Building a new speaker Part 2

March 29, 2019
 by Paul McGowan

In yesterday’s post, we kicked off the first of our 4-part series on speaker building. In this series, I wanted to share with you what we were trying to achieve when we designed the first in our new line of reference standard loudspeakers, the AN3, and how we went about achieving our goals.

One point of confusion with the AN3 is where it sits in our eventual line of speakers—it is in the middle. AN3 is one of three in the AN Series, but we plan two series. The second series will move downward in price and size from the AN line but will have a close family resemblance in appearance and performance. Eventually, there will be 6 models of full range loudspeakers ranging from (we hope) as low as $3K the pair to the mighty AN1. All will have servo powered bass, separate midrange and tweeter. The long-awaited Sprout speaker is also looming on the horizon though it will be bookshelf sized and not powered.

They say (whoever “they” are) that it’s best not to watch the process of sausage making. Perhaps designing loudspeakers would have the same cautionary warning attached but you know me, I cannot resist pulling back the kimono and sharing with you the entire story: the clip leads, the helter-skelter, the computer modeling, the successes and failures.

I know that many readers have this Pollyannaish notion of how speakers are designed—white-coated engineers bristling with slide rules and spouting knowledge like Dilbert, but I am here to tell you that’s a rare scene. I’ve spent many years of my life watching and practicing the art of design with some of the best and it ain’t that way for those that use our ears and calculators to design. It’s a combination of meters and screens and computers and sweat and patience and listening.

In the second of the four videos, which you can watch here, I spend the time showing viewers how Arnie’s 4-way system works and how each of the 4 parts actually sounds. So, if you’ve ever wondered how a multi-way crossover works and sounds this is an instructive video to watch.

The demonstration and the information we gain from this is important to the final design. That is because it’s easier to clone the sound of one system into another if it is done in small increments such as just the midrange and tweeters.

I think you’ll find today’s video illuminating.

Subscribe to Paul's Posts

25 comments on “Building a new speaker Part 2”

  1. A most convincing demo and sound, Paul. However I still have problems to understand how you get the drivers of these four ways phase-aligned. It should be even harder when using the ribbons for the midrange and get the tweeters and midrange-coupler playing phase-coherently at the crossover frequencies. I liked to see the step response and hear music with steeper transients. However if you don’t listen to percussion instruments or plucked guitar and harp strings phase coherency doesn’t seem to be that important.

    1. Yes, I should have done a more detailed video demonstration but we were pressed for time to get this ready for the show.

      The process is a tricky one because much is done by ear. Plucked instruments are a big part of the listening task for the very reasons you suspect. The crossover at the points where drivers meet is the most difficult in many respects.

      In a nutshell, we start with our best guess as to where each driver is going to be happiest based on a measurement of that driver. It’s pretty easy to see on a FR measurement where the driver’s resonant peak is, where it starts to fall off etc. At this point, depending on the same measurements for the driver it will couple to, we decide the quickness of the slope we need and design a network. Still staying on the measurement system we see what happens. If there’s a phase problem it pretty quickly reveals itself as a dip. We can then flip one driver’s phase to compensate or start adjusting the network. But all that is really not telling us how it will sound.

      That’s when we have to leave the metering system and just start listening. The problems become obvious quickly. We can then take the crossover up and down, change sloped, add networks etc. based on best guesses until instruments sound right.

      It’s a tedious process but one that pays great benefits and is actually quicker than any other way I know. Part of the problem with computer modeling which works so well with electronics is that we don’t have good models of everything. Imagine the complexity of building models of the drivers, the baffle, the room, the crossover components, etc.

      In the end listening is the only way to know.

      1. Many thanks, Paul. Obviously experienced craftsmanship meets artistry here. Textbook rules and theories only help creating a foundation for a successful voicing.

  2. Is the midbass coupler dipole or sealed? I may have missed that. If sealed, and if crossed over at 300 Hz, that would mean dipole highs, dipole mids on the BG ribbons, moving to forward dispersion of about 180 degrees at 300 Hz, progressing to omnipolar pattern with decreasing frequency. The frequency at which it becomes fully omni from midbass down will be determined by the cabinet width of course, which I don’t know. Or is the midbass coupler dipole as well?

    Seamlessly combining dipolar and omnipolar segments goes beyond phase and frequency challenges of course. Full range Omnipolar is rare and full range dipole is rare.

    Maggies are full range dipoles, MBL and very few others (Ohm and Audio Physik) are full range omnipolar.

    Possibly using the midbass coupler gives you a better transition from dipolar mids and highs to omnipolar bass radiation than say Martin Logans.

    I can always hear ML transition from dipolar radiating mids and highs to sealed, large omnipolar woofer, making Logans unlistenable for me.

    No offense to anyone who loves them. I understand why they’re loved and also why some don’t like the transition from electrostatic dipole mids and highs. I hear it and understand why I hear it, and I can’t unhear it now that I’ve heard it!

