Question on Super AA Crossover

[mikebse2a3]

I have a question for AL K or Dean that I've wondered about for awhile.

I've noticed the tweeter portion(input to 2.2mfd) of the network is being taken off of the 40mfd instead of being connected directly to the (+) input from the amplifier and was wondering what advantages are you receiving by this method?

Otherwise it seems to me that it would be better not to pass the high frequencies through the 40mfd for less chance of high frequency signals being degraded.

mike[]

[Deang]

LOL for reasons I won't discuss!!

Ours is not to question why or else get whacked by the Klappenstick.

[djk]

If there was a chance that the speaker was time aligned, then the group delay works out better with the tweeter picked off the 400hz HP.

Since there is no hope of time alignment with a passive crossover, it may sound better to have the tweeter HP go straight to the amplifier.

[Deang]

My thinking was like yours, so I did it both ways. Al's way sounds better to me.

Same deal with the A and AA compared to the AK/AL/AB series of networks. Something about the later filters in the squawker transition area doesn't sound right to me, and I wonder if this might have something to do with it.

[Al Klappenberger]

Mike,

The reason the high filters are tied to the 40 uF and not the input is very simple and very fundamental. My networks are two completely seperate "diplexers" tied one to the next. The 40 uf and 2.4 mHy forms the woofer / squawker cross and the rest of the network is the squawker / tweeter cross. This is true for the "Super AA" and all my network designs. To move the high crossver input to the common input completely louses up the low crossover! The is the main thing Klipsch gets wrong on ther networks. You can NEVER connect two similar type filters together and get a flat impedance. It can be done using two singly terminated filters of the same type but the impedance will ALWAYS be the parallel combination of the termination impedances over the common frequency range of the filters and will change as you go out of the range of one of them assuming the have different cutoff frequencies. BTW: It's my opinion that "group delay" is a red herring and makes no difference to anyhting. It's just a way of quantifying phase linearlty. Propogation delay, on the other had, is a major concern. May people confuse the two.

Here's a graphic that may help explain the idea of 2-port "diplexers" tied together to make a 3-mort "multiplexer":

AL K.

[Erik Mandaville]

Mike:

First and foremost, it's good to ask questions!

"I've noticed the tweeter portion(input to 2.2mfd) of the network is being taken off of the 40mfd instead of being connected directly to the (+) input from the amplifier and was wondering what advantages are you receiving by this method?

Otherwise it seems to me that it would be better not to pass the high frequencies through the 40mfd for less chance of high frequency signals being degraded."

I have an interpretation of this that's different from Al's or DJK's, and IMO Klipsch is not making a mistake in the manner in which they have connected the tweeter branch. Series connected capacitors are nothing new, nor are bandpass networks. I rather think that Klipsch, such as is the case with the type 'A' network, is (intentionally, not by mistake) using the series connected caps in order to both obtain the desired capacitance, and as a convenient tie-point for the physical connection between parts. In other words, in the schematic shown above, the voice coil of the tweeter really is connected directly to the input of the amplifier, only in this case by way of the series 40uf connection. I'll go and double check this, but 40uf in series with 2.2uf results in a series-derived capacitance of just around 2uf. Even though what WE see is a huge capacitor with a little one attached to it for the tweeter, that series connection essentially amounts (does it not?) to the equivalent of a 2uf (approx.) connected directly to the amp input. (edit) I meant that the tweeter branch as a whole really IS connected directly to the input.

Erik

[Al Klappenberger]

Erik,

Your analysis of the network is WAY to simplistic. This is an example of oversimplification or what would apear to be "common sense". I'm sorry to say that the truth is just not a simple! Specificly, you can't just assume the 40 uF in sereis with 2.2 uF means you could substitute 2.09 uF have the same filter. It doesn't work like that becasue the two caps are not just in sereis. There are other components between them!

Al K.

