Advantages and disadvantages of having the woofer-mid in a 2-way "Run Wild," with no Woofer Leg on the Crossover?

Jump to Latest
#42
markbakk said:
Anomalies such as metal cone resonances high up show as distortions on related lower frequencies. That is certainly clear on the measurement of the Fountek. But here one should question the validity and interpretation of the measurement method itself too.
Click to expand...
This doesn't only effect metal cones. It affects all types of cones. Paper cones have just as much breakup as metal cones, just that the breakup nodes have a low quality factor, i.e. the energy is spread over a wider range of frequencies rather than being a more pronounced narrowband spike. Untrained eyes will look at the datasheet of a driver like the SEAS CA18RNX and conclude that it has minimal or no cone breakup but they couldn't be more wrong.

If the CA18RNX had a 'resonance free cone' installed, the response would be rolling off already by 2kHz yet in the frequency response we see something that looks more or less flat out to 7-8kHz. Why is this then? Cone breakup is pushing the response up to flat from 2-8kHz. The inductance of the voicecoil is rolling off the response at the same rate the cone resonance is boosting it back up. Why is this bad then? Divide that range by 5 and you get 400Hz to 1.6kHz, then check out the 5th order harmonic in the plot I posted and what do you know, there's a broad increase of 5th order harmonics from about 400Hz to 1.6kHz. This is because any 5th order distortion that is created (by non-linearities of the magnetics, surround, spider) at those frequencies hits the breakup of the cone and therefore plays louder than it otherwise would with a truly resonance free cone.
Another indicator of cone breakup is to check out the impedance plot in the drivers datasheet and we see the impedance has doubled from the minimum of Re=~6ohm to 12ohms by about 2kHz. Therefore we'd expect the response to have fallen by 6dB at 2kHz but instead it hasn't fallen at all, therefore we can conclude the cone is breaking up by +6dB at 2kHz. If you don't understand why this is, we drive speakers with a voltage source so if the impedance has doubled the current has halved, and the pistonic movement of the cone is proportional to current through the voicecoil. Half the current is -6dB.

A revelation huh? You can look at just the frequency response and impedance plot in the datasheet and have a reasonable guess at how much higher the distortion will be at certain frequencies! This is a good trick to scoping out good drivers without having to buy and test them, and avoiding ones that aren't worth buying. If the impedance is going to the moon in the datasheet and the frequency response is not falling off a cliff in response, you can bet that the driver won't perform very well at least 2 octaves below where that happens.

Of course, what the frequency response and impedance in the datasheet can't tell you is how linear the magnetics/surround/spider are in the first place. If those parts are exceedingly linear then you don't care about cone breakup messing up the overall linearity of the driver since the harmonics won't be audible even being boosted up by breakup. Unless you are buying Satori/Scanspeak level drivers then this is seldom the case so knowing the above interpretation techniques will be a valuable skill to have in selecting drivers that deliver value for money.

Also since you doubt the validity of my measurement method, here's an article about that... http://feleppa.com.au/speakermeas.html
 
Last edited:
  • Like
Reactions: tktran303, cowanaudio, tmuikku and 3 others
#44
TMM said:
Almost every woofer or "mid-woofer" in existance has elevated distortion at higher frequencies. Don't let a smoothly falling frequency response lull you into thinking that you can run it as high as you want or avoid using a lowpass filter completely.

Consider the following frequency response. Looks pretty benign, right? No significant issues using this out to 6-7kHz?
KSgCKN6.png


Now consider the harmonic distortion of the same driver. There is elevated 3rd and 5th order distortion from 900Hz up to around 3kHz so you could not want to cross this woofer higher than 1kHz in order to maintain the lowest possible distortion
QunPNQJ.png


Another example with a less satisfactory looking frequency response. Should be ok as long as we can avoid that big peak starting at 5kHz, right?
kkoaTy8.png


Wrong. Distortion rises significantly starting as low as 1kHz.
gnxABPI.png
Click to expand...
"Almost every woofer or "mid-woofer" in existance has elevated distortion at higher frequencies."
almost
the key word here is almost.
 
#45
youknowyou said:
"Almost every woofer or "mid-woofer" in existance has elevated distortion at higher frequencies."
almost
the key word here is almost.
Click to expand...
If you don't measure, then you don't know. Measuring everything is very expensive. Understanding the data that we are given for free by the manufacturer is less expensive.

There are probably less than 10 manufacturers who make woofers where the above problem does not present an issue for non-linear distortion, and even then that does not apply to all of the woofers they make, just some of them. Generally if you see this behaviour, run away unless you can afford to purchase the driver and measure for yourself (and be prepared that it might suck), or there are already non-linear distortion measurements available for it online.
 
