Hi,
It looks like one of the terminals enters one side of the inductor. The other side of the inductor connects to a resistor and capacitor that are in series with one another. The end of the cap+resistor connects to the other terminal. If the mid/bass were connected to the non terminal connected side of the inductor and the terminal that the cap/resistor string connects to then you've got a standard 2nd order filter with damping resistor in series with the shunt cap. Or you've got a first order filter and a zobel.
To the left hand side of the terminal entry points you've got a sand cast power resistor and a small film cap.
The signal enters the small film cap then feeds directly into one of the terminals of what is most likely a variable resistor. The other side of the variable resistor connects to the sand cast resistor, which then connects/shunts, to the other terminal. If the tweeter were to be connected between the 'the other terminal' and the 'other side of the variable resistor' then you've got a 1st order electrical filter with an oddly configured variable L-pad.
This does not look like a series crossover to me.
It looks like one of the terminals enters one side of the inductor. The other side of the inductor connects to a resistor and capacitor that are in series with one another. The end of the cap+resistor connects to the other terminal. If the mid/bass were connected to the non terminal connected side of the inductor and the terminal that the cap/resistor string connects to then you've got a standard 2nd order filter with damping resistor in series with the shunt cap. Or you've got a first order filter and a zobel.
To the left hand side of the terminal entry points you've got a sand cast power resistor and a small film cap.
The signal enters the small film cap then feeds directly into one of the terminals of what is most likely a variable resistor. The other side of the variable resistor connects to the sand cast resistor, which then connects/shunts, to the other terminal. If the tweeter were to be connected between the 'the other terminal' and the 'other side of the variable resistor' then you've got a 1st order electrical filter with an oddly configured variable L-pad.
This does not look like a series crossover to me.
I just stumbled across this thread and the debate within and don't have a strong oppinion but would toss in a few comments.
The first system I designed over 30 years ago for a college senior project used a series crossover because the subtractive nature looked like it might offer better performance and I didn't kow enough about the fallacy of that oversimplification. I never tried a series crossover again.
I think the ESP reference on the subject covers matters pretty well. It seems that you can, for the most part, achieve a similar result both with parallel and series crossovers. What I don't like is interaction by section: in other words if I get the woofer and its crossover right and start adjusting the tweeter and its crossover, which shifts around the woofer and its crossover. With this interaction, how do I optimize a system?
The same would apply to high output impedance drive. Adjusting one section will impact the next. Optimization becomes very complex due to the interactions. Only through parallel crossovers and low impedance drive do we have independence between sections.
If the arguement is to use computer optimization, then it might be no more difficult. I have used Xopt for years and then Leap crossover shop, so I have no problem with using computer optimization schemes. I still tend not to use them with simple networks. There is a fair amount of setup required for computer optimization: impedance curves, careful driver measurements, taill correction etc. On the other hand, I can knock out a simple "2nd order and 2nd order" for any 2 way in less than an hour (getting to values that work and give a reasonable blend, I may then spend weeks finessing the balance a dB here and a dB there). Why go down a path that forces you to use computer optimization even if you might otherwise forego it for a simple system?
So if series connected networks make life more difficult and, ultimately, don't offer any real performance improvements, what am I missing about their appeal?
David S.
The first system I designed over 30 years ago for a college senior project used a series crossover because the subtractive nature looked like it might offer better performance and I didn't kow enough about the fallacy of that oversimplification. I never tried a series crossover again.
I think the ESP reference on the subject covers matters pretty well. It seems that you can, for the most part, achieve a similar result both with parallel and series crossovers. What I don't like is interaction by section: in other words if I get the woofer and its crossover right and start adjusting the tweeter and its crossover, which shifts around the woofer and its crossover. With this interaction, how do I optimize a system?
The same would apply to high output impedance drive. Adjusting one section will impact the next. Optimization becomes very complex due to the interactions. Only through parallel crossovers and low impedance drive do we have independence between sections.
