AA Xover Modification

[lynnm]

I modded the AA's in my 1982 Khorns this weekend: (With acknowledgement to ALK et. al.)

1. Disabled the zener diodes in the tweeter circuit by clipping a wire on one side of the diode.

2. Removed the screw that secures the tweeter inductor to the Xover board (used butyl rubber to stick it to the xover board).

3. Bypassed the inductor in the woofer circuit by wiring the woofer directly to the input.

A couple of questions for you real techies out there:

The xover slope in the tweeter circuit of a true Type A is 6 db - The AA uses a slope of 18 db. How can I mod the AA to yield a 6 db slope ? I notice that the True Type A does not have an inductor in the tweeter circuit and wonder if simply bypassing the inductor would change the slope but I would prefer not to do so without some expert input.

I have a schematic for the Type A but although I know that a copy of an AA has been posted here I could not find it.

Al Klappenger recommends paralleling a 3.9 uf cap with the 2 uf in the tweeter circuit to smooth the response curve. Could Al or someone else perhaps give me a little detail on what the benefit of doing this would be ??

As you may have gathered from the above I am not completely green in electronics. I can honestly say that I know enough to be a menace . Any guidance would be appreciated.

Limited listening impressions:

The change - If indeed there is any - Is subtle. That said, I find that there appears to be a slight improvement in the bass and possibly a slight smoothing in the upper midrange and low treble. I should say that others considering these mods take my assessment of upper frequency changes with a grain of salt as my hearing is poor in that range.

Overall my Khorns sound a little smoother to me with the mods - Not that they weren't lovely to listen too before.

[John Warren]

z

Edited December 17, 2013 by John Warren

[jhawk92]

Lynn-

If you can, would you post or send a pic of your zenier diode "mod." I think I may have them as well in my '68's, but before I do any clipping of parts, I wanted to be sure. Thanks.

[3dzapper]

John, does this sound right to you, not knowing the Q of the tweeeter:

10uf; 24mh; 8ohms?

With that peak just one octave below the crossover point I can see the usefullness of this filter in the circuit.If the values are correct, I'll play with it. It's easy enough to put in and remove.

Rick

[JohnA]

Lynn,

You'll have to remove the 245 uH inductor and one of the 2 uF caps from the circuit. Then you'll have a Type A. Personally, I'd build a second set of Type AAs using premium components without the zener diodes. They protect the tweeters better. In fact, to get the flatest response, I might build a 3rd-order Butterworth high-pass for the tweeter, at 5000.

[lynnm]

John

Do you have a copy of the AA schematic ?

Is the inductor in the tweeter circuit the 245 microhenry unit you are referring to ?

Some more critical listening is needed but I may reconnect the inductor in the woofer circuit because last night I thought I detected a bit of muddiness in the bottom on a couple of Saxophone solos.

On the other hand that may simply be the result of my concentrating on the sound of the speakers too much rather than the music

Damn this kind of fiddling is easier when the differences are blatantly obvious !

[Frzninvt]

Ryan, the reason for the removal of the screw that holds down the tweeter inductor is that it reacts with the coil and throws the value off. Al K came across this in his testing I believe. So it will make a difference with it removed.

[3dzapper]

Type in AA schematic in the search box above. John Albright provided the AA schemitic for us not long ago. rf3iicrazy

is the poster.

Rick

[lynnm]

3dzapper

Thanks - I got it!

[John Warren]

----------------

On 10/14/2003 6:51:13 PM lynnm wrote:

Al Klappenger recommends paralleling a 3.9 uf cap with the 2 uf in the tweeter circuit to smooth the response curve. Could Al or someone else perhaps give me a little detail on what the benefit of doing this would be ??

----------------

The added 3.9 uF cap makes the AA similar to a Butterworth. The stock AA is a Chebychev that provides a bit of "boost" at the crossover frequency of 6kHz. The frequency reponse curves of the T35 show a 3-5dB notch beween 6-7kHz. So is the Butterworth really "benefit"?

The impedance peak at Fs is what should be "keeping you up at night" because with it,

you don't know what you got!

