The "Elsinore Project" Thread

[ABO]

Hi Guys,

sorry to destract from the waveguide, but I hope you could help me understand some things:

- recession for the tweeter
Joe points out that the top 3 drivers act as a point source, seen from 1 meter distance. It is implied that that is a good thing. However, Joe also points out that he or anyone is unlikely to ever listen from that distance. Wouldn't it then be better to decrease the depth of the recession? For instance by not flush mounting the tweeter.
It would make the build easier and improve time alignment in one go.

- bevelling the baffle
Joe is very clear about the importance of baffle width: it should not be changed. Bevelling the baffle would change the width and is therefore not recommended. However, earlier Planet10 pointed out that it is the ultimate width that counts and that bevelling the sides does not change that. In fact, he says that it would reduce ripple, which likely is a good thing. Joe did never respond to that.
IMO, bevelling is important from an aesthetic POV and the now seem two different options:
- follow planet10's advice and just go ahead
- use thicker MDF panels to compensate for the bevel

I hope that any of you experts could elaborate on both issues.

[Luke]

Originally Posted by Joel Wesseling

The Black Paint I picked for the waveguide has a little texture in it, not heavy texture. Now wondering if smoother is better. There was an ultra smooth finish choice as well.

It's still possible to change if neccessary?

maybe Joe can answer this?

thanks, Joel

I think smooth looks better, but as Joe says, it shouldn't make any difference to the sound.

[Joe Rasmussen]

Originally Posted by ABO

Joe points out that the top 3 drivers act as a point source, seen from 1 meter distance...

- bevelling the baffle...

[OK.. add the sound of Joe stepping up on his soap box, please excuse me if I go into teacher mode...]

1st part: No, in fact they were measured and modelled primarily at 2 Meters. But measurements were taken at 1M on axis, 2M on axis, 1M 15 degrees off axis, 2M 15 degrees off axis, 1M 25 degrees off axis, 2M 25 degrees off axis. Each of the six represents a group of measurements of three accoustics measurements, that a total of 18 "farfield" measurements plus "nearfield" measurement (right on the cone) plus "nearfield" measurement of the port. These are combined as a composite responses. The farfield measurements contain the phase information and maintains the phase relationship between the drivers. The Tweeter requires no nearfield measurement, but the two nearfield measurements are common to all the other farfield measurements.

What I am basically saying is that I can look at all those six different positions with the same crossover with the same values in common to all of them, and get a broad handle on the overall behaviour. May I also add that the 1M measurements are at 1 Watt nominal (2.83V), but that 2M measurements are 4 Watt (5.65V), but I attenuate the 2M measurements by 6dB so as to calibrate the graphs to the standard we are familiar with, namely their sensitivity at 1 Meter (Metre to us non-Americans).

But basically the "offset" which is generally the correct term for the recess of one driver in relation to another, is less critical here than in other straight 1st order Butterworth designs that sums -3dB down. Where the drivers line up perfect in time they will sum 6dB, but if slightly further away (than 2M) they will sum just slightly less than that. There will hardly be a difference between 2M and 3M (but much bigger going from 2M to 1M). Butterworth will on the other hand come apart very quickly. Butterworth has a narrow window of opporunity time wise, whereas the Elsinores do not.

BTW, the notion of lining up Acoustic Centres is almost amusing to me, I get visual thought of somebody chasing and trying to catch the wind. The AC offset is not a singular constant but changes with frequency and also likely with dynamic swings. It is where the impulses line up that is important, time measurements never lie.

2nd part: I've got nothing against bevelling perse' - but I recognised that not all DIY'ers have the required skill or even tools. I wanted to make the design as accessible as possible. So the above described acoustic measurements were based on a cabinet that most people could build. To be strictly correct I would have to do all the above with a box that has been bevelled and then model the design, specifically the crossover, to make sure it works.

What I am saying is that I cannot be as precise and can only predict - and I am not going to do that. Slightly widening the box would most likely suffice.

[... sound of Joe coming off his soap box and almost stumbles to the ground with a wry smile on his face.]

Hope that helps.

Cheers, Joe R.

[tinitus]

Originally Posted by Joe Rasmussen

Slightly widening the box would most likely suffice.

Cheers, Joe R.

like thicker side panels, for make edge bevelling ?

flat front area/width remains the same/constant
so, no that probably wouln't hurt

but true, anything you do to box surface on a well adjusted speaker will be audible, one way or the other

hey, try to mount felt on outside back of speaker box
I'm quite sure you will hear that too
place something on top of speaker box
and you will hear that too
anything is audible
even the dammed mounting screws on tweeters
well, they are gone here
which is cool

[planet10]

Originally Posted by ABO

- bevelling the baffle

Where the bevel (or round-over) affects things is related to its size. I doubt you can get enuff bevel on elsinore to even reach down to the XO point.

Here we get to invoke page 23 of Olson again.



The basic dimension of these is 24" with a 6" wide bevel. With Elsinore and its 3/4" walls you can reasonably achieve about a 3/4" bevel which will have an affect down to about 4500 Hz and then below that will look square. It will help in the top 2 octaves. Beneficial, but should have no effect of BS ripple or the XO.

