As I understand it SET amps in particular do best with well behaved (flat) impedance loads. Is this primarily an issue in the bass and mid-range area? Specifically would a tweeter circuit that presents a rather high and/or rising impedance be a problem? My assumption is that the impedance concerns are primarily in the high energy parts of the spectrum but knowing what the first three letters in assume spell I want to be sure that there are not stability issues or something that I had not considered.
All Generalizations Have Exceptions.
Start with a particular SE amplifier design.
What is the output impedance?
Is it very low, medium, or very high; depends on the exact amplifier.
Speaker impedances vary all over the map.
What speaker?
And then there is not only the impedance of the speaker versus frequency, there are the capacitive and inductive reactances, and the resistive components.
Reactive = Elliptical Load Lines.
If you think some Push Pull amplifiers do not have those problems, they can too.
It often works better than all the above might lead you to think.
Remember, the early engineers could not understand how the bumblebee could fly.
But the Bumblebee went ahead and flew anyway.
Without specifics, we are like a blindfolded archer, who is rotated an unknown number of degrees, and then told to hit the target.
Start with a particular SE amplifier design.
What is the output impedance?
Is it very low, medium, or very high; depends on the exact amplifier.
Speaker impedances vary all over the map.
What speaker?
And then there is not only the impedance of the speaker versus frequency, there are the capacitive and inductive reactances, and the resistive components.
Reactive = Elliptical Load Lines.
If you think some Push Pull amplifiers do not have those problems, they can too.
It often works better than all the above might lead you to think.
Remember, the early engineers could not understand how the bumblebee could fly.
But the Bumblebee went ahead and flew anyway.
Without specifics, we are like a blindfolded archer, who is rotated an unknown number of degrees, and then told to hit the target.
Last edited:
Any amplifier with a highish Rout, which is most SETs, a portion of the impedance curve of the loudspeaker imposes itself on the frequency response.
Since almost all dynamic speakers have a rising impedance peak at resonance, it can be an issue, and a SET with the wrong speaker often gets complaints of boomy fat bass. But this can be dealt with when you selct your loudspeakers.
Here is an example, 1 driver, 3 different designs depending on what the Rout of your amplifier is. nd the same sort of thing for the higher end driver.
http://wodendesign.com/downloads/King-of-Swingers.pdf
http://wodendesign.com/downloads/Nostromo-Alpair11ms-reflex.pdf
Another approach taken in the Frugel-Horns for instance is a variable amount of damping that tunes the bass.
The more insidious issue is non-flat impedance higher up, almost always caused by an XO. This can boost the upper midrange (where it often occurs) and colour things. One of th ereasons FR drivers are often seen used with SETs.
An astute designer can do a flat impedance XO but most don’t care. Here a Klipsch stock and with an XO where attention has been paid to the end impedance.
dave
Since almost all dynamic speakers have a rising impedance peak at resonance, it can be an issue, and a SET with the wrong speaker often gets complaints of boomy fat bass. But this can be dealt with when you selct your loudspeakers.
Here is an example, 1 driver, 3 different designs depending on what the Rout of your amplifier is. nd the same sort of thing for the higher end driver.
http://wodendesign.com/downloads/King-of-Swingers.pdf
http://wodendesign.com/downloads/Nostromo-Alpair11ms-reflex.pdf
Another approach taken in the Frugel-Horns for instance is a variable amount of damping that tunes the bass.
The more insidious issue is non-flat impedance higher up, almost always caused by an XO. This can boost the upper midrange (where it often occurs) and colour things. One of th ereasons FR drivers are often seen used with SETs.
An astute designer can do a flat impedance XO but most don’t care. Here a Klipsch stock and with an XO where attention has been paid to the end impedance.
dave
Thanks guys. What I have in mind is that many tweeters have a pretty steeply rising response above say 5 or 6K and sometimes it seems that just adding a little series resistance tilts that down some as well as lowering the overall level. It seems like a simple solution but I wouldn't want to throw amps into hissy fits. Of course simple inline resistors have been used for a long time so I suppose it would be pretty benign.
