Interview with Joachim Gerhard of Audio Physic

By Lars Mytting
29 Jul 1996

Low tweeter crossover frequency

- In the Caldera, you cross the tweeter at 1300 Hz. That is one octave under what common theory recommends today?

- I think many cross the tweeter too high. Conventional theory has always argued that we should stay away from the fundamental resonance of the tweeter. On the other hand, mr. Thiele has presented new crossovers that are crossed exactly at the resonance point. I think that is very interesting. The resonance point is a very well defined area. The only you have to get rid of is the resonance itself, by choosing the right Q and a RLC circuit. Of course, the tweeter has to be very broad-banded, and very linear. But tweeters are more robust than many seem to believe. We have made tests at ScanSpeak with the tweeters used in the Caldera speaker. The voice coil was heated to 320 degrees and it did not break. We put 120 volt in, and we could see sparks in the dark, but it still did not break.

- What is your favorite tweeter for the moment?

- In the Caldera, we use a selected version of the Scan Speak 9500, a variant of the 9700. I feel that the copper ring in the 9700 causes an over-emphasized treble, so this 9500 has a aluminum ring instead. But it has to be selected samples. A problem with softdomes is that automatic coating by machines does not give optimum resolution. You have to do it manually by coating the edge and the top, maybe with different coating. So it depends a lot on skill. You can have two girls building the same tweeter. One of them makes tweeters with a tolerance of 1 db, another girl builds the same and the tolerance is 4 db. But when you want the best, there has to be some handwork involved. I am very satisfied with the Scandinavian driver manufacturers. They are very flexible, and they allow us to discuss different solutions. Just during the last six months, I have visited Seas three times.

- Do you find their new magnesium drivers interesting?

- Oh yes, very much. We will bring a new Avanti that has a variant of the magnesium 5" for midrange. Of course I was not very happy with the top resonance, and we had to use a special circuit to suppress it. When we solved it, it was the best sound I have ever heard. The energy storage is so low, only 1/3 of polypropylene. Also, it is very clean in terms of harmonic distortion, especially between 50 and 300 Hz, where it has 10 times less distortion than traditional drivers.

- How did you get rid of the resonance peak?

- I developed a special double tuning circuit. With it, we got able to tune the resonances very exact. But you have to be very precise, the values are very critical for it to work. It is difficult, but I can assure you that it is possible to tune this driver to a level where the resonances are not audible. We also found good correlation in the waterfall diagram when we used one-third octave measurements. When the time delay in the peak is less than 0.5 milliseconds, it has disappeared in the audible sense. But don't be satisfied until it is totally gone. If it is not, it will superimpose, so be extremely careful and honest when tuning it.

- And the $10.000 dollar question: What does the circuit look like?

(Gerhard answers the question with a quick drawing, here converted to ascii format:)

---------|-------------|---------- +
|             |
IND 1         IND 2
|----RES 3----|
|             | 
CAP 1         CAP 2
|             |
|             | 
RES 1         RES 2
|             |  
|             | 
--------------------------------- -

- The 5" magnesium drivers has two resonance peaks, so the secret is to use this double RLC-circuit to tune them away. Of course, I cannot give any component values, but any speaker builder without measuring equipment should use the following, old trick when tuning: Take a tuner and set it to an empty station, and listen to how the driver reproduces the noise. This makes you hear the distinct pitches of the cone, and it tells you a great deal about the resonance behavior. It is really a fantastic method. When correct, the driver should sound like water, or like wind, and have a very linear spectrum.

- How flat do you go in frequency response?

- I go to +/- 2dB. When I reach that level, it is fine. I don't go to extremes to get it more flat, because that adds more crossover components, which has disadvantages. We have made some experiments where we equalized the speaker to +/- 0,5 dB or even better. Of course, it was a little bit more coherent. But I found that +/- 2 dB is good enough. After all, frequency response is only tonal balance, energy level. It is not time domain behavior or distortion, and there is not so much gain in going to extreme lengths. Other things are just as important. For instance, I look very much into time domain behavior, because that tells you the energy storage, which is a crucial parameter.

Gerhard's first rule in speaker placement is to prevent the first reflections as much as possible. Instead of the usual rule of placing the speakers so that they correlate with one-thirds of the room dimensions, he suggests that the speakers are placed in the middle of the room.

- Room placement is totally underestimated. What is usual today is just an aesthetic placement. It is totally stupid. For me it is no question that speakers, except dipoles or line sources, benefit of being listened to in the near field. I don't like time delay; the sound should be heard only once. The only things that comes from the side is delayed and disturbs the image. The sound gets broader, more indistinct. And I am very concerned about the three-dimensional touching feeling in speakers. That is what makes you visualize the performance, so I find imaging a really fundamental information of the music. Tonal balance can always be discussed, but this feeling of being there is very crucial.