If the arguement is to use computer optimization, then it might be no more difficult. I have used Xopt for years and then Leap crossover shop, so I have no problem with using computer optimization schemes. I still tend not to use them with simple networks. There is a fair amount of setup required for computer optimization: impedance curves, careful driver measurements, taill correction etc. On the other hand, I can knock out a simple "2nd order and 2nd order" for any 2 way in less than an hour (getting to values that work and give a reasonable blend, I may then spend weeks finessing the balance a dB here and a dB there). Why go down a path that forces you to use computer optimization even if you might otherwise forego it for a simple system?
So if series connected networks make life more difficult and, ultimately, don't offer any real performance improvements, what am I missing about their appeal?
David S.
Hmm, I dunno. Looks like the WLM tweeter is drawing it's drive from the voltage across the coil. Which is quite series really. But I wouldn't bet my life on what is going on there. Might be a bit hybrid series/parallel. Not quite enough to go on there. 🙁
Hey maybe we should respect WLM anyway. It's a nice creative design, for sure. Very Acoustic Research with that variable pad too. 😎
Hey maybe we should respect WLM anyway. It's a nice creative design, for sure. Very Acoustic Research with that variable pad too. 😎
Last edited:
I heard the La Scala's for an extended listening session at my friend's house with all his high end gear and they do sound good. He favors vinyl and tube gear which may be one reason he like the 93 dB/watt WLM's.
The La Scala can't be 93DB though. That applies to the Eminence coaxial/horn based units which are an entirely different animal. 🙂
I've just redrawn a coil and a capacitor and a wirewound plus the AR-style tweeter pot, and I can see a simple series crossover in there with a resistance across the bass rather than a capacitor. Not quite sure what the thing with the black shroud is though. Could be a little coil across the tweeter...
Thing is the Visaton cone tweeter is much more efficient than the 8" Visaton bass, so you can do attenuation. WLM are cool! 😎
I've just redrawn a coil and a capacitor and a wirewound plus the AR-style tweeter pot, and I can see a simple series crossover in there with a resistance across the bass rather than a capacitor. Not quite sure what the thing with the black shroud is though. Could be a little coil across the tweeter...
An externally hosted image should be here but it was not working when we last tested it.
Thing is the Visaton cone tweeter is much more efficient than the 8" Visaton bass, so you can do attenuation. WLM are cool! 😎
Here's a response I got from Troels Graveson.... (1 down, 1 to go)
"I regard speaker building as a life-long education and sometimes I make series filters to hear what they can do. Sometimes I find them better, sometimes not.
I don’t think it’s a matter of series vs. parallel as such, rather that a combination of drivers and crossover may incidently perform well with either topology.
Actually I have stopped making series filters because a lot of people don’t understand them and can’t find out how to wire it correctly ( = lots of mails!).
best regards
Troels"
Yes, the La Scala is 93 dB efficient. Visit the WLM web site for the specs.
Here is a schematic for the infamous AR3a loudspeaker. The mid/tweeter is series wired.
Attachments
Yes, parallel filter. 1st order with L-PAD for the tweeter. "2nd" order electrical for the woofer. The small size of the C (after big L), also a series 22R suggest that it is a zobel, so a 1st order LPF.
EDIT: Ups The C is not small at all actually (I thought it was a MKP).
Last edited:
I'm glad I'm not the only one who sees that. The second pic with the wiring helps to confirm it though.
The signal enters the red terminal and goes straight into the inductor. The other side of the inductor connects directly to the thick red wire which I am assuming is going to the mid/bass. The black wire from the mid/bass then connects directly to the black terminal. Even without any extra components we've got a first order parallel filter. The cap is obviously connected in a shunt position and from the second picture it looks like the resistor might have been omitted, creating a second order parallel filter, but as we cannot see around the other side of the capacitor that clearly it's hard to tell what could be there.
And, here's a response I got from Tony Gee. Not quite the same as Troels.....
"
I first built series crossovers purely out of curiosity. The idea being that parallel crossovers have the "fault" that the music signal is chopped into seperate parts and then stuck back together again. With a series crossover all is connected together, nothing "gets lost". For example you can't bi-wire a series crossover. A simplistic and naïve thought but enough to get me into exploring series crossovers. I also like to build them because they are more of a challenge to get right due to all the interaction that is going on.