[Erik Mandaville]

Al:

Thanks for your response. You certainly know more about crossover design than I, however I would like to submit the possibility that your explanation may be slightly overly more complex than necessary. There are frequently other components involved, just as the woofer coil also has one end in common with the other parts of the network.

I can of course stand corrected if I over simplified this, and that may be the case; however I don't think it is. The combined capacitance is as you say, about 2.85uf (or 2.9), and that's what the tweeter is seeing. It's essentially the same, IMO, of having that value connected directly to the input. If 2.0uf were connected directly to the (+) input, it would still have other components connected to it, but not in a way that would alter the desired reactance. As is, it still doesn't effect the crossover to the tweeter, since the needed capacitance is simply (yes, I agree it's simple) derived from the series connection of the two caps -- and the opposite end of the 40uf is in fact connected to directly to the input. Thus, the combination of those two parts is really the same as it it were a single capacitor, just as two caps in parallel can behave like a single cap of the additive value of both.

If I am wrong, I would like to learn more about this! Can you explain to me how the other components are influencing the desired crossover frequency for the tweeter?

Thanks,

Erik

edit: I think that series capacitance results in something like 2.085, not what was mentioned above.

[mikebse2a3]

Thanks all for your explanations.

Erik

Yes anytime two capacitors are wired in series the equilivant value is less than the smallest value. I knew just by a simple theoretical calculation of the impedance of the 40mfd that for all intents you could consider the 2.2mfd as tied to the input's (+) and thus my question of what other benefits was there to this hookup.

So considering that the theoretical is affected by real world conditions(ie like capacitors that don't measure as theory would calculate at higher frequencies for one example) a large value capacitor might exhibit some problems at the higher frequencies and that there might be an audible benefit to actually hook the 2.2mfd directly to the input.(even if the value needed to be tweeked slightly say to 2mfd in your example.

djk

Yes that is what I was thinking that somehow there might be some phase advantage for this hookup but considering the phyiscal difference in locations of the tweeter and squawker that any benefits of this was lost in the case of our klipsch speakers.

Dean

LOL! I like youe style!!!

I would have to have done as you and tried the other connection for any audible differences I might have perceived.

AL K

Thanks for taking the time to explain this. So if I understand you correctly this hookup in theory will give you the smoothest impedance curve of the crossover network as its main benefit in our klipsch speakers?



AL K said: The 40 uf and 2.4 mHy forms the woofer / squawker cross and the rest of the network is the squawker / tweeter cross. This is true for the "Super AA" and all my network designs. To move the high crossver input to the common input completely louses up the low crossover!

How would moving the tweeter hookup position screw up the low crossover? By low crossover are you refering to the woofer's crossover or possibly the Squawkers?

So my orginal question and thought was if the 2.2mfd connection was moved to the (+) input terminal then it seemed to me that the impedance would only be effected slightly(If this assumption is correct) with the benefit of not passing all the tweeters signal through the 40mfd and this could have had an audible benefit that some might want to try. (note: Dean did try this and thought the original sounded best.)

Again AL K thanks for explaining your network design.

mike[]

[Deang]

Erik, you're leaving out the reactive component, and not factoring in the impact on impedance.

Ratio = [ ( FqH / FqL ) 1 ]

FqL2 = ( FqL * FqH / ratio )^1/2

FqH2 = FqL * FqH / FqL2

C1 = 159155 / ( FqH * Rt )

C2 = 159155 / ( FqL2 * Rm )

L1 = 159.155 * Rm / FqH2

L2 = 159.155 * Rw / FqL

According to your understanding of this the Type A has a crossover point of 12kHz.

[Erik Mandaville]

Al:

For what it's worth, I used to address this very same aspect of the Klipsch networks. However, what I have come to conclude is simply that those series connected capactors in common (sort of) with the midrange and tweeter function in a way that's similar to a series resistor voltage divider, where the desired working voltage for one load (say the output stage of an amplifier) is derived at the junction of the two series resistors, and the next needed voltage is obtained from the combined resistance of both resistors.