Last edited:
  • Like
Reactions: morbo
#49
TMM said:
This doesn't only effect metal cones. It affects all types of cones. Paper cones have just as much breakup as metal cones, just that the breakup nodes have a low quality factor, i.e. the energy is spread over a wider range of frequencies rather than being a more pronounced narrowband spike. Untrained eyes will look at the datasheet of a driver like the SEAS CA18RNX and conclude that it has minimal or no cone breakup but they couldn't be more wrong.

If the CA18RNX had a 'resonance free cone' installed, the response would be rolling off already by 2kHz yet in the frequency response we see something that looks more or less flat out to 7-8kHz. Why is this then? Cone breakup is pushing the response up to flat from 2-8kHz. The inductance of the voicecoil is rolling off the response at the same rate the cone resonance is boosting it back up. Why is this bad then? Divide that range by 5 and you get 400Hz to 1.6kHz, then check out the 5th order harmonic in the plot I posted and what do you know, there's a broad increase of 5th order harmonics from about 400Hz to 1.6kHz. This is because any 5th order distortion that is created (by non-linearities of the magnetics, surround, spider) at those frequencies hits the breakup of the cone and therefore plays louder than it otherwise would with a truly resonance free cone.
Another indicator of cone breakup is to check out the impedance plot in the drivers datasheet and we see the impedance has doubled from the minimum of Re=~6ohm to 12ohms by about 2kHz. Therefore we'd expect the response to have fallen by 6dB at 2kHz but instead it hasn't fallen at all, therefore we can conclude the cone is breaking up by +6dB at 2kHz. If you don't understand why this is, we drive speakers with a voltage source so if the impedance has doubled the current has halved, and the pistonic movement of the cone is proportional to current through the voicecoil. Half the current is -6dB.

A revelation huh? You can look at just the frequency response and impedance plot in the datasheet and have a reasonable guess at how much higher the distortion will be at certain frequencies! This is a good trick to scoping out good drivers without having to buy and test them, and avoiding ones that aren't worth buying. If the impedance is going to the moon in the datasheet and the frequency response is not falling off a cliff in response, you can bet that the driver won't perform very well at least 2 octaves below where that happens.

Of course, what the frequency response and impedance in the datasheet can't tell you is how linear the magnetics/surround/spider are in the first place. If those parts are exceedingly linear then you don't care about cone breakup messing up the overall linearity of the driver since the harmonics won't be audible even being boosted up by breakup. Unless you are buying Satori/Scanspeak level drivers then this is seldom the case so knowing the above interpretation techniques will be a valuable skill to have in selecting drivers that deliver value for money.

Also since you doubt the validity of my measurement method, here's an article about that... http://feleppa.com.au/speakermeas.html
Click to expand...
One other reason come speakers have a rising response is the rising radiation impedance.
 
#50
Let's resurrect this thread since I've got a related question.

I've got a Monacor SPH-145HQ and a ScanSpeak D2606-9220 and modelled with XSim a crossover at 2.5Khz and and other solution is without a crossover at approximately 8.5 Khz.

Both had acceptable phase shifts ( between + and - 15 degrees ) however the one having the woofer running without high frequency cutoff (I put in baffle correction) needed to have the tweeter's polarity reversed. The 2.5Khz solution had both drivers running with the same polarity connection.

In this case I wonder which of the two would be the better solution, personally I'm wondering about the off axis response at higher frequencies of the mid-woofer before the tweeter kicks in.

Looking forward to your input.

Thanks, John.
 

Attachments

#52
Thanks for the reply, it answers my dilemma, 2.5Khz it will be.

The drivers in the 2.5Khz crossover overlap for the whole region where both contribute to the sound. The 8.5Khz setup has a gap of approximately 90usec between them at crossover frequency.
 
#53
EL506 said:
Looking forward to your input.

Thanks, John.
Click to expand...
You give me far too much credit!

I am a "Big-Picture" guy and (I hope) a Team Builder. I farm out or always at least have a collaborator on crossover design. Winslow Burhoe, btw, the most recent time I saw him, referred to me as "The King of the Midrange," but, if you give that more than a moment's thought... that is not entirely a compliment, is it? (Meaning, this guy is clueless in the lower octaves.)

