Beyma CP25 questions

[Daddy Dee]

Dean,

Thanks for the link to JW's mod.

Al,

Thanks for the BA contact info.

[WMcD]

Very interesting. It may back up my thesis to a small extent.

However, I'm going to let Al give his interpretation of the data and any subjective evaluation . . . should he care to do so. Yeah, Al, wisely, is careful not to do.

Smile,

Gil

[Daddy Dee]

This is one of those times when an audio reference glossary would be a fine addition to the forum.

John was talking about "transients" and how wider bandwidth would yield better transients.

Could anyone help me with what transients are? And talk REAL slow please.

[dubai2000]

It's not quite clear to me what is exactly meant by 'better transient behaviour' (does it translate to a faster/more open/smoother sound?), but to my ears the K-77 sounds kind of harsh (plus not quite extended enough) once you compare it to some other tweeters (the JBLs in my case). So any 'slowly spoken clarification' would indeed be welcome.

Thanks.

Wolfram

[Al Klappenberger]

Hi guys,

"Transients" are simply quick changes in amplitude. You could call them spikes or impulses. Fast changes in amplitude (volume that is) can only happen with wide bandwidth. Any waveform can be broken down into a sereis of sine wave frequencies extending up in frequency (it's called fourier analysis). If you remove the high frequency components a waveform can no longer change quickly. An extrem example is a 1 KHz square wave. It is a 1 Khz sine wave along with every OTHER harmonic ( also sine waves of 3KHz, 5Khz, 7, 9 ... on up) decreasing in amplitude. If you put a filter in the circuit that cuts off everyhting above 1 Khz the square wave will actually become a sine wave at 1 Khz! Every "transient" requires these high frequency harmonics in orter to BE A TRANSIENT! So.. a tweeter than passes more highs has better transient response.

Here's another interesting thing, a square wave that is missing it's extreme high harmonics displays ringing on the top and botton where it should be a flat line! It also shows a slower rise and fall time. That it is, it's vertical edges become slanted or sloped. Another word for this is "integration". A lowpass filter is an "Integrator".

Al K.

[Deang]

Apparently, this principle or concept would apply to amplifiers as well.

[dubai2000]

Thanks Al, the fog is beginning to lift....slightly .

Wolfram

[formica]

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

On 5/12/2004 8:35:35 PM William F. Gil McDermott wrote:

To digress, there is a bit of a debate on whether a speaker frequency response is best characterized by on axis reponse (alone) or overall acoustic response. The latter included on axis response and the sum of all power at off axis.

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

I think there is more and more research (and ABX testing) that shows that the "overall acoustic response" is preferred... as it makes the speaker placement much more flexible. After all, when listening to a speaker in-room versus in an anechoic chamber, it's FR (and primary reflections) are the sum of the on-axis and off-axis components.

It would help to explain the improvements heard with the tweeter upgrades.

Rob

[Al Klappenberger]

DeanG,

"Apparently, this principle or concept would apply to amplifiers as well"

The fact is that this principle applies to EVERYTHING! tTis includes amplifiers, mikes and even YOUR EARS! If you can't hear beyond 14KHz (as I can't), the same integration is happening in your head! Now there's something to think about!

Al k.

[Deang]

Yes, definitely something to think about. Too bad I can't think over 15Khz.

[WMcD]

I think I agree with Al about transient response and frequency. But disagree about "ringing".

There are two ways to look at perfomance of a system. One is to put in a pulse or a step (electrial input) and then watch what the output does (acoustic output).

You will note that this is response in the time domain.

A similar technique is to use a square wave. This has a sudden step up and a sudden step down. NOS has published time domain response of his amps by feeding a square wave and then seeing how faithfully the output reproduces the square wave as displayed on an oscilloscope. Say, moving from 0 to +1 to -1 and back to 0. This is what a square wave does.

Al discusses Fourier construction of a square wave. A square wave can be built up of the series of sine waves. It does take the higher frequency components to get the sharp edges of the square wave.

To some extent this makes intuitive sense in the high frequencies have quick rise rates. One of implications of high freqencies (short wavelengths) is that rise quickly in time. The inference is that if the square wave is not reproduced well at the edges, it is because of lack of high frequency response ability of the system.

- - - -

The other way of examining a system is frequency response, as contrasted to time response (which is transient response). Now we're talking about something common in hi-fi. Everyone wants to know, what is the frequency response of an amp or speaker.

Some spec may say, "Well it goes from 50 to 15,000 Hz with some bumps and slumps in frequency response. The bumps are twice the average power of the average and the slumps are half the power of the average." Hence we have a plus or minus 3 dB as a standard of allowable lack of perfection.

This is usually done with a swept sine wave. More modern systems do use impulse response and crank the math.

