Are you (open) baffled yet?

[Variac]

The Fostex FT17 tweeters are good sounding, very flat, very efficient, and go up to 30khz. KYW has used them for projects
I believe.

 

[zygibajt]

Visaton TL 16 would be great,also some Fostex.

 

[Nuuk]

Thanks guys, I'll do some investigating!

 

[PaulHilgeman]

I'll just point you over to a thread where I have pics of my open baffle dipoles.

http://www.diyaudio.com/forums/showthread.php?s=&threadid=21829

-Paul

 

[qwad]

be better if it was australian or hungarian IMHO cheers tomcat

 

[Nuuk]

Please don't tell me to 'get a life' but I have just re-read the whole of this thread!

One aspect that was discussed was how much the baffle adds to the sound. Well, from the initial OB experiment I did with a driver mounted on a frame, and only polystyrene sheet for the baffles, I would say that it does add something.

Using only the polystyrene sheet, there was nothing added and my only 'complaint' was a perceived lack of bass. On the positive side, I have honestly never heard such a clear sound from any loudspeaker (including the 2300UKP ones that I heard the other night) despite using cheapo (10UKP) drive units!

So I would say as usual there is a compromise, this time between amount of bass (or perceived bass) and clarity.

More details

 

[Nuuk]

After fuurther experimenting with my OB's, I have found that it is best NOT to have any gap between the bottom of the baffles and the floor.

That may be obvious to some, but the baffles shown on the PHP site do call for a gap (although they are much bigger baffles).

Has anyone else tried with a gap and without? Also, what are your finding as regards placement of the baffles in relation to the rear wall?

 

[Chris8sirhC]

I think ive read through this thread three times now, and I have a couple of ideas which mighnt prove usefull (or maby not...?) concerning the snail shaped baffles. In all the pictures shown the strait line where the large radius ends and the small radius begins is always facing directly down. Would it be possible to tilt this 30 or so degrees to the side(then "slice" off the bottom to create a flat surface) , then you would have the snail shaped baffle along with a suitable base to keep it from tipping over. I'm thinking that at this point the baffle would be far enough away from the driver that you would only be dealing with lower frequencies and thus it would only help. Also, would it be possible to vary the slope of the conical shape in order to cancel out any peaks in the drivers frequency response, and to lessin the troughs in its response. I'm thinking you would make the shape expand faster over a given frequency distance to strengthen a trough, and make the shape expand much slower over a frequency hill in order to cancel it out so to speak. Just my two cents, and im curious if im completely off base here or there may be something to this.
Also not considering the varying slope idea, would it be necessary to slowly increase/decrease the slope across the frequency distance since at lower freuqnecies (longer distances from the driver) the sound will not be as loud as it is closer to the driver? (sound dissipates over distance)

 

[Svante]

I have also had some thoughts regarding baffle shapes recently. This is going to be a rather long and theoretical post, sorry about that.
Some of you have already tried my hack
http://www.tolvan.com/diffract.exe
that I wrote during my learning about this thing with baffle diffraction.

I think the snail shape has much in common with my heart shape, in that it smears the reflected pulse in time. I call it a Hanning window in the program (which I realised yesterday that it really isn't, but it doesn't matter for this post).

I reason like this: There are two sources of sound on the front side of an open baffle, one is the driver and its mirror, the second is a line source situated around the edge of the baffle. The edge source has the same power as the driver sources in case of the open baffle, but is delayed and smeared in time. The edge source is the sum of edge reflections from the back and front side of the speaker.

At really low frequencies the driver and edge sources will cancel, and this is the big drawback with open baffles. That is why open baffles have to be so large. The response at higher frequencies is better understood by first looking in the time domain.
First consider the circular baffle with the driver at the centre. In this case the delay between the edge and the driver sources will be equal for all directions and the egde reflection will be seen as a secondary negative impulse in the impulse response of the circular baffle.

Those of you familiar with fourier transform might know that an impulse contains all frequencies. Here we have two impulses with a delay. This means that some of the frequencies in the impulses will cancel, some will add up. The end result is that the response from this baffle alters between destructive and constructive interference as the frequency increases.

Second, consider the non-circular baffle shape. In this case the edge pulse will be smeared in time with a peak corresponding to the most commonly occurring distance between driver and edge. The shape of the pulse will be completely determined by the distribution of the driver-to-edge distance.

