The Speaker Guru speaks: John Kreskovsky

Speaker Designer Questions

1. Without giving the standard "weakest link" answer, how important would you rate >speakers as component?
It isn't so much that speakers are flawed, it's more the task they are asked to perform: create an illusion of a recorded event. The speaker is nothing more than a transducer and it can be designed to meet a variety of different specifications with regards to bandwidth, dynamic range, distortion, radiation pattern, etc. However, no single design will fill all needs in all environments. Still, the real problems are the limits placed by the recording process itself. If we rely of 2, 3 or even 5 channel reproduction we are forced to work within those limits. We can create a very pleasing result, but it will always be a far cry from the real thing.

2. How have speakers changed in the past 25 years?
Personally, I don't think things have changed much. We have better components, better manufacturing ability, better means of measurement, but some of the best speaker from 25 years ago is still considered the best today. There has been significant technological advancement and much research in the field, but I don't feel this has resulted in better speaker systems across the board.

3. Do we have to spend a lot on woofers and tweeters to get good sound?
Yes and no. Begin by defining "good". A "good" sounding speaker can be had for a very modest price. What has to be considered is the quality of the driver, not the price. How well do they perform when pushed to the limits? Are they compatible with the design objectives? When trying to reproduce 30 Hz at 100dB+ with low distortion you are probably going to need to spend some dollars simply because of the physical requirements involved. But at the other frequency extreme there are some very high quality tweeters around that compare very favorably with tweeter costing 5 times as much. Still, in general price follows quality. The little gems are hard to find. But be warned, quality doesn't necessarily follow price.

4. What are the top three design parameters you use? (or the top three things you >worry about most)
Are there that many? I guess I could roll off the standard answers: Low distortion, high dynamic range, wide bandwidth, some specific polar characteristic and power response, and perhaps efficiency. How would that be? I'd place distortion, bandwidth and dynamic range at the top of the list. As far as polar response I say that I prefer speakers with symmetric polar response, regardless of the type of crossover implemented, which is why I favor MTM system. Power response is somewhat more illusive as far as I am concerned. Certainly one would hope the systems power response would be a smooth function of frequency, but should it be flat, rising, falling? What is right depends on the environment in which the system is used. Additionally, how much control do we really have over it? It's more a function of system configuration than anything else. By configuration I mean conventional box speaker, line source, point source, planar radiator, mono or dipole, etc. But I wax on. The bottom line for me is that I want a speaker that can play loud, has extended response, and is pleasing to listen to. After that, it is what ever it takes.

5. Do you design on measurement or sound?
It's obviously a combination of both, but over time one develops a recognition of the correlation between how a system measures and how it sounds. The starting point for me is always a good set of measurements. From that the crossover is developed and I use CAD digital emulation exclusively to audition prototypes. This minimizes development time and provides the opportunity to quickly audition changes.

6. What to you think is that "special something" is that makes some speakers sound so good?
Many inexperienced listeners will be taken by overly accentuated response at the frequency extremes but I think most experienced listeners would agree that a speaker that gets the midrange right is what makes a system. By right I mean balanced, open and uncolored. It's much more critical than one might think too. Two speakers that measure +/-1 dB can sound very different. For example, referenced to 1k Hz, a system that has the response from 100 to 1k elevated by 1dB and the response below 100 and above 1k depress by 1 dB will sound completely different than a system that has that response inverted, i.e. depressed between 100 and 1k.

7. What are your speaker audition strong points and weak points, i.e., what >problems do you find most difficult to pick up in listening tests?
Well, I make no excuses. I'm going on 58 and what's happening in the top octave just isn't as relevant to me as it once was. On the other side I find that I an overly sensitive to any honky or nasal sound in the midrange. For some reason I find myself becoming depressed by such colorations. As a result, it is extremely easy for me to pick up.

