Can't Reproduce a Square Wave.

[roddyama]

I didn't mean to hijack a thread, so I started a new one.

How do you make a speaker output a square wave?


Quote from Bill Fitzpatrick

Now, tell me why YOU think I need to go digital and time delays to get a square wave.

Square wave. The fundamental frequency plus all odd order harmonics. Now this you can reproduce. What I can't see is how to reproduce the harmonics with their original phase relationship to the fundamental frequency with a multi-way system.

If you go active xovers to eliminate the the passive xovers, you still have the frequency dependent phase shift.

If you run bandpass boxes with no xovers, you still have frequency dependent phase shifts associated with the inductive and capacitive components of a speaker in a box.

It doesn't make sense to make a multiway system with full range drivers. That might get you close in the speakers bandpass, but the system wouldn't sound very good.

There is no reasonable way to eliminate the frequency dependent phase shifts of analog circuits.
The only way that I can think of is digital FIR filters and digital time delays to contend with variations in the acoustic centers of the drivers. Even then, I fear that the best you'll get is a rough approximation of the original square wave, but if this could result in a phase coherent reproduction of the leading edge of the square wave, I would be very happy with that.

ok? Did I miss anything?
Rodd Yamas***a

[Bill Fitzpatrick]

This is a good thread. The holy grail, all other things being equal, is the launching of a square wave from a speaker system.

Let's have a lot of input here and please stick to the subject.

[f4ier]

John Kreskovsky has a writeup or two on the subject of reproducing good square waves out of loudspeakers

[roddyama]

Hi F4ier,

Interesting read. John K indeed has ample knowledge on loudspeaker and how they work. He frequents the Madisound speaker forum. He's a very solid participant there. Unfortunately, he has not reproduced a square wave with his system. He has done a number of simulations. He has also reproduced an impulse of 0.1ms (20kHz). This is only the tweeter responding to a high transient input.

What he has not shown is a phase coherent response from the real speaker that includes a low frequency (<500Hz) fundamental + all odd order harmonics up to 20kHz.

Thanks for the reply.
Rodd Yamas***a

[f4ier]

Hi Rodd,

Yup, that guy knows a lot about loudspeakers. Have you seen his prototype 2-way measurements? His square wave measurements can be found at http://www.geocities.com/kreskovs/CrossoverdocN2.html Though his prototypes may not be able to reproduce recognizable square waves accross a wide frequency spectrum, it is an achievement nonetheless; especially considering he uses analog components

He's bound to produce a digitally-filtered/EQd speaker set one of these days now that his favorite CAD program has digital design and simulation capability. Perhaps he'll produce a prototype out of curiosity We all know these things can easily be done in the digital domain, but where's the fun in that

Isaac

[phase_accurate]

Hi Rodd

I don't believe that a perfect sqarewave will ever be reproduced by a loudspeaker either (apart from the non-existence of perfect sqare-waves in the real world). But it is possible to get closer to it than >95% of the speakers can.

From the versatility point-of-view an all digital active solution would be the best one (time response, possibility to equalize driver's responses etc).
By the use of automated test procedures it would even be possible tweak batch production speakers so they are almost like clones, performance-wise. But this would require a HUGE development effort and expensive production facilities.
Me personnally, I prefer analog filters.

To come to the point: Also an all digital solution wouldn't allow the reproduction of a real square-wave. To achieve this, a reproduction system must have a flat frequency response from 0 Hz to infinity, which doesn't exist in real life.

The closest thing one can build is a system of very high bandwidth (< 20 Hz up to > 40 kHz, 12dB/octave slopes !!) that has a reasonably flat amplitude AND phase response in between. I.e. whose crossover doesn't introduce any additional phase shift.
It's square wave response would still not look perfect, i.e. the rising and falling edges will not be infinitely steep and the top and the bottom will be tilted. But it's response to falling or rising edges does at least not look like a series of responses, starting with the tweeter's response and ending with the woofer's one.

Regards

Charles

[ALW]

Electrostatics!

the only speaker that CAN reproduce a square wave.

[phase_accurate]

Hi ALW

I agree with you that electrostatics are amongst the best ones in this respect* but I would still say that ABSOLUTELY NO speaker is capable to produce a real square-wave. I.e. a signal with infinitely fast rising and falling edges and horizontal top and bottom (I think you can imagine what I mean by this, I don't know the correct English terms for it) because this would ask for an FR fom DC to infinity which not even the best ESL will ever achieve.
But some speakers are indeed getting closer to it (ESLs, FRs, some active concepts...) and some less (mainly the classic multiway with crossovers of order 2 or higher).


Regards

Charles


*I am listening to the MSW FR transducer whose manufacturer also claims that it should be able to reproduce a square-wave.

[navin]

yes - few if any speaker can reproduce a square wave. the problem lies with the physics of the issue. to reproduce a squeare wave one must have a very high acceleration factor (using dynamic drivers) that means a high BL and low Mms or Mmd. The speaker must also have very good braking. Now that is damn diffcult. lets look at Acceleration fctor only....

A high Bl/Mms ratio can be achieved within a narrow frequency band but over a band that for practical pruposes covers 30Hz to 24kHz it gets crazy esp. for lower frequenies.

make a driver with low Mms and the low frequencies suffer. Make a driver with high Mms and the highs suffer. XOs only make matters worse.

Braking factor means the magnet / motor are in absolute control all the time. It means the cone has negligile (if any) inertia. See the problem?

BTW why would one want to reproduce a square wave?

Regards
Navin

[phase_accurate]

Hi Navin

Nobody wants to listen to square-waves, me neither. But only a speaker that is capable of faithfully reproducing a square-wave is capable of reproducing ANY waveshape.

OTOH a speaker whose focus is ONLY the reproduction of perfect sqarewaves isn't perfect either.

Regards

Charles