Do you generally use zobels to reduce rising impedance on woofers and midranges? It adds more reactive elements but the intent of course is to reduce the overall reactance.
Do you generally use zobels to reduce rising impedance on woofers and midranges? It adds more reactive elements but the intent of course is to reduce the overall reactance.
Last edited:
That's right, and in order to take out this normally un-called for source of EQ you would be aiming to take out the reactance of the load at the same time. It's not generally a bad thing to do, if necessary and if you do it properly.
I learned this the hard way in my horn based speaker system, zobel networks may be effective in taming the rising impedance with frequency, and doing what you can to lower source impedance driving HF drivers also helps. (Like using global feedback, 8 ohm taps on 16 ohm drivers, etc.)
Zobels in a multiway are an XO thing, in a FR if there is rising impedance response (typically only ones without copper capped pole pieces) than a zobel could be useful.
dave
Thank you Dave, I've found this thread ( Dean's Super AA - Technical/Modifications - The Klipsch Audio Community ) together with others, but is it specific to La Scala speakers?
Not one specific, just a general discussion about that crossover (and another aspect I was not aware of: the importance of its impedance). The plot shown by Dave comes from here: ALK Engineering
The LaScala example just shows that flat impedance is possible.
I have yet to hear a Klipsch i’ve liked.
Elsinore is another multiway with flat impedance
dave
I have yet to hear a Klipsch i’ve liked.
Elsinore is another multiway with flat impedance
dave
Yes, every crossover variation is different. If you set this up then every time you make a crossover change you need to come back and adjust the conjugate.
It's easy enough to do.
That there idea is a myth. So-called engineers who know nothing of bumble bee's might think it so, but weigh one and measure its wing area and it becomes obvious that they will have no issue taking flight.
cheers,
Douglas
I don't think it was an issue of the bumblebee's lift/drag/weight, i.e. its ability to glide. Any dude in a cape can do that🙂 It was propulsion by means of flapping wings, which has only recently been duplicated by drones.
May I ask you how to start working on a crossover to flatten its impedance?
Are there some rules based on speakers impedance curves and the number of frequency ranges?
I hope to still be on topic,
thank you.
It's the same as when you perform Zobel compensation to individual drivers, except that you do it to the whole speaker after making the crossover.
Have you made a crossover before using impedance curves?
Have you made a crossover before using impedance curves?
I've played with Zobels and the general concept is clear (at high frequencies the speaker's impedance raises, so I add in parallel an RC that will drop its impedance accordingly to keep the overall response as flat as possible), but never put my mind on the design of a crossover for multiple drivers.
Yes, well it's no different really. There are two common types of compensation, the RC you mention and the RLC used for peaks. A crossover usually causes a peak.
I don't think an absolutely flat impedance is anything like required. Avoid it doing wacky stuff where the amplifier is otherwise compromised is a good thing. It is the overall response that matters.
Aside from that, active, line-level frequency division is my preference also. If a clipping( and thus delivering more power into the HF ), amplifier is not connected to a tweeter, that tweeter is safer… 🙂
It is also easier to build a steep-slope X-Over at line level with op amps...delivering a 24 dB/octave power-handling cross over is slightly cumbersome. Not that the active x-over is not slightly cumbersome, but I am not building that beast...heh-heh-heh.
cheers,
Douglas
Aside from that, active, line-level frequency division is my preference also. If a clipping( and thus delivering more power into the HF ), amplifier is not connected to a tweeter, that tweeter is safer… 🙂
It is also easier to build a steep-slope X-Over at line level with op amps...delivering a 24 dB/octave power-handling cross over is slightly cumbersome. Not that the active x-over is not slightly cumbersome, but I am not building that beast...heh-heh-heh.
cheers,
Douglas