Given the waveguide loaded tweeter effective benefits will be low (ie mostly cosmetic).

dave

PS: Joe the link taking you to the next page at the bottom of this page is broke.
The Elsinore Name

[ABO]

Joe and Dave,

Thanks for replying. I conclude that bevelling, or rounding, the side edges of the baffle is not objectionable perse.
I would only round off the edges with a diameter <1cm which should be no problem, it seems.

Joe, I still don't understand something. You say that the mere idea of lining up the acoustic centre of the drivers is useless, funny even.
But I thought that was the very reason why you gave the tweeter its 'offset'; to make a point source?

Is there another reason the tweeter is recessed?

[Joe Rasmussen]

Originally Posted by planet10

Where the bevel (or round-over) affects things is related to its size. I doubt you can get enuff bevel on elsinore to even reach down to the XO point.

Here we get to invoke page 23 of Olson again

Hi Dave

So in your opinion, would a 1/2" bevel not made that much of a difference? I can see your reasoning and it is appreciated. I read into your comments that it would be negligible.

Those graphs you've shown are very similar to what's in John L. Murphy's book I have here. Yes, the size of the bevel comes in to the equation and if it is large enough it has undoubtably very significant benefits. But are small bevels are over-rated?

We do have another mechanism at work with the Elsinores, the multi-driver array with large over-laps in frequencies, this will do some degree of fill-in. The top driver suffers the most IMO, but the driver below the tweeter undoes to some degree the problem of the top driver. Not perfectly mind you, but helpful.

Below 1KHz the overlap of the four identical drivers is far greater in the Mk5 than earlier versions, so much so that the overall sensitivity has gone up at least 1dB, in fact it looks like 1.5dB, the modelling has also shown that is has reduced the diffraction effects further (below 1KHz) and also that the often severe and ignored problem of floor bounce is pretty much nonexistent because you have drivers sharing the same frequencies all emanating at differents heights relative to the floor. In many 2-way (and even 3-way) designs the problem of floor bounce is often messing up diffraction effects - because they can usually show up at the similar frequencies - and this is completely missing in the Elsinores. Bevelling would do pretty much nothing to solve this.

What I am saying, and I am speaking to the other guys out there, I stress I am NOT against bevelling, in fact I LIKE the idea, but the Elsinores required a whole bunch of systemic decisions that has to work together. Also I wanted as many as possible to be encouraged to be able to build them. So I chose a square box.

Originally Posted by planet10

PS: Joe the link... is broke.

Fixed, thanks.

Cheers, Joe R.

[Joe Rasmussen]

Originally Posted by ABO

Joe, I still don't understand something. You say that the mere idea of lining up the acoustic centre of the drivers is useless, funny even.
But I thought that was the very reason why you gave the tweeter its 'offset'; to make a point source?

Is there another reason the tweeter is recessed?

No, the reason is to time align the drivers so that they do behave as a point source.

What I was saying is simply this, to look at an individual driver in your hand and point to a physical position inside that driver as the driver's AC. That simply is a flawed assumption that the driver has such an AC. So what do you align if not the driver's AC? Their time response.

That's all I am saying.

Cheers, Joe R.

PS: If a driver had such an AC, then the manufacturer could easily publish it and put all of us out of our misery. But all they can point to is the physical location and dimensions of the Voice Coil. And when we talk about AC's, what about the effects of the crossover and how the AC would be delayed in time (crossover can only ever delay), that is simply not predictable. It's all enough to make your head spin - no, build it and measure it, look at the time error and then make the correction. The only sure way to do it.

[planet10]

Originally Posted by Joe Rasmussen

So in your opinion, would a 1/2" bevel not made that much of a difference? I can see your reasoning and it is appreciated. I read into your comments that it would be negligible.

Pretty much acoustically. At least a little bevel or roundover, is nice aesthetically,

Those graphs you've shown are very similar to what's in John L. Murphy's book I have here.

Of course they do... he stole them from Olson too. He did thou omit some of the most important ones.

Yes, the size of the bevel comes in to the equation and if it is large enough it has undoubtably very significant benefits. But are small bevels are over-rated?

Indeed, on the Fonken, where the bevel is on the order of 3" it makes a really significant contribution,

Small bevels have some small benefit on flush mounted tweeters -- the same ones that have issues with raised screws.

dave

[Joel Wesseling]

Picking up Painted WAVEGUIDES tomorrow

I think smooth looks better, but as Joe says, it shouldn't make any difference to the sound.

I did ask, prior to painting, about the finish I selected and was told it was a very fine texture so should be nice.


FILLER BOARD

I checked my MDF and it is 3/4" or 19mm so A bit of sanding of face or back is required.
I thought about making the filler 162mm by 232mm and then you could easily sand the approriate sides to allign off center tweeter while completly filling the opening.. Any suggestions? I can also go with 158x228
when I make the circle template it will have to be for one size board .

Joel