autoformer general question

[thumperxes]

on autoformers with regards to tap questions can the tweeter and squaker be on the same tap setting or must only one tap be used per speaker. If only one speaker per tap is there anything wrong with using 2 autoformers.

[JohnA]

You can place both drivers on the same tap if you need to. The caps or inductors may have to be sized for the change in impedance. There is also nothing wrong with using 2 autoformers, except cost.

[WMcD]

I agree with John.

If you move the tap for the midrange, the cap at the input will have to be changed. A resistor can be added if you're going to tap with less output.

Generally there should not be a problem moving things so that the feed to the tweeter crossover is on the same tap as the feed to the midrange.

There is a larger issue about how the load presented by the mid does not work as a parallel resistance with the load presented by the tweeter to mess up the value of the cap feeding the autotransformer.

The answer is that internal inductance of the mid acts an inductor (a low pass crossover filter) and so the load goes up in impedance above, roughly 6 kHz. Essentially, in the tweeter range, there is no load presented by the mid. Very rough approximation here.

The tweeter has a crossover filter, if even a single cap. It acts so that below 6 kHz, the tweeter plus its cap, presents no load below 6 kHz. Again a rough approximation, for discussion.

This is one of functions of analog filters which is seldom discussed. We normally think of the filters of not passing certain frequencies (an open switch) and passing others (a closed switch). We're happy that this frequency selective switch passes only some freqs where it acts as a closed switch.

But the reason these filters work is,that the "open switch" area presents a very high impedance to the input from the amplifier (as a real switch would). In other words, it is not quite so much that the crossover filter shunts power away to keep it from getting to the driver. Rather, the input to the filter is incapable of taking power from the amplifier.

Therefore,they don't work as a parallel load if designed properly. The switches to the mid and to the tweeter (and this extends to the woofer) are never closed at the same frequency.

This is why having them connected to any tap on the autotransformer does not mess things up. The same is true of a proper design without an autotransformer.

Gil

[boom3]

This is the most germane crossover discussion I've seen in a while. We tend to think of the autoformer as just a level control without considering that it is after all a tapped inductor...

Is there a chart that relates the Klipsch autoformer tap settings to cap value needed for the K-55 family and associated horns?

[Groomlakearea51]

Don't know if this is what you need, but this is what Bob Crites shows on his site.

T2A(3110A) -3, -6, -9, -12db

T3A(3465-M) -6db

T4A(3485) -4db

T5A(3496) -6db

T7A(3504) -10db

T8A(3507) -8db

T9A(3540) -3db

T10A(3542) -9db

T11A -12db

[WMcD]

Well, I did not mean to give the impression that the autotransformer acts as an inductor. In my view, it just acts as a two-winding transformer would.

When there is a resistive load on the secondary, the primary sees a resistive load. It does not act as a choke for high frequencies.

Maybe at some high frequency, inductive effects come into play. It is my understanding that transformers are sometimes wound with parallel windings (e.g. tri-filar). We know that parallel inductors reduce what would otherwise be their inductance, like series caps. And at some very high freqs, the interwinding capacitance and the inductance of the windings can resonate.

The autotransformer is passing the the tweeter freqs, something we know for a fact in the A or AA, etc. And of course in amps with output transformers the treble gets through.

On the other hand, you can use a transformer winding as an inductor. That is not a contradiction. One time I tested this with a multimeter which measures inductance.

Gil

[Erik Mandaville]

"This is the most germane crossover discussion I've seen in a while. We tend to think of the autoformer as just a level control without considering that it is after all a tapped inductor..."

Agreed it's a good discussion. The fact that the autoformer is a multi-tapped inductor has been actually brought up a number of times in the past. It's somewhat related to how it functions as an attenuator. As was pointed out above, an eqaully effective crossover can also be designed without it. I know that from experience myself despite the number of times Mr. Klipsch's work to the contrary is brought up in response. (please note that I have great admiration for his work, just a different pair of ears that I'm trying to please). Autoformers have been described as sort of more 'transient perfect,' however I didn't find that in comparisons I've made using both types of the same network order using the same crossover frequencies. ANY of the older Heritage networks can be made using a pair of resistors rather than an autoformer. It consists of a bandpass (series connected coil and cap) on the squawker, followed by an L-pad calculated for the impedance of the driver and desired reduction in output. The tweeter is connected directly to the input by way of a single capacitor. Very simple; very effective. One just needs to keep in mind that, when calculating the desired values of capacitance and inductance, the actual impedance of the driver is used rather than the reflected impedance imposed by the autoformer. You can also design a mixed order network fairly easily, such as using a lower x-over frequency for the tweeter accompanied by a sharper cutoff, such as 18dB/octave. That's what I use on my own networks at a frequency of 4,000Hz. Best not to do that with the K-77, though. For reduction in tweeter gain minus the autoformer, an additional fixed or variable L-pad is needed, however, which is something I admit makes the convenience of the autoformer attractive.