Series crossovers have a major downside: due to the fact that all parts are interconnected, they must all be of high quality. This makes them relatively expensive. For example the coil that is parallel to the tweeter can be a cheap, thin wire type in a standard parallel crossover. In a series crossover this inductor is in series with the woofer and therefore determines the quality of the bass and midange (in a 2-way system).
Also not all drivers are suited for series crossovers, "easy" drivers that don't need much correcting are the best, again because the component quality in correction networks is very important with a series crossover. Also correcting on driver can influence the other driver's output as well, where as with a parallel network the individual sections don't influence each other.
Subjectively speaking, I find series crossovers to sound more coherent, spatial and dynamic. I don't know why this is.
regards,
Tony"
In thinking about Tony's last comment, I wonder if his liking for series xo's may be due in large part to the high quality drivers his uses with them as he pointed out in the previous paragraph.
The bottom line here for me, at least, is series xo's are similar in performance to parallel types, but do require high quality drivers to be successfully implemented. I think we can all agree there is no one perfect xo topology and that each must be matched to the drivers chosen and the general geometry of the driver placement on a BB. Interactions between drivers when implementing series xo's can present a complex problem. I've built both types recently in a high end 2-way with the aid of a virtual crossover box to impirically determine the component values. One speaker had series xo and the other a quasi-parallel one with no cap ala the Sonus Faber Extrema type. Each had a minimalist xo in the and. I ended up going with the SFE type only because it was better suited to my listening room acoustics because it had a BBC dip vs the flat response of the series xo.
Very nice views on series from Mr. Troels Gravesen and Mr. Tony Gee there, speakerdoctor. Nice work. 🙂
Few loose ends to tie up. Jay has it right on his WLM La Scala analysis, I reckon. This is what I get roughly... PARALLEL:
Seems to make sense...😎
That AR-3A crossover is parallel for sure if you re-examine it. And 93dB SPL seems a bit optimistic for WLM considering the Visaton 8" bass unit is around 88dB at 1W. But let's not nitpick. The very interesting topic is series crossovers.
Visaton - Lautsprecher und Zubehör, Loudspeakers and Accessories
Getting back to series, I see elegance lurking in this Tony Gee design if you make the RC element of the input of a simple series filter equal to the Zobel network for the bass and simplify dramatically:
That has gotta be worth trying at some level.
Few loose ends to tie up. Jay has it right on his WLM La Scala analysis, I reckon. This is what I get roughly... PARALLEL:
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
Seems to make sense...😎
That AR-3A crossover is parallel for sure if you re-examine it. And 93dB SPL seems a bit optimistic for WLM considering the Visaton 8" bass unit is around 88dB at 1W. But let's not nitpick. The very interesting topic is series crossovers.
Visaton - Lautsprecher und Zubehör, Loudspeakers and Accessories
Getting back to series, I see elegance lurking in this Tony Gee design if you make the RC element of the input of a simple series filter equal to the Zobel network for the bass and simplify dramatically:
An externally hosted image should be here but it was not working when we last tested it.
That has gotta be worth trying at some level.
Hi,
Sorry I cannot see that being the case. You can't zobel the bass/mid
from the wrong side of L3 and 10R at the input would over attenuate
the tweeter.
You can say the bass mid zobel is fairly redundant when you
consider the relative impedance of C3 to it at all frequencies.
It doesn't do much at all, removing it makes little difference.
rgds, sreten.
The white thin wire from the tweeter clearly connects to the black terminal on the binding posts.
Well, I did say GUESS and ROUGHLY....😀
An externally hosted image should be here but it was not working when we last tested it.
What about THIS?
An externally hosted image should be here but it was not working when we last tested it.
System7, how were you able to get all the xo values for the La Scala to show on your schematic?
I hope that was not in reference to me. I didn't design any other xover for it, just the AR and by ear. If you think I'm an inexperienced builder, you need to look at my photobucket pages....
Here's the FR plot (though there was some HVAC interference in the rooom where it was taken):
...and the measured impedance:
Now- how's that look for designed by ear?
Later,
Wolf
It's been confirmed by my WLM-owner friend (who asked the U.S. distributer) that the La Scala topology is parallel.
- Status
- Not open for further replies.