Suffering a PC virus (of sorts!)

[prickears]

It would seem to me, Nowdays, Its just too easy to rely too much on PC programs to 'Design' speakers, without understanding, the basic science/physics/art behind it.

I'm the first to put my hand up here, And I doubt I'm the only one.(Que here)

See, the thing is, (I think), a fair chunk of the problem is, different schools.
Not everybody out there see's eye to eye on every issue, and therefore, solution to the issue. ART, if you will. (consult PC here).

This is clearly because, there is generally more than one solution to the problem.

Case in point. I have a project in its infancy at the moment. DIYmember 1 says,
use the natural rolloff of the mid, to a, keep the XO simple and, b, out of the 'mids'.

My school of thought exactly.

DIYmember 2, recommends crossing an octave lower, to help transient response (I think), and power handling.

Not my school of thought, but no doubt equally correct.

And my PC sits, clearly baffled by the question, let alone, the solution!

1st solution breaks the rule 'thou shall not break the decade rule!' (ie, if three way, and woofer mid xo @ 250hz, then mid to tweet xo shall be no higher than 2500hz (250 x 10= decade 2500)). And XO becomes 5000hz(but keeps out of the midrange).

2nd solution breaks the rule 'thou shall not XO in the mid range!' (for the point of this eg. 500 to 5000 hz) and crosses right where a small mistake in the XO calc = very audible X0 (but retains transient (I think) response and power handling.

At the end of the day thats what makes this whole journey so much fun I reckon.

And my PC, doesn't understand 'fun'! But it still seems to know more than me!

So, who's driving who, in your PC vs real Knowledge, relationship?

Why not consult your computer!

[LineArray]

You express exactly what i am feeling sometimes too.

But i must admit: There is a german forum - name des not matter -
where the ratio of contributions like
"I have simulated that problem for you ..." is very high.

I hate it, when it gets too much.

For shure many problems can be simulated with more closeness to
reality than in the former dacades, since computational power and
algorithms improved.

I am a great fan of simulating electro acoustic problems using
electrical equivalent circuits, e.g. the work of Thiele/Small
is based on that.

But no program does the d e s i g n for you. Which BR alignment
to choose may depend on several factors, including the
preferred position of the loudspeaker in the room.

Having knowledge and computational tools in the background
as "helpers" and also knowledge of "rules" is worthful.

But a good design is mostly about h o w rules are bended, broken
or circumvented.

There is no software having a button
"Build optimum speaker now?" on the GUI, is there ?

Software -to me - is useful in rapidly testing ideas
"What would happen if i made the baffle that small ..."

Most Software is only good at a certain aspect of the problem.

But in electroacoustics a manyfold of aspects play a role.
Example:

Strategy A:
Minimize error (distortion or whatever) by unweighted measurement.

Strategy B:
Design error to be least audible.

Question:
Does your Software in use know about auditory effects ? (Normally not.)

Is your software able to predict how this or that particular change
in frequency response is related to the response in a particular room,
taking into acount polar dispersion ?

What does your software know about this certain driver being
used above it's "pistonic" frequency range of movement ?

Of course there is software which can do ... but you have to feed it
with lots of data which is often not available.

Design and building of a good loudspeaker in the end has to deal
with the real thing, that is for shure.

Kind Regards

[prickears]

Yes,

Thats exactly it.

PC is a useful tool, it can tell you what you want to hear, but can't tell you what you will hear. Such is the art of the loudspeaker.

PC says, I can tell you what you want as long as, You have the ability to test a driver using me as a tool, then YOU must must also like what you hear (not always the case!).
then I may plot you a solution that maybe correct, BUT may not be the sound your looking for, due to inherent sound of driver!

If only it was possible to sample sound charactaristics of a given driver!

I wonder if that is why there is so many different schools of thought out there, After all we all know beyond any doubt, what we like to hear!

What sounds like chocolate to me, may not be to the taste of others, and viki verki.

Having said that, some designs, clearly have more followers than others.

Have any of these classics been built almost entirly from computer modeling, or painstaking listening and tweaking?

Both would be my best guess.

Art AND science.

Thanx for your input here! And best regards to you too.

[gornir]

I couldn't agree more. Very well described.

[prickears]

Thanks Gornir,

I feel like an alien sometimes, nice to know i'm not alone!