Erik

[Erik Mandaville]

"On the other hand, you can use a transformer winding as an inductor."

That is very true.

[BEC]

Well, I did not mean to give the impression that the autotransformer acts as an inductor. In my view, it just acts as a two-winding transformer would.

When there is a resistive load on the secondary, the primary sees a resistive load. It does not act as a choke for high frequencies.

Maybe at some high frequency, inductive effects come into play. It is my understanding that transformers are sometimes wound with parallel windings (e.g. tri-filar). We know that parallel inductors reduce what would otherwise be their inductance, like series caps. And at some very high freqs, the interwinding capacitance and the inductance of the windings can resonate.

The autotransformer is passing the the tweeter freqs, something we know for a fact in the A or AA, etc. And of course in amps with output transformers the treble gets through.

On the other hand, you can use a transformer winding as an inductor. That is not a contradiction. One time I tested this with a multimeter which measures inductance.

Gil

Gil,

You are right about the autotransformer passing the tweeter frequencies but you gave the wrong reference for that. The type A or AA do not use the autotransformer in the tweeter circuit. The Type B and E do use the autotransformer in the tweeter circuits.

Bob Crites

[WMcD]

Whoops, I typed without brain engaged. BEC is absolutely correct. As PWK said, I now disagree with the author of my writing.

Gil

[Arizona Dan]

I saw this thread on autoformers, and thought I would use this chance to ask a question. Hope you don't mind. My McIntosh amp (MC402) has autoformers built in. Is it ok to use the AA crossover with another autoformer, or would a modification to the crossover be better.

Dan

[WMcD]

Maybe there IS a situation where an autotransformer can serve as the L in a crossover. It is annoying that I've written something which is not technically accurate. But better to correct it. Something in the back of my mind compelled me to research the issue tonight. (Didn't I read something somewherre?)

Attached is a PWK patent on a crossover. It seems that PWK submitted a patent application on the bass horn and crossover as one application. People here will recognize it as the LB bass horn.

- - - -

This legal matter is somewhat unique to US patent law. Related but different inventions can be covered by one US patent. Usually it is a matter of a substance and means of making a substance. That is not true in other countries.

The USPTO, even under US law, must have found that these inventions (bass horn and crossover) were not related enough to be included together as a single patent. Therefore this patent on the crossover became a division of the original filing. This is why there are two patent documents where the description and drawings are similar, if not identical. The claims, though, are different. This one claims the crossover. Another claims the bass horn.

- - - -

Incidentally, the specification is a wonderful description of PWK's thoughts on how his horn loaded speakers are designed. A valuable description of his design process. Very much a must read.

- - - -

But, regarding crossovers.

Here we see a description on how values of the multi-pole (having L and C components) crossovers can be have values which peak up output This is something not discussed much at all. Filter circuits are usually created so that they have rolloffs without peaks. However, it is possible to make passive crossovers with peaks to compensate for driver output sags.

This peaking has a price. This is somewhat contrary to my "switch" theory in a post above. That was simplistic, for purposes of discussion.

The price, or method of operation, is that the filter actually has a lower input impedance where the output has a desired voltage peak. Essentially, if you want more power output from the filter in a narrow band, it has a lower input impedance in the same band to draw more current out of the amp. A form of equalization.

Per the description by PWK this can be accomplished without an autotransformer in the mid and woofer.

- - - -

But let us look at the tweeter circuit.

1) The autotransformer is being used as a step-up transformer. This is different than the typical Klipsch set-up where the auto transformer is being used as a step down transformer. This would work without L-C peaking. And it would lead to a lower input impedance to the auto transformer (or any transformer) regardless of peaking from LC effects.

2) The tweeter circuit is described as a three-pole device. We see two caps. Therefore, the third pole must be caused by the inductance from the autotransformer. Therefore, we should not think that autotransformers can never have some inductance. My statements to the contrary are overgeneralizations; based on this description by PWK which speaks to the contrary. I wanted to put up this patent with these claims because some of the claims imply an L effect. An LC circuit is described.

3) I believe that autotransformers are used in other Klipsch crossovers and SS Mac because the windings are well coupled to each other by the magnetic core and the continuous winding. I'm holding my ground on this issue.

4) It is interesting that the autotransformer described in the tweeter circuit is a T2 with part of the iron removed. My guess is that this leads to a reduction of magnetic coupling and therefore there is some inductance created which would otherwise not be present.