- - - -

There is much math to be used if you have time domain information and want to go to frequency response data. It is the Fourier transform. (Some of calculations are redundant. If you can do away with the redundancies, it is called a Fast Fourier Transform or FFT.)

There is an equal amount of math if you have frequency response data and want time domain data. It is the inverse Fourier Transform.

- - - -

I'd like to point out two things which look similar but are not the same. That is ringing in a phyical system, and square wave distortion in the Fourier construction of a square wave from sines.

When you do a mathmatical summation of sine waves, it is true that there are things called "ears" at the sharp edges. The high power EE text books do call them ears, and not oscillations. These are caused by lack of high frequency components, not their addition.

But what Al says may lead one to the conclusion that there is ringing shown in the math. That is not quite true.

Here is why. The ears are not necessaraly resonance as we know it.

One reason we would think (incorrectly) it is a resonance shown in the math is that at the rise of the square wave there seems to be an overshoot. It is a "not quite perfect square wave." Then the little squiggles are damped down somehow and we see a flat plateau of the square wave input. And if we look at the squiggle, it has harmonic components. (Acutally these are the missing harmonic components. So the wave form distortion is from stuff which is not there, rather than things which snuck in.)

"But, there, you see, we have a system resonanting with harmonics in the math," someone will say.

Well enough. But not the whole story.

The top of the square wave in the math settles down to a constant level of 1. All is happy. But not for long.

Now let's look at the summation where the expected square wave is going to transition from the 1 to zero (0). In the math, some TIME (if short) before the drop, there is another ear. And oddly, it starts up before the theoretical drop and gets loudest just at the drop.

- - -

This can not be describing a ringing in a physical system by any reasonable definition. The squiggles appear, magically, when the top of the "not quite perfect" square wave is tracking 1.0 perfectly and there no change otherwise.

- - -

I'm going to dwell a bit on this second ear, and time, to discuss why it is not a ring. It can't be a ring in response to the transition from 1 to 0. This is because the ramp up in magnitude of the purported ring occures BEFORE the event which purportedly causes it! (i.e the drop from 1 to 0). Therefore the Fourier construction math, to the extent there are ears, does not describe an added resonance.

A bell can't ring BEFORE it is hit on the transition from 1 to 0. Also, by inference, the effect from 0 to 1 is not a ring, either, in the math. Even if it looks like one.

- - - -

In fact the squiggles of the ear in Fourier construction is not a ring from ADDED frequencies. It is only that some high frequencies are not there at all. Coming back around the bend, we see squiggles in time. We have to be very careful about making judgment that there are oddball frequencies added by ringing when we look at response in time. Rather, they are not there at all.

Best,

Gil

[Shock-Late]

Another few questions concerning Beyma CP25 ( don't wanna order a brand new pair and pay for them just to find out my old K77M sounded better on my Scalas! )

OK so I'm pretty sure those CP25 ARE better tweeters. They have a very good reputation, and there's no way i could afford the baby cheeks from JBL so beyma seems to be the best "second choice" (after all, I still need a good sub, and here in Belgium I can almost buy a good sub for the price of the JBL tweeters...)

Still i'm concerned about the relative "lack" of efficiency of the Beyma compared to the T35. It's true that on some music i wish i could slightly tame the treble down; the more so since I put the 1uF bypass cap that makes the treble more "hot" and defined...

but i wouldn't want my system to sound "too soft" either. Just want it to sound right.

That said, i'm using EV SM120A midrange horns; those babies have a wide and smooth 120° horizontal pattern; the Beyma could be a good match; i also feel that maybe those horns sound a little bit less "loud" than the K400, and maybe a slightly less sensitive tweeter, such as the Beyma, could just do the trick.

Donno... a bit confused here... any thought welcome

Frans

[Al Klappenberger]

Shock,

Don't worry about the assumed lack of sensitivity of the CP25. The Khorn and Belle are actually closer to 100 dB than the claimed 104 anyhow. The 104 is under optimum conditions. I would bet it's inflated a bit too! I will probably be crusified for saying that, but in any case, you will not notice the difference!

My Belles are modified, but they measure 100 dB at 1 meter with 2.8V input. They are 4 Ohms however so the measurement is sensitivity, not efficiency. Efficiency is only 2.82V at 8 Ohms for 1 Watt. Klipsch heritage speakers are NOT 8 Ohms. They go from 4 Ohms to 30 Ohms! Now how can you claim ANY number as absolute "efficiency" at 1 Watt when the impedance ins't a flat 8 ohms? The answer is that you can't! 104 dB at 1 meter for 1W is just a rough number.

Al K.

[Deang]

"My Belles are modified, but they measure 100 dB at 1 meter with 2.8V input"

See what happens when you start messing with PK's stuff.

