Tips for beginners in loudspeaker development
(Version of 13.03.2003)
Back to "Welcome" page
Since my homepage in online I've got many requests of people who have never build a loudspeaker system their own. Many of them directly want to design their personal loudspeaker and ask for my support.
For many people the reason for DIY is the perception that they can save money by DIY. This is only partly true because you need to spend a lot of spare time. To come to the point very quickly:
If you are not willing to spend a lot of your spare time then loudspeaker DIY is not the right thing for you.
According to my opinion there a 3 rational main reasons for loudspeaker DIY or 3 different groups of interested persons, respectively:
- The do-it-yourselfer/designer:
Despite a huge offer of assembled loudspeakers he doesn't find the "right" thing. This often concerns the size (e.g. must fit into a gap), the outer shape (he hates "boxes") or the appearance (e.g. colour, type of wood) of the loudspeakers. For the DIY/designer the integration of the speakers into the home is essential, technical aspects are less important.
This group just needs a good suggestion (loudspeaker + network + drawings) and a bit of background information in order to allow little modifications that don't deteriorate the
performance.
- The sound maniac:
He just don't find the loudspeaker that sounds "right" for him or he cannot afford his dream speakers. Often the outer appearance is less important, only the acoustic performance matters.
These guys need a warning: even DIY loudspeakers can't do magic! In order to be able to design the desired sound you must be intensively engaged in this theme, incl. basic theory, a bit of measuring technique and a lot of try out.
- The technique freak:
He wants to dive deep into subject. He wants to know how loudspeakers work in detail. This group is the most grateful one, because it has the "right" prerequisite for successfully building a DIY loudspeaker::
- much time and inquisitiveness
- wants to try out things and learn from the results
- does not only go for the loudspeaker itself but also includes other factors (recording quality, room acoustics etc.)
This is just a rough classification and different combinations are possible. Although each group needs a specific support they all should follow some general advice:
- For the first project I urgently recommend to build a proposed project slavishly . This can be a project of:
Only slight modifications either on the dimensions of the enclosure (max. 10% per dimension as long as the toal volume is kept constant) or on the used type of wood (chipboard instead of MDF, possibly 3mm thicker) are allowed. If "bigger" deviations from the original concept are desired some theoretical background will be needed (e.g. studying the book of W. J. Tenbusch Grundlagen der Lautsprecher).
- The selected concept should be adapted to the desired performance:
- Small 2-way bookshelf speaker with a bass-/midrange chassis of 13cm or 17cm diameter for moderate loudness and/or people with limited space, e.g. AUDAX Kit Pro 13 Ti, Cheap Trick 192
- Small 2-way floor standing speaker with a bass-/midrange chassis of 17cm or 20cm diameter with good all-round properties, e.g. ALCONE Lagranche/Laplace, AUDAX Pro 17 TDS, Cheap Trick 175/Mivoc SB180, VISATON VIB 170 AL
- Bigger 3-way floor standing speaker if higher loudness is requested temporary, e.g. MONACOR Straight or Cheap Trick 173 (Visaton)
- Or concepts with a full range speaker, which often sound very musically and spatially for little money but are limited in loudness, e.g. Cheap Trick CT193, Hobby-HiFi Pico Lino or the "Buschhorn" with Visaton FRS8
In all cases I mentioned inexpensive examples which are ideally suited as a base for modifications.
- When you have selected a concept that fulfils your needs it needs to be build. Especially when welding the network you need to proceed very carefully. Errors may not only kill your amplifier (short circuit) but also damage tweeters and squakers if they are not protected from low frequencies as planed.
Ideally you first measure the DC resistance at the connections of the fully wired box with a multimeter in order to ensure that no short circuit has occurred (you should measure at least the DC resistance of the bass loudspeaker). Then you should apply a noise signal with low level to the box but connect only one loudspeaker chassis at a time to the appropriate outlet of the network. This is to ensure that really a limitation of low and/or high frequencies occurs. Only then everything is assembled and you can acoustically evaluate your own loudspeaker for the first time.
- After the first enjoyment that the new loudspeakers work you will ask the question: was it worth all the labour, do they sound better than the old ones? Even if you as a do-it-yourselfer are only too willing to believe that your own work sounds better than anything else the critically DIY will ask himself very soon whether this sounding different is really better? And this is just the beginning because then you will realise your judgement depends on:
- which peace of music you are listening (see evaluation criteria)
- where the loudspeaker is located within the room (see Room acoustics)
- where you are listening (see Room acoustics)
- an so forth
- At this moment the DIY are separated:
- some people are quickly satisfied with the new, different sounding speaker. They are proud of their work and defend it with all their power from ignorants and unbelievers. These people don't need help because after all they are the centre of the universe . . .
- other people are sceptically, wont to understand the differences, to quantify the divergence in order to improve their work step by step. These doubters now have the difficult task to obtain evaluation criteria people in order to distinguish reliably between better and worse.
The evaluation criteria
First of all it must be determined, which piece of music is "well" recorded (from a technical point of view) and how it is reproduced on a "recognised good sounding equipment" in a "favourably shaped room".
This alone is a very lengthy process. Of course also the individual musical taste and the tonal preferences play a part in this, because not everybody listens to music in the same way: some prefer a rich, warm sound others like brilliant high frequencies or a clear midrange. For some people the dynamic aspects of music reproduction or the "timing" are more important and so forth. Therefore it is necessary to build up a collection of - technically - "well recorded" tracks which cover the critical concerns of your personal music taste and way of listening. As an example the human ear is very sensitive to differences concerning the reproduction of human voices as we are trained on this for our whole life.