5) We do, though, see that PWK is describing a resonance in the tweeter crossover in which the loosely coupled (my term) autotransformer forms a part. An "L."

6) Again, I believe this is not an issue in other designs using a well ironed autotransformer like the unmodified T2.

7) People who favor air core inductors will find some solice in that these are favored in the spec.

I'm being a bit picky and this is a bit OT. OTOH, it is food for discussion and insight.

Best,

Gil

USP004237340.pdf

[Erik Mandaville]

"There is also nothing wrong with using 2 autoformers, except cost."

Of course this is true, and cost certainly is a significant factor. If one were to maintain the use of the autoformer for the squawker, a far more common and widely used approach to reduce tweeter gain is a variable L-pad. It can be adjusted by the user 'on the fly' to dial in the exact amount attenuation desired, and requires no change in related passive part since the original load impedance is maintained. Suggesting that two autoformers can be used, to me, implies that the tweeter circuit can be treated individually and does not necessarily have to be associated with the primary capacitor between the network input and autoformer (EX: Type 'A').

Can we identify the specific reason for not connecting the tweeter filter directly to the amp input?

Erik

[boom3]

"There is also nothing wrong with using 2 autoformers, except cost."

Of course this is true, and cost certainly is a significant factor. If one were to maintain the use of the autoformer for the squawker, a far more common and widely used approach to reduce tweeter gain is a variable L-pad. It can be adjusted by the user 'on the fly' to dial in the exact amount attenuation desired, and requires no change in related passive part since the original load impedance is maintained. Suggesting that two autoformers can be used, to me, implies that the tweeter circuit can be treated individually and does not necessarily have to be associated with the primary capacitor between the network input and autoformer (EX: Type 'A').

Can we identify the specific reason for not connecting the tweeter filter directly to the amp input?

Erik

Well, Paul disliked L-pads...and he felt the need to attenuate the tweeter so it had to be adjusted by an autoformer

I appreciate the attenuation chart provided, but I was asking for a correlation between the change of inductance created by changing taps on the autoformer and the resulting change to the series cap required for the K-55 series driver when used with the K400/401 horns. Seat o' the pants tells me this change is probably slight between any two adjacent taps. This is probably of academic interest only since I don't forsee changing out expensive caps to change the values by a small fraction just to track the target crossover frequency a little better.

[Erik Mandaville]

"Well, Paul disliked L-pads...and he felt the need to attenuate the tweeter so it had to be adjusted by an autoformer"

Sure, I'm aware of that. Simple enough. However the example I gave had to do with a specific network ('A') which does not attenuate the tweeter. Rather, the tweeter is connected in series AFTER the capacitor between the amplifier input and autoformer. In theory, this might possibly also increase related ESR, which is a characteristic that has been addressed in the past in relation to potentially lower tweeter output and perceived HF response with older capacitors. I was simply making the point that the tweeter filter in THAT particular network should not have to be associated with the autoformer in any way in order to functional properly and effectively.

The question you are asking is something I have done quite a few times. The main job of the autoformer is to increase (or decrease) the load impedance 'seen' by the amplifier, which in turn will bring about either and increase or decrease in gain. Since the values of capacitance and inductance in crossovers are in part based on driver impedance, it would be correct to assume that alterations of impedance imposed by the autoformer would require an appropriate change in values of associated capacitance. I have switched between tap 4 and tap 3 on the autoformer many times without changing the value of the primary capacitor, and the only change in response that I could detect was the resulting attenuation brought about by the impedance increase from one tap to the other. The response might not measure that way, but there is not always a direct correlation between measured and perceived response as it relates to sound quality.

The breadboard version of the network below was to test the quality of a type A network that used a resistive L-pad to attenuate the squawker rather than the autoformer. It was an experiment. The result was a design that to me had an obvious increase in high high frequency response and presence. The tweeter filter, which at that time was a first order design, was also connected directly to the input of the amplifier rather than behind a 13uf capacitor. My question had to do with whether or not there was a possibility that the perceived increase in the amount of HF energy and presence had anything to do with the tweeter being separated from the midrange branch of the crossover network.

I do know that Paul Klipsch did not care for resistive L-pads, and certainly respect his choice in that as a designer. I'm reporting the change in sound that I witnessed with the removal of the autoformer.

[Erik Mandaville]

And for the record: The illustration of the network above is for informational purposes. I in fact would prefer to get rid of it completely and never look at it again -- despite the fact that to me it looks kind of neat on its own. The aesthetic issue has ZERO to do with the fact that I can't get away from the fact that it is nevertheless still a passive dividing network, consisting of hundreds of turns of thick wire, capacitors, and resistors.