I've done my own little experiments, and your number is about right -- however, it makes sense in my case because I'm using tube amps and running off the 4 ohm taps. I don't understand how your sensitivity number can be so low running solid state.

[Colin]

Yeah, at the Klipsch gathering in Arkansas (May, 2004), IB Slamming was explaining sensitivity to Stream when he said it was nominal, I thought meant it was number. I laughed because it seems like in the back-handed way of science, IB Slamming was saying it is only a number! This industry keeps tweaking audiophiles in the dark by NOT including impedance graphs on the back of all their loudspeakers. Only when hobbyists can visually see the loudspeakers electrical load as a continuously variable requirement can tweaking audiophiles truly understand the vital concept of matching an amplifier to a loudspeaker. How can any loudspeaker with wild impedance angles or 2-ohm dips be solely described as efficient?

Nice lesson, professor Gil!

Some spec may say, "Well it goes from 50 to 15,000 Hz with some bumps and slumps

in frequency response. The bumps are twice the average power of the average and

the slumps are half the power of the average." Hence we have a plus or minus 3

dB as a standard of ALLOWABLE lack of perfection.

Of course we know that neither method of examining loudspeaker performance is accurate:

1. Square waves do NOT exist in musical instruments (do they exist in nature?).

2. A frequency response with half or twice the energy (± 3dB) into some parts and NOT others is far from accurate.

It seems to me that Fast Fourier Transform or FFT is closer to the actual waveform of a note. That is what we should be concerned about. How closely does a home movie and music reproduction system reproduce the actual musical note in our living room. How CLOSELY can it match the waveform of the note?

Does your lesson Gill mean that the Beyma, or any, tweeter needs physical or electrical dampening?

[John Warren]

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

On 5/14/2004 10:11:09 AM Colin wrote:

1. Square waves do NOT exist in musical instruments (do they exist in nature?).

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

Actually, pure sine waves do not exist in nature either(?).<?xml:namespace prefix = o ns = "urn:schemas-microsoft-com:office:office" />

In fact, the complexity of musical notes are quite similar to the fundemental frequency and the odd, 1/n, harmonics of square waves.

                   1           1           1

   F(t) = sin(t) + - sin(3t) + - sin(5t) + - sin(7t) + ...

                   3           5           7

[ZAKO]

John.....As I again pick myself up off the floor...You never cease to amaize me. But why is the square wave their any way?

[WMcD]

The question may be why it is of any interest that square waves can be built up out of sine waves. Actually, the other Fourier contruction is a triangle wave made up of even harmonics. But same issue.

The bottom line is that any (I recall) periodic wave form can be shown to be made up of sines.

When you do a frequency response curve, you are measuring how the system responds to pure sines. So, to some extent, if it can pass through all the sines of various frequencies, then you know it will also pass all complex waveforms too.

Similarly, if there is limited bandpass (i.e. high or low frequencies are not passed) then you can predict how a given complex waveform can't be reconstructed perfectly.

- - - -

Underlying some of this is an issue of how to predict the response of something like a crossover having inductors of H and caps of C.

You might know that inductors and caps are little calculus machines. For example, the current through a cap is the slope of the voltage across it. That is to say, the first derivitive. And the voltage across an inductor is the slope of the current through it.

Consider that in a resistor, the wave form (in time) of the voltage is exactly the same as the resulting current. But scaled according to Ohms law. The wave forms of V and I are exactly the same, except for the scaling. None of this slope and calculus is needed.

Now there is something very interesting in calculus. The slope of the sine wave is a cosine function. And the cosine is just a sine shifted 90 degrees. An easy way of expressing this is that the current has shifted 90 degrees from the voltage and is still a sine. This assumes the voltage is indeed a pure sine.

A common misconception is that the phase shift is a time shift. It is not. As long as we're looking at just the H or C.

Now, all of sudden, the calculus problem of finding a slope and the relation of I and V, got far more easy to express. We just say they are related as a phase shift. If we are dealing with a pure sine.

The kicker is that the slopes of complex waveforms would be a real bear to determine. But if we can construct it out of combinations of sines, then things get more easy.

Gil

[Deang]

I used to be pretty good at math, until they started mixing letters in with the numbers. Something about that just didn't seem right. As usual, I have no idea how to work through it all -- but I get the jist of it.

[John Warren]

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

On 5/14/2004 5:49:32 AM Al Klappenberger wrote:

I would bet it's inflated a bit too! I will probably be crusified for saying that, but in any case, you will not notice the difference!

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

The audio industry is ripe with pure lies. The combination of no regulatory body and a technically inept customer base make it so.

No good engineer would ever work in the audio business UNLESS he is willing to work at about 70% of his nominal market value.