So when placing speakers, just think about yourself. When you are going to talk in a room, do you go to the corner? To the back wall? No, you go to the middle. Try it out yourself. Make someone talk in your listening room while they walk around, and find where the sound is most pleasing. Good placement is for free. But people don't seem to realize it. They buy speaker systems for 50.000 dollars and then they say: Oh, I can't put my speakers where they sound best, because my wife does not like it!

- Is it mostly the radiation pattern and the absence of out-of-phase dips in the movement of air that makes good imaging qualities?

- I would say so, yes. Horizontally and vertically. Therefore, we also measure the response from the side and the top. We place the speaker on a turning platter and measure it from several angles. What we found is that the more air that moves in phase; the better it sounds. It is a kind of bubble in time and space, which the listener is immersed in. The two drivers add into a kind of acoustic hologram. So my ideal radiation shape would be a kind of kidney-like pattern.

- Why don't you use the D'Appolito placement of drivers?

- I don't like it, because I want freedom in the radiation pattern. I think that is more what it is like in nature. I designed some kits with D'Appolito configuration for Germany in 1981. It was a big hit; the power handling is great, you have two woofers, and you get the typical vertical beaming that creates tremendous projection. But all designs I have heard has this very typical effect of a band that is constricted vertically, and that pushes the image forward. So I find the focus too strong, I lose the depth. That is subjectively, of course. I don't say it is bad, some people use it in all their designs, but I find that it is difficult to tune, it creates one problem more.

Several of the larger Audio Physic speakers has two woofers mounted low on each side of the speaker. Gerhard explains that this is done to get a low-range response with better phase behavior.

- We invented this system because we wanted an in-phase signal in the bass. I have always tried to avoid the double signal pass that you usually get from conventional systems; with one pass going directly from the membrane, the other over the floor and the ceiling. This placement near the floor also gives you approximately 3 to 6 dB gain. So one of the advantages is that you can use lower mass in the woofer and stronger magnet systems. But from 200 or 300 Hz you run into problems. You get a resonance. The magazines say that you shall not put the woofer to the floor, because it gets boomy. But we found that the problem in fact is in the deeper midrange region. So if you make a very low crossover frequency, say 125-180 Hz, you can avoid this problem and have the advantage of the room gain. Also, you benefit from the fact that the center of the gravity is very low. The system is push-push, so the energy inside is canceled, and the loudspeaker gets pneumatically stabilized. And the effective baffle is large, because the woofers "see" the side of the speaker. So you have tremendous advantages. The only problem I have found, is that the crossover has to be very steep to get rid of the boominess.

- How crucial is the localization near the floor?

- Very crucial! In the Virgo, we found that from the floor to the middle of the woofer, 40 centimeters was ideal. But place the woofer 5 centimeters different, and you get a totally different result. I would encourage people to build a simple box, fasten it on a string or something, and then raise it and lower it while listening and measuring from the listening position. Maybe you will find some interesting results!

- What part of the frequency band do you find most difficult to tune?

- Oh, the midrange! Of course. The ear sensitivity is at the best there, and you are in a wavelength of between 10 to 30 centimeters, which also is the dimensions of the baffle. So you get dips and peaks. Also, when you measure the energy in natural music, you will find a peak at 100-800 Hz.

- What solution do you have to get enough energy in the octaves between 80 and 320 hz?

- First and foremost, this is a range where you really can hear the advantages of harder cones, because of the demanding dynamics in these octaves. But I try to cover this with a regular midrange driver, with a crossover frequency at maybe 150 Hz. I find this better than trying to let a larger driver, maybe a 8", cover it. Because then you usually have to cross it over higher in the midrange, and that is very difficult. I find simple things so challenging that I see no reason in going into even more complicated things. First I have to master simplicity, Gerhard says.

Epilogue: The do-it-yourself part of this interview
All interviews with famous personalities usually include small talk like "he says thoughtfully, lighting a cigarette" / " ... so that brought an end to it, she said, sipping slowly to her glass of Calvados..."

Since this interview is antiseptically clean for chic phrases, details on ties, coffee etc., we will give the reader the following text (which also is true) to fill in wherever he/she feels that a poetic turn is appropriate: (Use the ctrl+x and ctrl+v buttons in most wordprocessors):

"says Gerhard, a man whose passion for high-tech becomes evident through his enjoyment when winding his mechanical watch. No batteries, please. No cheap solutions, please. We're German".

Lars Mytting

About the
author

Print this
article

Email this
article