This is similar to how I believe the tweeter/mid branches of these specific crossovers are functioning, only that the capacitors deal with frequency division rather than voltage division. In effect, the needed capacitance for the midrange section is derived from the first of the series caps (which is referenced to the input of the network), and the second (for the tweeter) is obtained from the series connection of the two. I don't see that the two interact in any way more than if the series obtained value for the tweeter had been connected by itself to the positive input to the network.

Erik

[Erik Mandaville]

Dean:

I have lots of books here, too. However, would you please identify the value of capacitance needed based on the impedance of the driver we are discussing?

To simplify: What is the value of the capacitor needed?

Erik

[Erik Mandaville]

Mike:

I have made several of these networks doing it both ways, and I hear no difference between them.

Erik

[Al Klappenberger]

Erik,

The reason you can't just add those two caps together is very simple': THERE IS A .3 MHY INDUCTOR BETWEEN THEM!!

The 40 uF cap sets the squawker low frequency limit (400 Hz). The .3 mH sets the squawker upper frequenc limit(6000 Hz). The 2.2 uF and the other parts in the tweeter filter set the lower limit of the tweeter (also 6000 Hz).

Al K.

[Erik Mandaville]

Al:

What I said above, once again: "If I am wrong, I would like to learn more about this! Can you explain to me how the other components are influencing the desired crossover frequency for the tweeter?"

You indicate that:

"THERE IS A .3 MHY INDUCTOR BETWEEN THEM!!"

Ok, I can hear that clearly enough. As I also said, if I'm wrong, I absolutely stand corrected. However, that inductor is not really 'between them' per se'. Rather it has one end (the input) side in common with those two caps, leaving the 40uf and 2.2uf in series with both each other and the input. How does that single coil influnce the series capacitance we have been discussing?

"The 40 uF cap sets the squawker low frequency limit (400 Hz). The .3 mH sets the squawker upper frequenc limit(6000 Hz). The 2.2 uF and the other parts in the tweeter filter set the lower limit of the tweeter (also 6000 Hz)."

I understand that. But, that upper limit induced by the coil is obtained on the opposite side of the coil -- the end that connects to the input of the autoformer, right? The parts associated with the tweeter, from what I can see above, are the 40uf/2.2uf series connection, the parallel coil to common, and then the last capacitor in series with the positive side of the tweeter.

Al: If you don't want to hassle with this, of course that's fine. I'm just trying to add to my understanding is all.

Erik

[NOSValves]

Dean an ignore feature on this forum would be really useful. You know the kind where you simply have no way of seeing the particular persons post.

[Al Klappenberger]

Erik,

All this could be called "Filters 101".

Here's the first lesson: Mentally plug each part in the Super AA network into the 4 highpass and lowpass filter boxes in the graphic I posted earlier. HINT: One single part in the network corresponds to one single box in the graphic EXCEPT the tweeter filter that includes 3 parts which all go in a single box.

How here an exercise: If you combined the 40 uf and the 2.2 uF caps, on which side of the new 2.09 uF cap would you connect the .3 mHy inductor. In other words, Try to fit that single cap into two boxes!

AL K.

[Erik Mandaville]

Al:

Sure, we can call this "Filters 101" or whatever you like. I'm just trying to understand, since this is an aspect of the Klipsch networks I've looked at before.

You are not understanding the question I posed to you. One can't simply use the single 2.09 cap for this. The midrange must have the 40uf. This is where either you are not following me, or more likely, I am not explaining myself clearly.

It's alright, I have enough books on crossover design to brush up on this.

Thanks,

Erik

[Al Klappenberger]

Erik,

Your's right, I am not following you! You MUST fit the parts into the boxes to understand this. Until you have done that you are spinning you wheels.

AL K.

[Erik Mandaville]

Al:

You're also right: I spin my wheels sometimes, but usually (hopefully) they eventually catch. I have a mental block on this for some reason, but I'll give it some more thought sometime. I've got another more serious audio-related electronics problem that needs attention right now.

Erik