    The transition problem has lead many to believe that a cone woofer can’t keep up with electrostatic mids/highs. It does sound like a transducer speed issue. But it’s not.

    The omnipolar bass is bouncing off of many more room surfaces than the dipolar mids and highs. Those additional and different bass reflections are multiple, delayed, and sobarrive later than the dipole mids and highs.

    Since that is occurring at the crossover from stat to come woofer, the additional and delayed bass reflections from the Omni sounds like a speed issue.

    The Maggies may not punch at the low end like a Martin Logan. But dipole bass, mids, and highs means the reflections of all those frequencies bounce off the same surfaces, so all arrive at the same time.

    That is why Maggies are thought to have fast bass. It dies sound faster, but it’s a room delay issue, not a transducer speed issue.

    Cheers if PS has figured out a way to produce a similarly seamless sound with differing polar patterns. If anyone could do it, Arnie Nudell could.

    As several famous examples illustrate, it ain’t easy.

    1. Arnie’s is a dipole through the midbass. The ribbon is open at the back, the midbass and tweeters are dipoles through using front and back out of phase drivers.

      We chose not to go that route and only have a rear mounted out of phase tweeter.

      One of our long term goals is to provide our customers with the ability to place the loudspeaker pair close to the front wall. Arnie was never a fan of this and demanded people just figure out a way to pull the speakers out into the room. That was a source of endless arguments between the two of us.

      Our AN3 is a sealed monopole design and we use a combination of variable baffle step and midbass coupler to adjust for the distance to the front wall.

      1. I used to think a 4 way system could never sounds as coherent as a good 2 way system or one way in the case of full range driver. Dynamic drivers and crossover technology have come a long way folks. I have a 4 way system the NHT model 2.9 with a crossover of 100-350 Hz to a 6.5 midbass lower midrange or bass coupler driver from a side firing subwoofer to a 4″ midrange driver and there’s no coherency problem. Strings sound beautiful and clear. The entire frequency response is linear, seamless and musical.

  3. Loved this video. It was an Ear Opener.
    I think that both you and Arnie may both correct about whether the mid or high frequency range is the most important part of the of speaker’s sound reproduction. I’ like to add bass into the mix as well because “The whole is always the sum of its parts”.

  4. I am a man more than books, that of the internet, in texts about what was formerly known as High Fidelty, such as those of: J. H. Newitt, William Willman, William F. Boyce, Dr. N.V. Franssen, Norman H. Crowhurst, M. Marin Bonell, agree that when several drivers are installed, for example a pair, what is achieved is an increase in efficiency in 3 dB (SPL), that if one were used alone, and so, although this response is not linear, until today, there are algorithms that automatically enter the number of drivers with their specifications in the respective box, you get the total efficiency of all those drivers, when they are outdoors, but It is also possible to determine the efficiency of a set of drivers already installed in the chosen cabinet. (Wonders of the digital age)

    Now, in the present video, the moderator indicates that by increasing the number of tweeters, a response of lower frequency is obtained by effect of the set of them, than if a single tweeter is used.

    According to the mentioned books, this is not the case, the lowest frequency of the set is given by the lowest frequency that any of the drivers can deliver, (assuming that they are identical) the same happens with the woofers , when increasing the number of them, what is obtained is the increase of the SPL, but does not diminish the response of low frequency of the set, when using several of them.

    Except in the case that the signal is manipulated by equalization, which would produce excessive twisting in the cone, which is not very pleasing to the purist audiophiles

    Everyone can remember what they sounded like, the speakers of yesteryear, especially the Japanese when they were subjected to the tone controls of receivers.

    In short, the use of several identical drivers in a speaker system, what does is to increase the efficiency (sensitivity) but does not lower the lowest frequency that any of them can give, unless using equalization, by which the driver is forced to reproduce frequencies for which it has not been designed (abuse of the Xmax)

    In any case, the lowest frequency that a driver can give is based on the design and consequently on some of the T.S. parameters.

    Reason, does not ask for strength. When a composer needs to emphasize the sound of the contrabass, he does not use a single instrument and tells the musician to exert unusual force when playing, no, he simply uses more basses to obtain the desired effect, for example Beethoven for 9a. Symphony prescribed 8 basses, but that does not mean that by using 8, each instrument will reproduce frequencies lower than 34.168 Hz, (Eb) that the instrument can reproduce.

    The use of 2 woofers in the above indicated model, among other things, is to obtain the tonal balance, and this is achieved with greater efficiency of the woofers, for which two are used, in order to balance the sensitivity of the midrange and tweeters.

    It should be noted that since the 4-way QLS came on the market in 1976, it has taken the Nudell team almost 43 years to take up the important concept of 4-way speaker design.

    Better late than never.


    1. The need for a high crossover point for a tweeter is often tied to it’s lower frequency power handling, not its frequency response. Multiple tweeters effectively increase the low frequency power handling, enabling the safe use of a lower crossover point.

