Symmetrical Crossovers - Why? - Advantage/Disadvantage

[DaveC]

Magico also claims symmetric, elliptic xo topology so I'm guessing there is some advantage. I've seen the FR of an elliptic but how is it done? I can see it's use for hard coned drivers as the slope gets steeper and steeper as you move away from Fc.

 

[opus112]

Magico also claims symmetric, elliptic xo topology so I'm guessing there is some advantage. I've seen the FR of an elliptic but how is it done?

To do high slope elliptics calls for high Q inductors (low losses) and low ESR capacitors. The stop band is made up of zeroes in the transfer function - resonant LC circuits. You can find an example filter here : http://www.whatsbestforum.com/showthread.php?8858-Digital-that-sounds-like-analog&p=154399&viewfull=1#post154399

Note that this one's a line-level filter and its at a much higher frequency than a typical XO. To do it at speaker-level currents would require much larger components. Probably the inductors will need to be gapped ferrite to achieve low enough losses.

 

[DonH50]

Filters can get complicated (messy) quickly but fundamentally it is how you define the poles and zeros in the transfer function that determine if it is Butterworth, Chebyshev, elliptical, etc. Butterworth filters have no ripple in the passband or the stopband. Elliptical filters have ripple in the passband and stopband but provide steep rolloffs and less attenuation at the passband edge. Chebyshev puts ripples in the passband or stopband, not both, and you decide which band is appropriate for your application. Bessel filters are linear phase, meaning constant group delay, which means better pulse fidelity (impulse/time-domain response) than other types. Bessel and its variants provide linear phase without ripples but have low roll-off. And so forth and so on...

This Wikipedia link has some pictures that show what I am babbling about: https://en.wikipedia.org/wiki/Filter_(signal_processing)

Perhaps someday I could write an overview but I am working far too many hours lately (did the simulation whilst awaiting test results, in the lab, on Sunday, fourth in a row!) And I'd have to find my old grad textbooks, and I didn't really like the last one that much... And frankly I am waiting for the drama to die down.

Speaker crossovers use passive LC circuits as opus112 said so they can handle the high power. I have long espoused that an active design provides better overall sound but there is a lot of resistance to that idea. These days DSP provides excellent filter response but of course you must have a DAC in the circuit (and an ADC, someplace).

By the way, the simulation I showed did not use work tools (get in trouble for that), or even my personal simulator (for which I paid good money), but a free program called LTSpice available from the Linear Technology website. Anyone is welcome to download and play with it, using my files or whatever they choose.

 

[opus112]

elliptical filters are linear phase, meaning constant group delay, which means better pules fidelity (impulse/time-domain response) than other types.

Did you mean to write 'Bessel' filters here Don? Because elliptical filters have the worst ever phase response of any kind of filter I've seen. Not that its particularly important for audio.

I have long espoused that an active design provides better overall sound but there is a lot of resistance to that idea.

Its a very interesting topic - I think a lot of the resistance to active speakers comes from the fact that the filters aren't implemented to sound good. Rather they typically just use opamps in text-book style without concerns over their SQ. Plus the fact that active speakers are normally a single box and hence not obsolescence-proof in any sense.

 

[DonH50]

Good grief. Bessel, yes, should not post from memory at the end of a 90-hour workweek, thanks for catching that.

Bessel filters have maximally flat phase (delay) response, Butterworth have maximally flat frequency response, elliptic the steepest roll-off (slope).

Back when I started having to design filters image parameter design was still used quite a bit. Nowadays it's mostly DSP FIR and IIR filters.

Opamps can sound great but it takes a lot of care to select the right ones to use in a filter circuit, and of course it depends on the type of filter, and the filter design must be "sound" as well. Filter circuits impose wide impedance variations at various places in the filter and at different frequencies.

OK, off to bed, when I start writing puns it's been too long a day (was at work at 5:30 AM this morning, blah).

 

[LarsS]

Thanks all for your input, highly appreciated!

Some more information.

In one of the reviews, the reviewer managed to drive an amp (Marantz PMA-6005) into protection mode while testing the Gauder Akustik Arcona 100 (same as I have). Two other amps managed to drive speakers without problem (vacuum tube Westend Monaco and T-A PA 3000 HV).

I've managed to drive two amps into protection mode while playing extremely loud (see Disco loud) - Vincent/ShengYa A17CS (hybrid integrated amp 150W, 22 Kg) and my Norma Revo IPA-140 integrated amp (140W, 36 Amp cont & 150 Amp peak, 25 Kg).

/Best
Lars

 

[DonH50]

High-order filters can present a very reactive load to an amplifier, especially out of band. That is, rather than looking like a simple resistor, the speaker starts looking to the amplifier more like a pure capacitor or inductor. Amplifiers often do not like highly reactive loads; it can make them unstable, in the worst case oscillating and (hopefully) shutting down. The real worst case is when they do not shut down and ultrasonic oscillation destroys the tweeter (and perhaps other things).