8. Do you have a theory on crossovers or use whatever seems to work?
Well, it's no secret that I am a proponent of transient perfect design, not that every system I design is transient perfect. It is a design goal that I strive for in specific applications. My ICTA system which is under development will achieve that goal. I know that there are many that feel the approach irrelevant but I don't believe one can hold out the idea that distortion should be minimized and then brush the type of wave form distortion that is introduce by standard crossover under the rug. Would you by an amplifier or preamplifier that introduces a 360 degree phase shift across the frequency band? Why should you settle for that in a speaker system? So, if the phase distortion associated with a crossover can be eliminated without introducing other deleterious effects, then why not eliminate it? One specific criticism of the approach is that this phase distortion can only be eliminated on axis. While this is true, I would also point out that when the response of a correctly designed transient perfect (or minimum phase) system is examined off axis the wave form distortion introduced is still typically less than that introduced on axis by any standard crossover. Ok, so much for defending TP designs. Other than that I shy away from higher order crossovers and try to avoid low mid to tweeter crossovers. While many hold that the optimum crossover would be a brick wall type to eliminate driver interaction I don't find them attractive. The abrupt change from one driver to another seems reasonable if only the on axis response is considered. But when off axis considerations enter the picture the abrupt switch may also result in an abrupt change in the polar response about the crossover. A more gradual transition between the drivers can be beneficial here. When higher order minimum phase filter pairs are used to form a crossover (not to be confused with a minimum phase or TP crossover) greater group delay is also introduced across the system's useful bandwidth with possible peaks in GD at the crossover point. This is also related to why I prefer pushing the tweeter/midrange crossover as high as possibly since the GD introduce for a give crossover is inversely related to the crossover frequency. That is, a 4th order crossover at 2K Hz introduces 1/2 the GD that a 4th order crossover at 1k Hz introduces. In a three way system it's hard to eliminate the GD introduced by the crossover, but by keeping the woofer to mid crossover as low as possible the GD introduced in the midrange can be minimized at the sacrifice of greater GD in the lower octaves. So what this says is that to me it is important to configure the system so that the midrange driver(s) can span as broad a frequency range as possible. This is another reason I favor the MTM format since it allows using two smaller midrange drivers which can safely achieve the same woofer to midrange crossover point as would be possible with a single larger midrange and at the same time allows a higher midrange to tweeter crossover.

9. Do we overstress the bottom octave and area above 20K Hz?
Yes and no. Again, it depends on the objectives. Recordings with information below 30 Hz are hard to find. And flat response in a sealed box to 20 Hz will overload most rooms. But this is a benefit in that the lower the system cut off the more linear the system will be above the cut off point. All in all, low frequency extension has to be considered in conjunction with the environment in which the system will be used. Blindly buying a system based on low frequency response is a bad choice.

10. What one piece of advice would you give every new speaker builder?
First ask your self why you want to do this. If you are interested in building one pair of state of the art speakers and think you can save a bundle designing them yourself, forget it. You are better off buying a commercial product, kit or DIY plans of high regard. If you are interested in a hobby or learning experience, first educate yourself. Then assemble the required tools and from there begin your project. I don't think there is much sense in buying a kit as the most you would learn from that would be how to follow directions. I don't think you have to start with a particularly simple design. If you goal is a full range 3-way system, start there. It will provide the same learning experience as starting with a simple 2-way and working up to the 3-way, you won't accumulate a room full of speakers you can't get rid of, and you will probably save money as well. Oh, and beware of the new driver syndrome. I see more new drivers on the market than cars introduced by GM, Ford and Chrysler in the '50s.

11. When you get cornered at a party by a speaker builder, what is the question >they ask most often?
Can I have a ride in the Ferrari?

12. Will we ever find the Holy Grail in sound reproduction?
Absolutely! I don't think it's as far off as you might think.

13. So, which is it, tubes or SS?
Ever hear of the Fetron tube?

14. What does the future hold for speaker design?
I think the more pertinent question is what does the future hold for recorded music?

speaker icon

Copyright Peter Jay Smith 2005 Return to