      1. What a pity that you confuse the power handling capacity of a set of tweeters, with the lowest frequency of response that that set can offer.

        You are confusing: SPL with Fs.

        Few symphonic works use several piccolos at the same time, but according to your reasoning, the mere fact of using several of them at the same time would reproduce a lower frequency of response, to the point that many of them at the same time should sound like flutes, which is a nonsense.

        By using several piccolos at the same time, only a higher SPL is achieved, but the instrument’s lower response frequency does not change. The same happens with tweeters.

        This is elementary.

        1. That is an intriguing comparison. Any individual piccolo has a limited frequency range, and an array of piccolos will still be limited to that, only a lot louder! A tweeter is not limited in the same way; you can put low frequency through it but it will not be very effective at converting that to sound. Would a matrix of many tweeters be more effective? My intuitive response would be no, but that seems to conflict with the notion that multiple sub drivers are better for reproducing extreme low bass than just a single one of the same units. I honestly do not know, and would be happy if someone more knowledgeable could chip in.

          1. The tweeters, are designed to reproduce high frequencies within a range of frequencies previously established by the manufacturer, in accordance with the design and type of them.

            In my case, I have a 4-way speaker system designed from scratch by me, whose high frequency section is made up of a line array of 8 Fountek ribbons, Neo 3.0 model, per channel.

            The manufacturer recommends using a lower cut-off point of no less than 3.2KHz, so I can not violate this recommendation, on pain of damaging them, by forcing them to reproduce frequencies for which they have not been designed or built.

            Now, it does not matter if I use 8 as in my case, or 16 per side, they are built to work from 3.2 KHz to 40 Khz, using any other larger number of them, will not cause the lower frequency of response of the set lower of the 3.2 KHz design, even use any amount.

            The use of more or less of them is based on other factors, just to mention a few: the horizontal dispersion pattern that the designer wants to obtain, the SPL needed, depending on the size of the room and its acoustic conditions, the minimum excursion of the ribbons for a given SPL, and in general the coherence and the tonal balance with the rest of the components.

            Let’s apply a bit of logic, if using multiple tweeters would lower the lower response frequency of the set, then the mid-range would not be needed, since the first ones could replace the second ones by increasing the number of them, which It would be cheaper, and we all know that this is crazy.

            With the woofers the same thing happens, increasing the number of woofers in a given installation, only leads to the increase of the SPL, but not to the decrease of the lower limit of the response of each woofer, since this is mainly given by the Fs, the magnetic structure, the Sd and finally the rest of the parameters TS

            Not to put more woofers, I will have underground bass, yes, if I abuse that, I can be deaf soon, by excess of SPL, by increasing the Xmax.

            Do not believe what I said here, investigate it.

    2. “It should be noted that since the 4-way QLS came on the market in 1976, it has taken the Nudell team almost 43 years to take up the important concept of 4-way speaker design.

      Better late than never.”

      Not really. Plenty of Infinity/Genesis/Nudell-designs have been 4-way. IRS Beta, IRS Sigma, IRS Epsilon, Kappa series (bigger models), Renaissance etc. etc.
      I owned some of them. Still miss them..:)

      1. When Nudell sold Infinity to Harman International, after having bought my RS1-B, I lost contact with Infinity, and I never had it with Genesis, so you may be right.

      1. Awesome Paul, my NHT 2.9 only reach 29 Hz and don’t have a built in amplifier. I would have to biamp my speakers to come close but even then I wouldn’t have an amplifier built to match my side firing 10″ sub bass driver the way the AN3 do. I believe NHT did make a subwoofer amplifier and it’s said to be tuned to their speakers so I will keep an eye out for a pair of those mono’s on the used market. I’m presently running the speakers with a single stereo Creek amplifier. I’m sure these speakers have much more capability then the power I’m giving them. The NHT 3.3 use a 12″ side firing sub in a bigger cabinet and those can reach a flat 20 Hz so no sub needed on those. I might upgrade to those someday. My 2.9 can use a sub for home theater or even some musical material.

  5. Seldom have I ever anticipated the opportunity to view the latest videos on YouTube, but this series is the exception to the rule. Awesome insights into the nitty gritty of the design of this new speaker. I truly enjoyed the early visual in the monthly yesterday, and the look of these is stunning!

    Since I am likely more able to swing the costs of the Stellar series, I look forward to seeing those on an intimate level as well.

    Thanks Paul (et al).

Leave a Reply

Stop by for a tour:
Mon-Fri, 8:30am-5pm MST

4865 Sterling Dr.
Boulder, CO 80301

Join the hi-fi family

Stop by for a tour:
4865 Sterling Dr.
Boulder, CO 80301

Join the hi-fi family

linkedin facebook pinterest youtube rss twitter instagram facebook-blank rss-blank linkedin-blank pinterest youtube twitter instagram