By restricting yourself on this selection you are able to give a valid judgement on a reproduction chain very quickly. However this judgement cannot cover all aspects. A detailed discussion of this subject can be found on the page Limitations of music recordings.
Currently I use the following pieces of music to judge the quality of a reproduction system (I mainly look for sound balance and prefer a rich, warm sound):
No | Artist | Title | CD | Year | What to take care of |
1 | Bobby McFerrin | Blackbird | The Voice | 1984 | Plausible voice and noises (e.g. breathing, claps) |
2 | Jefferey Smith | Eleanor Rigby | A Little Sweeter | 1997 | Very rich voice, muffled piano in background |
3 | The King's Singers | Back In The U.S.S.R. | The Beatles Collection | 1986 | Vocal bass not too dominant, discrete contributions easy to localise |
4 | Vocaleros | Superstition | Vocaleros | 1997 | Really funky, it must be easy to localise each contribution |
5 | Brent Lewis | Mumbo Jumbo | Pulse . . . Where The Rhythm Begins | 1995 | Very wide spatial reproduction, easy to localise |
6 | Talking Horns | Johann, der Tango kommt | Fisch im Wasser | * | Recorded with 2 micros only |
7 | The Oscar Peterson Trio | You Look Good To Me | We Get Requests | 1965 | Rich walking bass |
8 | Oscar Peterson | Dream Of You | Reunion Blues | 1972 | Bass left and behind speaker base, sober percussions, VERY wide vibraphone (single tones must be followed easily, even across centere), muffled piano may not sound muddy |
9 | Melissa Walker | I'm A Fool To Want You | May I Feel | 1997 | Very spatial without much "pseudo" reverberation |
10 | Jennifer Warnes | Somewhere, Somebody | The Hunter | 1992 | Very wide virtual stage, both singers in centre (follow their spatial interaction) |
11 | Keb' Mo' | Just Like You | Just Like You | 1996 | Very earthy, is must be able to separate the (co-)singers |
12 | Holly Cole | Jersey Girl | Temptation | 1995 | Very fat bass, all instruments very proper, warm voice, slightly purred/grumbled |
13 | Mighty Sam McClain | Too Proud | Give It Up For Love | 1993 | Very crisp cymbals, even at loud sections not annoying (singer should not be mixed up) |
14 | Marla Glen | Personal | This Is Marla Glen | 1993 | Throaty, dark voice, shiny trumpets, sounds addictive |
15 | Hugh Massekela | Stimela | Hope | * | Loud sequences must have punch, realistic voice |
16 | Yello | Tied Up | Flag | 1988 | The masterpiece of the Swiss sound puzzlers. Must blow your brains! |
Modifications
Only when you have worked out these evaluation criteria it makes sense to think about modifications. Examples are:
- Modifications on the frequency deviding/manipulation network (FDN)
- Exchange of a chassis (e.g. tweeter) together with necessary adaptation of the FDN
- Higher quality components for the FDN
- Different low frequency strategy (e.g. bass reflex instead of closed box) or bigger/smaller enclosure, respectively
- Higher effort on enclosure construction (stiffening/damping)
Essential with these modifications is to manipulate only one loudspeaker and to keep the other as a reference (ensure in advance, that both "base conditions" are really identical). This allows to make before/after comparisons to judge the effectiveness of the modification. This is necessary as the acoustical memory of humans is pretty bad an lasts only for a few minutes.
However, these modifications require some basic theoretical knowledge of the respective field (e.g. FDN or enclosure design). This can be obtained by studying special literature or internet pages.
All in all this is a quite time consuming task. But this stage is necessary to design your own loudspeaker system successfully. That's why my slogan is:
- Start with a small, manageable project, learn by modification and then start a bigger project. Champs don't fall from heaven! (I hope this 1:1 translation from German makes sense in English, too)
Measurement technique
In general there as several development stages:
- Rough layout (which chassis are promising, how big is the enclosure etc.)
- Detail design (how to position the chassis on the enclosure, basic layout of FDN)
- Fine tuning (how the final FDN looks like)
While the first 2 steps can be done on a desk for the 3rd step the continuous listening, modifying, listening again, modifying and so on is required. This 3rd step in effect is a loop that is followed as long as the result is satisfying.
For this it is essential not to loose the "right" direction. Only if you have a reliable starting point the following optimisation can be successful. According to my opinion measuring technique can help to find this reliable starting point quickly thus speeding up the whole development process.
At least until today measuring technique cannot yet describe the hearing sensation in all details. This is why at the end the subjective impression is always more important than objective measurement results.
The answer to the question "measurement technique - yes or no" essentially depends on the costs. Who is really willing to spend 500 EUR on measurement equipment when a loudspeaker system for 300 EUR should be developed? Therefore the amateur wants inexpensive systems. Most available systems are based on a PC with (built-in) sound card. Although the sound cards and the cheap microphones are not ideal they are still good enough to find a reliable starting point for the further subjective optimisation as non ideal frequency responses (at least for the sound card) can often be compensated. More difficult are non linear frequency responses of microphones because individual calibration spectra are not feasible in the amateur price range below 100 EUR and errors are typically within +/- 3 dB.
Overview of very inexpensive measuring systems (arranged alphabetically):
Name | Remark |
AudioTester V2.0 |
Win 9X, 25 EUR, very powerful but not easy to use (stereo input required) |
HobbyBox V3.0 |
Win 9X, old version 48 EUR, not easy to use |
Praxis |
Win 9X, free limited version, also good for car HiFi |
SB_OCT |
DOS, 20 EUR, good for room acoustics, proposal for measuring microphone |
SpeakerWorkshop |
Win 9X, freeware, not easy to use |
WaveTools |
Win 3X, freeware, not for room acoustics, data export via clipboard to EXCEL |