What do the transformers do in Heritage x-overs?

[Deang]

Well, I know what transformers do -- but I'm having a hard time understanding what they do in a crossover.

[Deang]

Never mind.

Amazing -- I actually got the search function to work!!

[Jim Cornell]

Yea bud i just asked Trey the same thing down in odds and mods, they lower the power to the squawkers!

Jim

[Al Klappenberger]

DearG,

Transformers reduce the level to the driver by reducing the voltage applied to them. They also increase THE BACK VOLTAGE generated by the driver as seen by the amp. This impoves the amps daming factor directly to the driver. A resistor pad (L or T pad) divorces the back EMF from the amp and makes the damping factor ineffective. The bg trade off is a transformer cose 20 dollars compared to the L pad that costs 20 cents! Which ONE do you think the manufacturer will want to use! You can't blame them for choosing the L pad!

Al K.

[WMcD]

You can read about it in the form of an article by the Celebrated Dr. K.

Page 1 attached, the other pages follow.

Gil

[WMcD]

Page 2 attached.

[WMcD]

This page 3, the last one.

Good reading,

Gil

[Al Klappenberger]

Thanks for posting the PWK article on the subject Gil. The bottom line seems

to be that the transformers just sound better than resistor pads. I have a

good friend who is a retired acoustics engineer and a veteran of many

loudspeaker projects. I made a deal with him to swap a set of the UTC 3619

autotransformers I use in my netwroks for an item he had that I wanted. After

he replaced the "L" pads in a set of speakers he built that resembled the

Heresy design, he was astonished! He claimed he couldn't believe what he was

hearing! Neither of us expected it to make that much difference, but it

apparently did. The damping factor explanation is my theory for what is going

on. It is only my theory. I don't believe there is any real scientific proof

that this is the reason though.

Al K.

[Cut-Throat]

ALK,

Which transformer do you prefer the UTC 3619 or the T2A?

[jnorv]

Are these transformers wound just like standard transformers? With both a primary and secondary set of coils? If so, what are the typical winding ratios?

Jim N

[WMcD]

To answer one question, an autotransformer just has a single winding which is tapped in different places.

There is something odd going on with calculations. I remember reading this long ago. It arises from the fact that the resistive element of the windings act like a potentiometer or classic voltage divider. Just a little bit. Not to be confused with the gross effects of an L-pad.

For example, in the step down wiring mode, the input is across all the windings and the output is across, say, half of them. If that was done in a normal transformer, this would result in half voltage. But, in the single wiring mode, part of the voltage across the "secondary" is due to resistive dropping. But there is also the magnetic coupling.

There is something to be said for magnetic coupling (it is better) as compaired to a two part transformer.

The autotransformer can also be used in step up mode. PWK got a patent on that in connection with the "Valerie" type bass horn. That is why you'll see two patents which look the same except for the claims. I can look it up if anyone is interested. The "divisional" patent used a step up mode to increase voltage sensitivity of the K-77. Evidently the K-33 bass driver and bass horn did not have much loss, it matched the K-55, and the K-77 was the weakling.

There is a good technical reason for the use of the autotransformer. As Al. K. points out, it is related to the damping factor, but a little different.

Sorry for the lack of diagrams, if someone is interested, I can put some together.

The original Wente and Thuras (or maybe just one of them) patent on the Bell Lab 555 (PWK named his mid driver in honor of it) driver shows how the acoustic impedance of the driver should be just in the middle of horn load it is looking into. This give best energy transfer. You'll see equations and descriptions on how the optimum throat size is set by electrical parameters of the driver.

But there is something just as important; and the same thing in a different way.

The acoustic load from the finite length horn is not constant down near cut off. Harry Olson shows this. It bounces up and down. So, it is best to get the driver acoustic output impedance in the middle of the high and low spots of the horn load. That way neither the highs or lows of the impedance hurt you too much in energy transfer. Now we've done that optimization anyway, but it is an issue.

To which you say, "Fine, but what does that have to do with autotransformers?"

The remarkable thing is that the acoustic output impedance of the driver is influenced by the impedance of the electrical source driving it.

You'll note that the PWK patents on the Valerie and MCM-1900 cite to Wente and Thuras's article on Audio Perspectives (part of the Klipsch papers). Buried in there are equations which lead to one, not set out there, for optimum throat size based on basic electrical parameters of the driver. There is an "n" factor. That involves the electrical output impedance of the driving amp.

Generally, if you have an amp with zero or near zero output impedance, n=1, and drops out of the throat size equation. But if the output impedance is equal to the voice coil resistance, n=2, (this is from memory). The optimum throat size doubles!

There is a way to use T-S parameters of the driver to calculate optimum throat size. Don Keele gives them in one of his AES preprints. They are derived from the W&T article. He notes the "n" and sets it at 1 for "modern amps." Bruce Edgar brought the preprint to our attention in his SpeakerBuilder articles. There is a bit of math, not shown by them, which gets you from one set of equations to the other. It took me a while to figure it out.

You have to do some imagining here. This means that the acoustic impedance of the driver output shifted, perhaps away from being at the midpoint of the bumps. You've messed up the optimization by playing with "n", which is based on the electrical driving impedance.