Erik

[Don Richard]

The tweeter filter, which at that time was a first order design, was also connected directly to the input of the amplifier rather than behind a 13uf capacitor. My question had to do with whether or not there was a possibility that the perceived increase in the amount of HF energy and presence had anything to do with the tweeter being separated from the midrange branch of the crossover network.

Hooking the tweeter filter cap to the junction of the autoformer input (pin5?) and the 13uf cap causes the tweeter roll-off to react as a second order network electrically rather than first order. Dr. Eugene Patronis calls this an "improved 3 way network" in the 3rd edition of Sound System Design by Davis and Patronis. Hooking a tweeter filter cap directly to the xover input will allow more tweeter output below cutoff.

[Erik Mandaville]

"Hooking the tweeter filter cap to the junction of the autoformer input (pin5?) and the 13uf cap causes the tweeter roll-off to react as a second order network electrically rather than first order. Dr. Eugene Patronis calls this an "improved 3 way network" in the 3rd edition of Sound System Design by Davis and Patronis. Hooking a tweeter filter cap directly to the xover input will allow more tweeter output below cutoff."

I pointed out once in the past the notion that I thought the original tweeter connection on tap 5 was possibly of a higher order than its frequently referred-to first-order status. As would be shown in a schematic diagram, the "input - 13uf - #5 tap - 2uf - voice coil" arrangement closely resembles a third order design (not saying that's what it is); however,my point at the time was simply to state that I found a greater sense of presence and HF energy, as well as clarity overall, with the autoformer completely out of the equation. For me, that was a more significant improvement overall than what was derived by simply connecting the tweeter filter to the input IN FRONT of the 13uf cap. With the tapped choke gone, the tweeter, as you pointed out, would be connected directly to the input, and a true bandpass would be connected in series with the squawker. That is another benefit, IMO.

In any event, a response plot of the original setting in the type 'A' was posted, indicating an essentially first order slope (if I recall correctly). It was posted in response to my suggestion that the autoformer might be having some influence on the behavior of the tweeter. I also pointed out the value of capacitance normally used for a 6dB tweeter network @ 6kHz would generally be different from what is used in the type 'A' example -- which has certain implications...........

What matters is that we are satisfied with what we have, and that is the case for me until I am able to sort of 'liberate' the system of passive chokes, caps, and resistors.

Erik

[JohnA]

I appreciate the attenuation chart provided, but I was asking for a correlation between the change of inductance created by changing taps on the autoformer and the resulting change to the series cap required for the K-55 series driver when used with the K400/401 horns. Seat o' the pants tells me this change is probably slight between any two adjacent taps. This is probably of academic interest only since I don't forsee changing out expensive caps to change the values by a small fraction just to track the target crossover frequency a little better.

Since the T2A is a tapped Transformer (autoformer) I doubt there is a significant change in *inductance* from one tap to anither, but the windings create a voltage ratio that dramatically changes the apparent impedance of the driver at each tap. You'd need to change the high-pass cap for that change, but not for a change in inductance from the transformer itself.

Hooking the tweeter filter cap to
the junction of the autoformer input (pin5?) and the 13uf cap causes
the tweeter roll-off to react as a second order network electrically
rather than first order. Dr. Eugene Patronis calls this an "improved 3
way network" in the 3rd edition of Sound System Design by Davis and Patronis. Hooking a tweeter filter cap directly to the xover input will allow more tweeter output below cutoff.

That's what I figured. The autoformer creates a node, so the caps act independently, but my one engineering circuits class did not prepare me to understand how that all works. I suspected the second cap steepened the skirt of the tweeter's high-pass filter, but I was never sure.

FWIW, when I rebuilt my mismatched crossovers, I went the classical route (parallel filters) and connected the tweeter circuits of my Type AAs to the amp input terminal. I can't say I heard any difference.

[thumperxes]

I did not realize that there would be so many posts about this. I wanted a simple answer but more knowledge is always better thanks everyone. Now I will add some of my brain dead input that I have with no proof of fact. If the tweeter cap is in line with the mid cap the value in type e would be 1uf if in series, type A would be 1.73uf if I am correct.

I thought I saw some one that said that you could directly connect to input in this case it would be just simple 2uf

Type AA would be 3.29uf?

How much does each change the xover point? To little to worry?

Wouldn’t if I used tap 3 and 4 it be just a parallel connection through the autoformer on the input any way the same as using the same tap or does the resistance value of lets say tap 4 change if 2 speakers are connected to it.

[*-)] [*-)] [*-)] [*-)] [*-)]