[Randy Bey]

John,

that's interesting because my in-room response showed a peak at 5-6K with both AA and ALK crossovers.

Switching to type A crossovers didn't eliminate the peak, but reduced it to a bump.

Much better sounding, IMHO.

[lynnm]

I have just finished rewiring my AA's and setting them up as A's.

Reintroduced the woofer inductor

Removed one 2 Mfd cap from Tweeter circuit.

Removed the Zener Diodes from the Tweeter circuit

Removed the 245 Mh inductor from the Tweeter Circuit

By so doing I have reduced the tweeter Xover slope from 18db @ 6k to 6 db @ 6k ( I think )

This allows for a smoother squawker -> tweeter transition but also slightly increases the tweeters vulnerability to damage from overdriven ampflication. My tube amp is rated at 30w/ch and I suspect that my horns are rarely subject to power spikes that are problematic.

I had tried wiring the woofers directly to the input but found that the bass while seemingly a little richer sounded a bit muddy on occasion especially when saxophones dipped into their lower regions..

I have listened a bit to my reconfigged Xovers tonight but I will give a more detailed report once I have had time to give the new setup a reasonable amount of listening time.

[John Warren]

z

Edited December 17, 2013 by John Warren

[Al Klappenberger]

The reason the "A" sounds slightly better than the "AA" is because the

tweeter filter in the "AA" is not properly designed and has a bad response

curve in the tweeter's frequency band. The single capacitor in the "A" is so

simple that all the prototype filters (Chebyshev, Butterworth, Gaussian,

etc.) all intersect in a first order filter. You can hardly define what type

of filter it actually is! The source and load impedance define it for any

given element value. In fact, first order filters are so sloppy that the

tables of normalized filters in the "Handbook of filter synthesis" by Dr.

Anatol Zeverev (a filter designers bible) start at second order for all

responses.

The tweeter filter in the "AA" is NOT really a 18 dB / octave filter. That

is, it is NOT a "Butterworth" response. Furthermore, it is a "doubly

terminated filter", which means it is intended to be terminated in equal

impedances at either end. This can be seen from the fact that its element

values are symmetrical. It has a 2 uFd cap on both ends. A speaker driver is

assumed to be some constant impedance, usually 8 Ohms. This means THE SOURCE

IMPEDANCE SHOULD ALSO BE 8 Ohms! An amplifier's source impedance is MUCH

lower than that. In this case a "singly terminated" filter is required. If

you reverse engineer the tweeter filter in the AA you will find that it is a

doubly terminated Chebyshev filter having .2 dB passband ripple requiring a

load AND source impedance of 10.7 Ohms. The 3 dB cutoff would then be 4700

Hz. In an 8 Ohm system, with a voltage source, like and amplifier, it simply

becomes the wrong filter for the job!

The correct design for this filter, in stand-alone form, would be a singly

terminated filter designed for a third order Butterworth response to operate

between zero Ohms (a voltage source) and a constant impedance of 8 Ohms and

having a cutoff frequency of 6000 Hz at 3 dB. From the table of normalized

filters in the Zverev filter handbook, the values are 1.5, 1.3333 and .50.

These values scale out to 2.21 uF at the input, .159 mHy to ground and 6.6

uFd at the output. With this filter, the tweeter will not be subjected to the

extreme diaphragm excursion at low frequency that it would be with a first

order filter and the passband will be a smooth curve like that of the

single cap in the "A" network.

Now, where have we seen values similar to these before?

Al K.

[sfogg]

Just out of curiosity has anyone ever played with first order series crossovers on Klipschs? As I understand it their primary benefit (with other negatives) is that when the crossover freqency shifts due to a drivers impedance change it effects both the high and low pass function so that they shift together to sort of compensate for the change. In a parallel crossover the drivers impendance effects just its own crossover point. As such when it moves (unless it is compensated for) it either creates a hole or overlap near the crossover point.

At some point I'm going to build some two way speakers (811b with Altec 902-8B and Lambda Acoustics TD-15M) as surrounds for my La Scalas and I was thinking for my initial testing I may give a first order series crossover a whirl. Not sure if the Altec driver can handle a first order crossover around 800-1000hz though.