Now, going back to the autotransformer. It is presented as a good alternative to L-pads used as attenuators. So what is wrong with L-pads? The answer is that they, effectively, put a resistance between the amp and the electrical input to the driver. (This is, in theory, like an amp with an increased output resistance (impedance); or poorer damping factor.) Therefore, the "n" is shifted up from 1.

Now, that doesn't necessarally mean that things are bad if the throat size has shifted accordingly, or the parameters of the driver are different and things are brought back into alignment. But it is a difference. Perhaps this has been accounted for in the new crossover and driver for the updated K-Horn.

On the other hand, the autotransformer set up will decrease the voltage to the midrange driver without inserting an appreciable resistance. In fact, things work out that it reduces the effective impedance presented to the driver.

This may well account for the reduction of swings in raw output shown by PWK in his TTWA publication. It may also account for differences heard by Al. K.'s associate.

Well, I've rattled on again. More than most wanted to know, less than some few wanted to know.

Best,

Gil

[Trey Cannon]

Look at the pic.

A resistor will change the output level of the speaker more at some frequencies than others.

The transformer will not do this due to the low resistance in the coils.

[WMcD]

I'll rattle some more.

The original question was what the autotransformer does in the cross over. You have to be up on cross over design. I noodled though this.

You may have a choice of putting the autotransformer before the crossover, or after the crossover, or in the middle of the crossover circuitry. There is more to this than I can explain here; if you have an E.E. background you'll see it.

Let's look at the A type and the midrange. The crossover is a single cap, and we therefore have a 6 dB per octave crossover feeding the midrange K-55. More accurately, it is feeding the autotransformer. Because it is wired in step down mode, the input impedance to the single cap crossover is roughly 100 ohms. This allows the power transfer to be reduced, or attenuated. This is generally shown in the impedance diagrams in the Audio magazine review by Richard Heyster of the K-Horn.

Now, you look at standard equations or programs to pick out where a 400 Hz cross over can be accomplished. Accurate ones ask for a speaker impedance. If it is 100 or so (higher often), the value of the cap is lower, than if it is 10 ohms. So we've found that we use a lower value cap, less expensive, if the crossover is before the autotransformer.

Plus, it has the benefit that if the amp has some d.c. offset, it is not going to saturate the windings of the autotransformer. And gross amp failures will not blow it out. The cap gives some protection.

But what if we need a second order filter (12 dB per octave) with an inductor. I think you'll find (in equations and programs) that the value of the inductor has to be quite large if the revised filter is "looking" at the input to the 100 ohm load presented by the autotransformer. What to do? We have a smaller cheaper cap, but an expensive, larger inductor.

The answer is to move the inductor to the driver side of the autotransformer, where we're in a setting of 10 ohms. The value of the inductor is lower. It is cheaper, and generally, less lossy. You'll see this is what is done in later K-Horn and other crossovers.

This means that the autotransformer is in the middle of the crossover. It does mean that standard equations and models don't work. A real headache. Yet it can be better in performance.

Gil

[WMcD]

Trey's post came in while I was working on my second one.

I don't disagree with him. Yet we're talking about different parts of the system.

Gil

[tubelion]

Hello trey,

If autotransformer is good, L-pad is bad. Why does the newer design use L-pad?

Thanks

Tubelion

[johnnyp]

Can different taps be used on the 3619 or the T2A to attenuate the driver more or less without altering the crossover frequency?

The reason I ask is because I prefer the sound of the K-55v without the the bugscreen but it is louder (too loud) without it.

[Al Klappenberger]

Cut-Throat,

To answer your question, I think the 3619 is better. It was designed by

Universal Transformer to be better at my request. The same company also makes

the T2A exclusively for Klipsch. The 3619 has heavier wire, more core area

and is bifilar wound to tighten coupling. It also has an additional output

tap allowing for more combinations of level settings. The bottom line though

is that you will probably not hear the difference between them. The biggest

practical advantage to the 3619 is that it costs less than the T2A because

you can get it directly from UTC!

Gil's discussion about the impedance matching may have something to do with

the audible difference between the autoformer and the L pads. I am not a horn

expert so I have no insight on that point. So research into what the

autoformer would do in a direct radiator loudspeaker might answer the

question.

I personally believe that the choice of L pad over transformer in the AK-4

was to keep the cost down. PWK is no longer there to force their feet to the

fire! I believe also that most people's knee-jerk reaction is that a

transformer causes ringing and all sorts of other band things compared to a

resistor. The top of the line McIntosh power amps use autoformers in the

output stage and the lesser models do not. They must not be too band a thing!

I don't think anybody would say that the big "Mac" amps ring. My old Mc250

doesn't. I have tested it myself.

Al K.

[Al Klappenberger]

John,

Yes, in the Klipsch designs you can't move a tap without other changes or the crossover frequency will move. This is where the 10 Ohm swamping resistor in my network comes into play. You can move the taps as you like becasue the impedance can't go above 10 Ohms no mater where to set the taps. Without the swamping resistor it can easily go to hundreds of ohms.

Al K.