 

[Folsom]

High-order filters can present a very reactive load to an amplifier, especially out of band. That is, rather than looking like a simple resistor, the speaker starts looking to the amplifier more like a pure capacitor or inductor. Amplifiers often do not like highly reactive loads; it can make them unstable, in the worst case oscillating and (hopefully) shutting down. The real worst case is when they do not shut down and ultrasonic oscillation destroys the tweeter (and perhaps other things).

Maybe those speaker manufacturers include dampening.

 

[LarsS]

Thanks all for your efforts in beating Google shedding light on Symmetrical Crossovers. Guess this is as far as we can get on this subject.

Quote from Dr Roland Gauder/Designer of speakers: " ... any amp will trip before my speakers do ... "

/Best
Lars

 

[cousinbilly]

I appear to be late. Sorry for reviving an old thread.

This will be a non-technical response.

Let's start with the basics:
A phono cartridge outputs a balanced signal, as does a reel to reel (tape). To keep it simple, a pure sinusoidal waveform is presented in an electrical state. Only, and I mean only, a dual differential (or true balanced) phono pre-amp will preserve the entire sinusoidal waveform. So far we are symmetrical. Imagine a sine wave.

As soon as the sine wave encounters an unbalanced electrical component (RCA interconnect), the 'falling' portion of the waveform is referenced to ground. This means the entire waveform is NOT preserved. We are now asymmetrical.

The same holds true for asymmetrical crossovers. The negative portion of the waveform is not preserved. This is not entirely true, but 'LarsS' in post #22 said it perfectly; the rear motion of the speaker (driver) is not preserved. I tried uploading relevant pictures, but no go.

Here's the problem; in order for the beauty of a symmetrical crossover to be truly realized, it must be driven by a symmetrical amplifier. Ideally the entire amplification chain will be balanced (dual differential). No one wants to hear this. No one wants to be told their unbalanced (single ended) amplifier is throwing 1/4 of the waveform away. Speaker manufacturers of symmetrical crossovers don't want to potentially lose customers because they don't have balanced amplifiers. Such is the world we live in.

If you own Gauder, Magico, or Evolution Acoustic speakers, do yourself a huuuuuuuuuuge favor, get hold of a dual differential amplifier, and listen. 50% of you will hear the difference immediately. Why 50%?, it has to do with 'spatial acuity', the way the brain interprets moving objects (sound waves are moving). Think of it this way; if you can parallel park, or if you never get honked at because you never cut drivers off, you have spatial acuity.

Sorry, the digital domain does not preserve the entire waveform.

I read somewhere; if you want to enjoy digital, never listen to good analogue.

 

[Speedskater]

Oh dear, oh dear, you should have let sleeping dogs lie. There are so, so many technical errors in your post, I don't know where to start (so I won't).

 

[DaveC]

Yeah, you're going to have to prove that's not all gibberish...

 

[Folsom]

I think that was meant to be posted in the joke/humour thread.

 

[cousinbilly]

Dang'it, wrong website!!!

Maybe someone with a Gauder or Magico speaker could humor me, get a dual differential amplifier, and then tell me how wrong I am.

Where are those meds????

 

[Folsom]

There's plenty of free information out there, and lots of great books. You're obviously interested in electronics. Right now you are just saying things that show you don't understand the basic fundamentals at all.

Magico's are elliptical, not symmetrical, btw.

So yes, you appear to be wrong about pretty much everything. No one is trying to be rude, but you are a middle school - Jr. High level trying to tell PHD's what's what...

 

[lscangus]

High-order filters can present a very reactive load to an amplifier, especially out of band. That is, rather than looking like a simple resistor, the speaker starts looking to the amplifier more like a pure capacitor or inductor. Amplifiers often do not like highly reactive loads; it can make them unstable, in the worst case oscillating and (hopefully) shutting down. The real worst case is when they do not shut down and ultrasonic oscillation destroys the tweeter (and perhaps other things).

Is there any way on the Amp side or Speaker side to solve this issue?

 

[DonH50]

Is there any way on the Amp side or Speaker side to solve this issue?

A number of ways, with a number of pro/con trades, but consumers do not usually need to worry about it... Note a speaker has an impedance over frequency dictated by the crossover, drivers, and cabinet design so does not "open up" though impedance can get very high or low at certainly frequencies. Most amplifiers state over what nominal impedance range they will tolerate and some state the lowest impedance they will tolerate. SS amps usually do not care if the load is high-impedance ("open") but tube amps can self-destruct without a load. Modern amps are designed to tolerate a wide range of loads and remain stable so this is rarely an issue these days. 40+ years ago, when I started messing with audio, it was a bigger deal.

To answer your question, some tube amps (or their owners) incorporate a high-value resistor, like 300~500 ohms, across the speaker terminals to keep it happy('ish) if the speaker is disconnected (e.g. intentionally to change a cable, by accident when a kid steps on the cable, or unintentionally when a speaker fuse blows, etc.) If you have a particularly nasty load that a SS amp does not like you can add a small series resistor (though that may cause other problems like changed frequency response and higher distortion) or a "snubber" circuit that is a resistor in series with a capacitor to damp high-frequency noise.

HTH - Don