Shawn

[JohnA]

Lynn,

The distortion you were hearing on saxophones MAY have been due to highs getting to the woofer. Try a high quality air-core inductor about 12 or 14 ga. That'll kill the highs like the iron inductor without the saturation problems and loss of output it has. Al tested on and found it changed inductance at just a couple of watts.

A 2 uF cap in series with a 7 ohm tweeter gives a -3 dB point of 11000 Hz. If the tweeter had an impedance of 14 ohms (16 nominal) the crossover point would be 5800. Early t-35s were 16 ohms; those would have been the ones Klipsch was using in the Type A era.

Like Al says, 6 dB crossovers are so gradual, an octave here or there is no problem.

While all of us are coming at you with different solutions and why, the use of premium components in the final design is the ultimate solution and cannot be over emphasized.

Try this circuit for a Type A design with a little tweeter protection and voicing similar to the Type AA.

[Al Klappenberger]

I recomend the network John posted over the circuit of the "A" hands down! That one is constant impedance. The "A" network is not! One small point though: For a tap setting fo 4-0 the 10 Ohm resistor is a smidge low. The calculated value is 11.55 Ohms. 10 Ohms is best for 2-5 or X-4 settings. The value is not really critical at all.

Al K.

[sfogg]

"I recomend the network John posted over the circuit of the "A" hands down! That one is constant impedance."

On looking at that circuit and thinking about it a bit more aren't the swamping resistors going to alter the amount of power that is delivered to the drivers based on the drivers impedance changes vs. frequency?

In the frequency where impedance is highest won't more power be burned through the resistor, path of least resistance, and hence less through the driver then at the points where the drivers impedance is lowest?

Won't this act as an EQ of sorts?

Shawn

[Al Klappenberger]

Shawn,

NO! The load presented by the K55 wquawker is a rather constant 13 Ohms. It actually only varies plus minus an ohm or so. The swamping resistor only absorbs the power not used by the squawker by the very act of cutting it down by the transformer. This is the also the act of lowering the impedance back to 8 Ohms after the transformer raised it to 26 Ohms. Tap setting 4-0 is 1.414 : 1 turns ratio which is 2:1 impedance ratio. 2 X 13 = 26. Parallel that with 11.55 Ohms and you have the needed 8 Ohms back and the level reduced by 6 dB. If the swamping resistor is 10 Ohms instead, it comes to a respectable 7.2 Ohms.

As to the swamping resistor, it is being fed by a power amp with is a low impedance source. An analogy would be antomobile headlights run by the 12V battery: Does one headlight get brighter if the other burns out? Answer: No! Each draws what it needs from the source which does not effect the other. Only the load on the battery changes, that is, the load impedance doubles when one burns out.

Al K.

[sfogg]

Al,

" NO! The load presented by the K55 wquawker is a rather constant 13 Ohms. I actually only verus plus minus an ohm or so. The swamping resistor only absorbs the power not used by the squawker by the very act of cutting it down by the transformer"

I was thinking more about the resistor in the tweeter circuit above and in general terms for a driver that has a larger changing impedance curve.

"As to the swamping resistor, it is being fed by a power amp with is a low impedance source."

Depends on the amp... some will have an ohm or two of output impedance.

"An analogy would be antomobile headlights run by the 12V battery: Does one headlight get brighter if the other burns out? Answer: No! Each draws what it needs from the source which does not effect the other. "

Because the source has more available power then both can draw together. If you instead fed the two bulbs from say a more limited/regulated fixed 10 watt source and one bulb burned out the other is going to get brighter.

Similar to the case of a dead/dying battery in a car. If you try cranking the engine with the lights on the lights will certainly dim as you are adding another draw on the source.

"Only the load on the battery changes, that is, the load impedance doubles when one burns out."

As such in a car when one burns out the source is now putting out less power. But if the source was limited (like an audio amplifier) more power could be going through the remaining bulb then when both bulbs were active.

Shawn