As with the circular baffle, the frequency contents of the driver impulse and the edge reflection will interfere, but now the frequency content of the edge reflection is different. Here we can see a great opportunity to control the interference by modifying the spectrum of the edge reflection. This we do by altering the baffle shape.

My heart shape stems from an attempt to shape the edge reflection to a hanning window, a function which is known to in some sense minimise the high frequencies. In this way, the interference towards higher frequencies should be minimised and we would be left with little but the 6dB/oct rolloff at low frequencies to compensate for.

The bottom line here is that it is not the shape of the baffle itself that is important, but the distribution of edge-driver distances that is. A snail, a heart, and even a curved star shape could yield the very same impulse response.

Maybe you all knew this, or maybe you all disagree, but I just had to write it down. I am pretty excited about the idea to optimize the edge reflection to a bandwidth limited version of the driver impulse, and I would love to have input from you if you have any. I may be completely wrong, but at the moment I am happy about it. We'll see when I come down to earth...

Sorry about the lengthy post. Comments?

 

[ThorstenL]

Konnichiwa,

The bottom line here is that it is not the shape of the baffle itself that is important, but the distribution of edge-driver distances that is. A snail, a heart, and even a curved star shape could yield the very same impulse response.

Or indeed a rectangular baffle with the correct spacing of the driver from the edge.... ;-) And of all given shapes a rectangular baffle will have the maximum surface are thus allowing the lowest reaching LF extension of any shape....

Good to know that sometimes theory reinforces the practice....

Sayonara

 

[Nuuk]

I posed the question of what effect a 'soft' baffle material will have on sound quality in another thread but got no answers.

So I'll raise the subject again here in light of the previous few posts. Surely, a less reflective baffle material will affect the sound for the same reasons put forward by Svante.

When I went from my test baffles which had a 25mm layer of polystyrene foam in front of the 15mm chipboard to my 'new' baffles which have a hard front surface, I noticed quite a difference in the sound. This could have been caused by the slight change in shape and size but I think it also had something to do with removing the softer baffle surface.

 

[sfdoddsy]

I would think any sound diffrerence, assuming the profile is the same, would be in the area of resonances. My dipoles vibrate a lot due to the bass excursions required. I have them decoupled and have rubberized the baffles. I don't think it makes any difference to the sound, but it makes me happier.

Steve

 

[Nuuk]

Surely reflections off the baffle front and edges must make some difference although I agree removing as much of the resonance is important.

 

[ThorstenL]

Konnichiwa,

Some of you have already tried my hack
http://www.tolvan.com/diffract.exe
that I wrote during my learning about this thing with baffle diffraction.

Great tool.

Could you "hack in" a few options, if possible?

1) Print and Export Data (then 2/3 can be done externally in Excel or such)

2) Include standard T/S derived (true sensitivity) Frequency response to be superimposed on "baffle" response

3) Include the influence of near room surfaces, especially the rear wall "bounce" and the "floor image" of the Baffle (other surfaces can be mostly ignored for the dipole option, not for the Box)

Alternatively, could you give the algorythm you use to calaculate and I update my excel model....

Sayonara

 

[Svante]

Konnichiwa,



Great tool.

Thx.

Could you "hack in" a few options, if possible?

1) Print and Export Data (then 2/3 can be done externally in Excel or such)

2) Include standard T/S derived (true sensitivity) Frequency response to be superimposed on "baffle" response

[/B]

I'll see about this. Some of it might happen.

3) Include the influence of near room surfaces, especially the rear wall "bounce" and the "floor image" of the Baffle (other surfaces can be mostly ignored for the dipole option, not for the Box)
[/B]

Would you know how this is usually simulated? I would take the first reflection from each wall and floor/ceiling except for the back wall. Is that it?

Alternatively, could you give the algorythm you use to calaculate and I update my excel model....