So, the answer I would have typed is, I always find it helpful to consult with the chart of "The Frequencies of Piano Keys." REALLY!!!

https://en.wikipedia.org/wiki/Piano_key_frequencies

For example, if my question is, "Cross over at 2.5k or 3.0k?," the chart tells me that that is the difference between D-sharp and F-sharp, that is, from one Black Key to the next higher Black Key. Three half-steps: D#-E; E-F; F-F#. No big deal.

So, to use my "crutch" on your problem, 8.5kHz is more than 200% the frequency of the 88th key on an 88-key piano (C8 = 4186Hz), and that just instinctively strikes me as not the territory for a woofer mid to be playing in.

But of course, the answer about Group Delay was Brilliant. Bravo!

BTW, today I am ordering a Purifi PTT5.25 woofer-mid. As soon as it comes here I will mount it in a literal shoebox and make a video of the same track I demo'ed the Eton 5-312 with.


ciao,

john
 
  • Like
Reactions: EL506
#54
some examples of woofers running fullrange- no crossover components:


http://web.archive.org/web/20070825...otropic.net/Projects/AudaxSpeakers/audax.html

he says, "The Euros are right. No matter how perfect you get the amplitude and phase with complex crossovers, they end up with that high-end sound: flat, dry, boring. Even when they had noticeable colorations, the first-order networks had a livliness and transparency that made them musically engaging. I ended up with the classic 3KHz 6dB deal in this system."

Tetra Speakers
https://tetraspeakers.com/
https://positive-feedback.com/Issue62/reader14.htm

Watkins Speakers
https://www.watkinsaudio.com/technical-details

diy
https://www.diyaudio.com/forums/multi-way/282141-6fe200-st200.html

https://www.caninialtoparlanti.it/minidiffusore1.htm

https://www.diyaudio.com/community/threads/simple-silver-flutes-2-way.232612/


----
see post #95
https://www.diyaudio.com/community/threads/devore-orangutan-clone.272810/page-5

https://www.madisoundspeakerstore.com/approx-10-woofers/seas-prestige-a26re4-h1411-10-paper-cone/

"This classic paper cone has a smooth response with wonderful roll off at 2000Hz. This is one of the only 10" drivers we know of that can be used in a 2-way by letting the woofer run full range without a crossover. The Seas A26 10" 2-way Kit PAIR (based on the classic A25) is availble with or without cabinets using this woofer. It can also be mated to lower priced tweeters, such as the Seas 27TDFC (H1189) to produce that magical retro sound."
 
  • Like
Reactions: EL506
#56
ChrisABC said:
some examples of woofers running fullrange- no crossover components:
Click to expand...

Thanks, some very compelling arguments for simplification.

In my uneducated opinion: If we use nothing in front of the woofer then we get the problem of baffle-step not being addressed. I suppose that in days past the boxes were either sitting on the floor or against the wall with the only execptions being electrostatics like the Quads. As such the baffle-step was less of an issue.

At one stage I had the AVI Neutron III which had a simple crossover at around 10Khz (the driver was 4") and that was highly acclaimed. I eventually stopped using it since imho the bass was sorely lacking. Baffle step correction is for me a must, potentially that could be addressed with EQ.

Hmmm, might leave the tweeter unhooked and only for decoration. Wonder how many would notice. (just kidding)
 
#58
EL506 said:
In my uneducated opinion: If we use nothing in front of the woofer then we get the problem of baffle-step not being addressed. I suppose that in days past the boxes were either sitting on the floor or against the wall with the only execptions being electrostatics like the Quads. As such the baffle-step was less of an issue.
Click to expand...
true... It is possible some old-school woofers had such a high self inductance that they would naturally roll off in a way which created something of a baffle step compensation. But I think that most of the people who like these kinds of speakers just simply prefer an elevated treble response.
 
#60
hifijim said:
true... It is possible some old-school woofers had such a high self inductance that they would naturally roll off in a way which created something of a baffle step compensation. But I think that most of the people who like these kinds of speakers just simply prefer an elevated treble response.
Click to expand...
Perhaps some older folks who have loss of hearing in the higher frequencies without being aware of it might like them. The Monacor driver that I mentioned earlier is often recommended for use in environments where treble is not a primary concern - after all it goes smoothly up all the way to 8 Khz or so.

AM radio sound reproduction was no higher than 4.5 ~ 5 Khz (depending if you were in Europe or the US: In Europe stations were spaced 9 Khz and in the US 10 Khz). Many FM receivers do not go much higher than 10 ~ 12 Khz, theoretically 15 Khz was achievable. In the days when FM and stereo was introduced High Fidelty frequency range was specified as 50hz to 15 Khz.
 
You must log in or register to reply here.
Top Bottom