Al Klappenberger

Marvel, "If the engineer's playback/monitoring was only 80db, but you listen to the recording at 95-100db or more, the bass and the treble are going to be way hotter than the way the engineer heard it." I don't think that matters. The recording equipment does not exhibit the Fletcher Munson effect. That's only in the human ear / brain at playback. The recording equipment has a flat response no matter the record level. The record level is set as high as it can be set without clipping to ensure the best signal / noise ratio. Fletcher Munson only comes into play when the recording is plaid back. I don't think any compensation for loudness is set into recordings. There is other compensation done though, like RIAA equalization for vinyl recordings, for example. After all, how can a recording engineer know how loud you are going to play his recording? It's up to the listener to adjust his tone controls or the "Loudness" knob some pre-amps have to his taste. Of course, I'm not a recording engineer, I could be wrong. Lee Clinton (Arcytype) might know. He does some of that work. Maybe he's reading this post. How about it? Al K.

Guys, All good insight. Keep it coming! One point I might make is that all Tom did was make the tweeter louder. That is only one of a LONG list of factors that determine what something "sounds like". Some we can easily measure, like frequency response and harmonic distortion. Others, like "coloration" you just can't measure. My question is why is it that something that you CAN measure, tweeter level, in this case, would sound better other than flat considering the "window" through which you listen to live music (no speaker at all) is perfectly "flat"? The higher level of the tweeter is making the highs louder than real. Its got to be compensating for something else you can't or haven't measured. I wouldn't think it would be Fletcher Munson effects if you set the loudness to the same as the real thing. Tom, do you listen at the same levels as real live music? If so, something else is going on. If you run it softer, the bass would need to be kicked up too. If you recall, I actually did do that by reducing the squawker level a bit when I was there! Fletcher Munson might actuall be the answer. AL K.

I'm curious about a point Tom makes here. It would seem to me that a persons' hearing would be a "constant" over everything he hears and would be the same no matter if he is listening to live music or reproduced music. Why is it that Tom would want to hear more highs over his stereo than he hears when listening to live music? I'm going to venture a theory: Maybe when listen to live music they subconsciously accept what they hear is an absolute and accept it as such, but when listening to a stereo they are really trying to entertain themselves by recapturing the days when they could actually hear 20 KHz. If this is the case, it sheds even more questions about the merits of listening tests over instrument tests. It suggest that even a person whose ears have been "calibrated" by listening to the real thing can still prefer something other than "true" reproduction. If this is true, what is a speaker manufacturer to do? Show they make a speaker have perfectly flat frequency response or make it suit what most people "want" to hear. Which philosophy will sell the most speakers? AL K.

These little portable digital recorders seem to be the latest thing. When I started looking for a digital recorder a while ago I couldn't find any of the older 4mm tape recorders that I was looking for. The only way you can get these is on eBay! I didn't need a portable recorder so I got a used Tascam DA-20 for $200. It's DEFINITELY NOT portable! I only wanted the digital recorder because it will record up to 2 straight hours on a single tape and then dump the recording directly to a CD recorder digitally, with no loss. I use it for recording off FM. The 90 minute analog cassette recorder I was using for that was real pain flipping the tape very 45 minutes! Even the new rack-mounted digital recorders use the flash-memory cards instead of tape. These cost over $1000 and I didn't want to spend that kind of money to record off FM broadcasts! AL K.

I think this is the recorder Tom is talking about. Here's more about it: http://www.zoom.co.jp/english/products/h4n/ AL K.

Larry, I don't think there is an "ideal" crossover frequency. From what I hear from others, the place to cross over is where the beam-width of the two horns are nearly the same. This makes for a smooth distribution of sound at the crossover. The frequency range of the horns and drivers chosen factors into it too. Al K.

Larry, I can't answer that one! I know that in my room the peak voltage with it louder than I heardly ever run the system, the voltage was 0.4V RMS maximum. Al K.

I do find that an interesting point! I'm not sure people are replacing them just because they sound bad, but for other reasons, like the fact that they are inconsistent and blow up easily. The one I used in the test was NOT a "true" K77M. It had no markings on it. It was a known good ElectroVoice T-35. People also replace them because they don't go up very high. I don't think that makes a lot of sense though since most of us can't hear beyond 15 KHz anyhow! BTW: This is one of the reasons I picked 5 KHz to use as a test tone. the 2nd and 3rd harmonics are 10 and15 Khz. Chances are you might be able to hear that. I once compared a 10 Khz sine wave to a square wave. I couldn't hear any difference at all! There is also intermod distortion. When you add in the 12.5 Khz tone the lower products added are also in the audible range but seem to go right along with the harmonic distortion in the severity. Basically, even the 3% worst case isn't that bad for a speaker. Impulse resonse. Yes, there's another bag of worms! I might just do some testing on that once I get cought up on other stuff! Al K.

