebyshev type I can be implemented in this way. When circuits that allow for zeros (to create a notch) are also available, you can create Chebyshev type II and Elliptic class filters (of which there are many varieties). Each filter type has its own set of selectivity (e.g. how steep the cutoff is), on and off axis response, phase and group delay, and time-domain properties (e.g. step or impulse response) and within any one of these categories there are tradeoffs to be made when choosing a filter and filter order.
So my plan for now is to to start with a general state-variable-filter based second order filter board (SVFB) that is very flexible and adjustable after you build it, since I think that this will be most popular and easiest to implement. After that, I plan to offer the "universal" filter board (UFB) that can be used to construct pretty much every filter building block but is not adjustable in the same way as the SVF filters. The SVF board has 2 independent second order sections. I haven't finalized the design of the universal board, but there will be at least 2 sections on it as well, potentially as many as four. I described the UFB in more detail in this post.
Now, back to your question: To make a 3rd order Butterworth filter, you need one first order section and one second order section. You could do this using:
There are lots of options and configurations. I will describe how to construct crossovers in more details once I have the boards ready to sell, so that people can see the different approaches. It sounds complicated, but really it's not!
Let me know if you have other questions.
-Charlie
Hi Alex,
Let me answer your question in a bit of a round-about fashion...
I prefer to conceptually envision filters for crossovers as being one of HP, LP, or all-pass. Building blocks made up from first and second order sections that have variable/adjustable corner frequency and Q can be cascaded to make any crossover order of Butterworth and Linkwitz-Riley, which are the most common used for loudspeakers. Also Bessel and Chebyshev type I can be implemented in this way. When circuits that allow for zeros (to create a notch) are also available, you can create Chebyshev type II and Elliptic class filters (of which there are many varieties). Each filter type has its own set of selectivity (e.g. how steep the cutoff is), on and off axis response, phase and group delay, and time-domain properties (e.g. step or impulse response) and within any one of these categories there are tradeoffs to be made when choosing a filter and filter order.
So my plan for now is to to start with a general state-variable-filter based second order filter board (SVFB) that is very flexible and adjustable after you build it, since I think that this will be most popular and easiest to implement. After that, I plan to offer the "universal" filter board (UFB) that can be used to construct pretty much every filter building block but is not adjustable in the same way as the SVF filters. The SVF board has 2 independent second order sections. I haven't finalized the design of the universal board, but there will be at least 2 sections on it as well, potentially as many as four. I described the UFB in more detail in this post.
Now, back to your question: To make a 3rd order Butterworth filter, you need one first order section and one second order section. You could do this using:
OPTION #1: one universal board (UFB 2nd order section + UFB 1st order section)
It's important to note that I am talking about a filter only, which is basically one-half of a crossover (e.g. only the LP part) so you would need to construct the complimentary filter (e.g. the corresponding HP filter) to complete the crossover. If you use OPTION #2, each SVFB second order section has BOTH HP and LP outputs because the SVF circuit creates these simultaneously. IN OPTION #1 the UFB can be wired as Sallen-Key second order sections that many people are familiar with, but these only generate a single filter function (e.g. LP or HP) at a time. or
OPTION #2: one SVF board and one universal board (SVFB 2nd order section + UFB 1st order section)
There are lots of options and configurations. I will describe how to construct crossovers in more details once I have the boards ready to sell, so that people can see the different approaches. It sounds complicated, but really it's not!
Let me know if you have other questions.
-Charlie
Last edited:
Ok, Alex. You're new here. I'm glad you're enthusiastic. Your buying an active crossover in the 80s implies that you are old enough to know how to be diplomatic. Had you phrased your post as "The design philosophy expressed here works really well for me" and then avoided belittling anyone who chooses another route your ideas would have been better received.
Charlie's crossover board topology allows all sort of configurations, and to my knowledge nobody had designated it as a Linkwitz-Reilly only board. I tried to point out that sometimes you can get a net even order system response with an odd order electrical filter. This implies that if you use an even order filter you could get an odd order net response. ALL that matters is the net response of the system, not the electrical filters required to get there.
There are topologies that minimize the number of active components the signal must pass through, but most potential buyers of this board would rather do simpler calculations of separate blocks for each function. I think Charlie is wise to do it this way. Those with enough savvy to split the crossover points and adjust each section's response to get flat response and phase alignment in a single stage probably have their own boards. I think it is telling that Linkwitz himself uses a building block approach despite being obviously savvy enough to combine stages. Building blocks allow easier adjustment for experimenters like this board's target audience.
Net: "ABCD worked great for me, try it" Good. "ABCD is the only way to go and you're an idiot of you go another way" Bad.
Charlie's crossover board topology allows all sort of configurations, and to my knowledge nobody had designated it as a Linkwitz-Reilly only board. I tried to point out that sometimes you can get a net even order system response with an odd order electrical filter. This implies that if you use an even order filter you could get an odd order net response. ALL that matters is the net response of the system, not the electrical filters required to get there.
There are topologies that minimize the number of active components the signal must pass through, but most potential buyers of this board would rather do simpler calculations of separate blocks for each function. I think Charlie is wise to do it this way. Those with enough savvy to split the crossover points and adjust each section's response to get flat response and phase alignment in a single stage probably have their own boards. I think it is telling that Linkwitz himself uses a building block approach despite being obviously savvy enough to combine stages. Building blocks allow easier adjustment for experimenters like this board's target audience.
Net: "ABCD worked great for me, try it" Good. "ABCD is the only way to go and you're an idiot of you go another way" Bad.
A 3rd order filter of any type can be constructed by taking a single-stage 2nd order and adding 1 resistor and 1 capacitor. Every extra gain stage adds more distortion, so a single-stage implementation would be best.
- Status
- Not open for further replies.