Mick.

[prickears]

O.K so lets look at it from a german car manufacturers point of veiw...........

BMW m3, is a classic example, newer v8 model eats the old straight six on paper, but........From a purist driver point of veiw, quite the opposite from what i'm led to understand. You NEED the computers on the new model to do what the old model could do without!

Again Porsche 911 vs Nissan GTR, GTR relies so heavily on its computers, that if one should fail mid corner at your local supermarket carpark, its all over! Meanwhile, the Porsche driver was probably a fraction slower to get his litre of milk, but lived to tell the tale should his computer ever fail!

So be honest, who represents Porsche and who represents Nissan?

Those lucky enough to drive both, generally have this to say....'The porsche is for the purist, it feels alive, The Nissan, is probably faster, BUT, it feels like its wiping your butt for you the whole time.'

Which would you buy?

For me its the Porsche every time I'm afraid!

That is, and always was, what made it a Porsche in the first place!

I feel, it's our ears that make the decision final, Not a computer, even if the PC say's 'yep thats a dead flat response and everything's zen'!

PC, Great tool to help get you where your going, but don't forget to feel your way through a bit too.

[5th element]

When it comes to a design there are a few factors that automatically dictate what you are allowed to do and where you can do it.

For example if you were going to make a three way out of a 6" bass driver a 4" mid and a good quality 1" dome tweeter. You might as well say, why bother with the mid then! And quite right so, good domes will cross over at around 1500hz with a steep crossover. (This places the crossover in the crucial mid-range. It is my experience however, that if the crossover is well designed, it is largely inaudible.)

However we are approaching this from the point of view of creating a three way.

Using such drivers would give you a lot of freedom. You could most likely cross between 200-1000hz, from the bass to the midrange and then from 1500-3000/4000hz perhaps, from the midrange to the tweeter.

Where to put the xover? Do you have any goals in mind? Do you prefer certain type of crossovers?

The first thing to do is of course measure the drive units, do they have any anomalies you'd like to avoid.

The bass driver might have a few wiggles in the impedance and show small resonances at say 600hz. As a direct result there is a slight hump in the harmonic distortion. Are you aiming for perfection? This hump might not be audible, but you'd rather not live with it. By default this will now set the upper limit of the bass to midrange crossover. You will probably want to keep it no higher then 500hz if you're using a steep slope and probably 400hz if you're going 2nd order.

So say we're now set at @ 400hz, 4th order just for the sake of things. This is a nice frequency to use. The midrange will have significantly less work to do as a result, and any beaming from the bass driver will be completely eliminated in this instance.

Measuring the midrange shows that it is comfortable with 400hz so that's not a problem. But you notice that arriving at a 4th order acoustic target is troublesome. This is going to require a 3rd order electrical network as the xover point is quite a distance away from the boxes natural roll off. Decreasing this to 250hz allows you to use a 2nd order electrical filter. As the midrange is comfortable with 250hz, you now decide this is a nice compromise.

Midrange - tweeter xover.

The midrange driver has a metal cone it has a bell like resonance at around 10khz. As a result we've now got peaks in the harmonic distortion at divisions of that peak - there is a significant increase in third harmonic at ~3khz and we want to avoid this.

This places a limit on the crossover to the tweeter. The tweeter is capable and has low distortion and enough Xmax to cope with a 4th order at 1500hz. This would entirely do away with that peak in HD at 3000hz, but would this work?

In simulation you find that matching the phase at 1500hz is a bit tricky and is very prone to slight component value changes. However if the xover point is increased up to 2khz things fall comfortably into place with asymmetrical slopes around 4th order. It is far less acceptable to component value deviation and does away with the peak at 3khz nicely.

This is just a small example. But it shows how things can go. It's always better to start out with ingredients that can give you as much potential flexibility as possible, because sometimes other factors come into play that place limitations on what you can do. And you don't want one of these to screw everything up.

[5th element]

far less acceptable? That should read susceptible. My spelling + spell checker for the loss!

[Moondog55]

5th that synthesis deserves to be posted in the WIKI

Very succinct and lucid explanation

[prickears]

Gotta second that. And spoken by someone who clearly does understand the basics, Then how to utilize (rather than rely on) simulators.