Who, I tested the Selenium several times to verify that. It's not a mixed up plot. Its distortion goes up, then down and then back up with level. Note that the B&C does the same thing, but only just slightly. Al K.

Barry, You can mount the crossover networks anywhere they will fit. This goes also for any attenuators. I was told hashimoto is gravely ill. Is he still with us? AL K.

Larry, Distortion goes down as loudness goes down. The question is HOW loud is loud! I think mosty any of the tweeter I have tested can be considered low distortion at normal home levels. If you are going to use your speakers in a large auditorium, I would avoid the Beyma and JBL. For the home stereo, any of them will do nicely.

ka7niq, Low Q components cause loss in the passband of a filter. This will normally show up as a slope in the passband of the filter as it gets near the passband edge. It seems to be most noticeable at the point of maximum phase shift with frequency change (maximum group delay). The crossover will change slightly, but that's not a major factor. As stopband skirts are normally defined as the number of dB attenuation from the defined passband corner frequency (usually the 3 dB down point), the stopband tends to suffer in relation. The fact is, the attenuation in the deep stopband is not effected much by low Q, only the passband suffers noticeably. So.. The main thing you loose with high ESR components is forward energy going to your drivers. That's why putting god caps with low ESR in a tweeter filter results in a crisper sound. More energy is getting to the tweeter. Al K.

After a lot of experimenting, I noticed that I get more consistent distortion reading if I position the mike right in front of the tweeter mouth, on center and 2 Inches away. Moving slightly off center causes the dispersion pattern to reduce the high frequency harmonics making the distortion read lower than it really is. Mike was B&K 4133 AMP: McIntosh MC50 Analyzer and singal source: HP 3563A AL K. This is the tabulated results:

It's hard to beleive, but it looks like the old T-35 . K77M is the lowest distortion tweeter. This might be because it's the highest efficiency. I can hear PWK's ghost saying "I told you so!"! The highest distortion are the two tweeters with the 1/2 inch aperture to improve dispersion. that's the Beyma CP25 and the JBL 2404. As usual, it's always a balance of compromises! Considering that I found .5V RMS to the maximum level into my tweeter, all of the tweeters have acceptable levels of harmonic distortion. You just can't use the Beyma or the JBL at extreme levels. This makes me feel a bit more at ease considering I have be recommending the CP25 for years and am using them in my own speakers. I ordered B&C DE10s to replace them, but I may not bother doing it now. BTW: If, after comparing these plots, anybody sees someting that makes no sense let me know. I am prone to getting things mixed up. You might see a plot that is identical to one of the others that's grafted into the wrong paste-up. Al K.

B&C DE10 This tweeter shows the same tendency to have a level of lowest distortion, like the Selenium, but not as much so. I did a 1.5V RMS test to this one too. Maybe this is normal?

Selenium D220TI This tweeter increased distortion as level was increased, then went down and back up! I did a test a 1.5V RMS (280 mW) into to see how it would act. I don't understand this!

JBL 2404H

Beyma CP25

Here's the results of the harmonic distortion tests. The input levels were 1.0V (.25V), 0.5V (31 mW) and .1V (1.3 mW). The mike was 2 Inches away. The top of the scales needed to vary with the efficiency of each tweeter and the level. First the good old T-35 / K77M AL K.

Herb, No. I have been looking at the Beyma version of that tweeter but have never even seen one. Someday maybe! Al K.

Here is the first tests on the JBL 2404 tweeter. This is the on-axis plots (black) and 45 deg off-axis vertical and horizontal. The two are virtually identical. This is the obvious dispersion champ! The front trace is anechoic using a short spike, then ignoring everything after the mike hears it to compute the response. Al K.

dkalsu, The JBL tweeters are here. I haven't opened them yet but the boxes look undamaged. Marion, The plots are "dumps" directly from the screen of the HP 3563A analyzer through an HPIB cable to a software plotter emulator. They are on a linear scale because a log scale magnifies the low frequency. Look at the high frequency area is more accurate on a linear scale. Al K.

Nat, I think something is wrong. Look at this definition of harmonic distortion: http://en.wikipedia.org/wiki/Total_harmonic_distortion It shows the calculation using voltage ratios. The 29 dB down 2nd harmonic distortion would be 10^(-29 / 20 ) * 100 or 3.55% distortion. The scales on my HP 334A distortion analyzer are -10dB/30, -20db/10%, -30dB,3% which confirms this. I think that upsets some of you conclusions. Do you agree? AL K.

Dean, You are all heart. I'm touched! AL K.

Nat, Based on the calculations you just did, suggest a realistic level and frequency to do harmonic distortion testing. I was going to use a single 5 KHz tone at 0.5 V RMS. That amounts to 31 mW of power assuming the tweeter is 8 Ohms. BTW: Unlike IM distortion, my HP 3563 analyzer will actually calculate distortion in percent for total harmonic distortion. AL K.