Sayonara [/B]

Sure, here is the code that does it (Delphi 7.0):


for x:=0 to xaxis.width do begin;
f:=xaxis.pixel_to_value(x);//get the frequency from my axis component

c1:=complex(0,0);
for i1:=0 to high(distance) do begin;
fi:=distance[i1]*2*pi*f/c; //calculate phase shift from delay c=345m/s, distance holds driver-to-edge distances
c1:=cadd(c1,cmul(complex(cos(fi),sin(fi)),complex(1/(high(distance)+1),0))); //add contributions, including phase shifts
end;
case boxtyperadio.ItemIndex of
0:begin;
c1:=csub(complex(2,0),c1);
end;
1:begin;
c1:=csub(complex(2,0),cfac(2,c1));//open baffle has diffraction from the back too.
end;
end;
d1:=cabs(c1);
y:=yaxis.value_to_pixel(20*log10(d1));
if x=0 then moveto(x,y) else lineto(x,y); //plot it!
end;


It involves a few functions of my own for complex math.
function complex(re,im:double):tcomplex;
begin;
result.re:=re;
result.im:=im;
end;
function cadd(c1,c2,c3:tcomplex):tcomplex;overload;
begin;
result.re:=c1.re+c2.re+c3.re;
result.im:=c1.im+c2.im+c3.im;
end;
function csub(c1,c2:tcomplex):tcomplex;
begin;
result.re:=c1.re-c2.re;
result.im:=c1.im-c2.im;
end;
function cmul(c1,c2:tcomplex):tcomplex;overload;
begin;
result.re:=c1.re*c2.re-c1.im*c2.im;
result.im:=c1.re*c2.im+c1.im*c2.re;
end;
function cabs(c1:tcomplex):double;
begin;
result:=sqrt(c1.re*c1.re+c1.im*c1.im);
end;

HTH
/Svante

 

[Svante]

Surely reflections off the baffle front and edges must make some difference although I agree removing as much of the resonance is important.

Difrraction reflection is hardly possible to avoid, unless the drivers are mounted in the wall. Otherwise this "baffle step" is unavoidable, and has to be compensated for.
BTW, this is interesting to realise when comparing response curves for drivers, some manufacturers mount them in a wall, some mount them in a box. The wall mounted ones usually *looks* better.

 

[Svante]

I posed the question of what effect a 'soft' baffle material will have on sound quality in another thread but got no answers.

So I'll raise the subject again here in light of the previous few posts. Surely, a less reflective baffle material will affect the sound for the same reasons put forward by Svante.

When I went from my test baffles which had a 25mm layer of polystyrene foam in front of the 15mm chipboard to my 'new' baffles which have a hard front surface, I noticed quite a difference in the sound. This could have been caused by the slight change in shape and size but I think it also had something to do with removing the softer baffle surface.

Surely reflections off the baffle front and edges must make some difference although I agree removing as much of the resonance is important.

I don't know the answer to this. I could imagine that an absorbing surface on the baffle could "eat" the wave that propagates toward the edge, but I dont know to what extent. It could be completely negible for any available material, or it could be quite noticeable. I don't know. Buyt clearly, the wave propagating on the back side has equally large magnitude as the one on front side, so both sides would have to be treated.
Also, the effect is likely to be greater for high frequencies.

I will probably set up some tests in the near future to measure these edge effects, and treating the surface one option that i have in mind to test.

If the baffle flexes, that is a completely different matter, and given the large radiating surface, this could very well make an audible difference, which if not properly controlled, IMO mostly would be bad.

 

[Svante]

Konnichiwa,



Or indeed a rectangular baffle with the correct spacing of the driver from the edge.... ;-) And of all given shapes a rectangular baffle will have the maximum surface are thus allowing the lowest reaching LF extension of any shape....

Good to know that sometimes theory reinforces the practice....

Sayonara

Erhh... Maximum surface? Given what conditions? But you are right, rectangular appears to be better that most people think, given an asymmetrical placement. Circular, or partially circular baffles with the driver in the centre of the (partial) circle appears to produce more FR ripple. That is one of the experiences my little hack has given me.

 

[ThorstenL]

Konnichiwa,

Would you know how this is usually simulated? I would take the first reflection from each wall and floor/ceiling except for the back wall. Is that it?

Well, I have been thinking about this (Norbert too and I cribbed some ideas from him).

First, if the baffle is in "free air", meaning with a substantial gap between baffle and floor, we will get fairly little effects from either sidewalls, ceiling or floor, simply because all these surfaces are in a "null" of the response. So only the sound radiated from the rear of the dipole becomes relevant to the room interaction.

What will happen is that the sound will be reflected back into the room with virtually no attenuation at low frequencies.

The rear output from most cone drivers plumets above around 1KHz with a 2nd order function down to a shelf quite notably below the nominal output, perhjaps 12 - 20db lower or continues to drop (depends upon the construction of the driver). Further, the higher components of the rear output become progressively more directional, which together with a slight slant from the baffle and possible toe in ensure that reflected rear radition is quite a bit down at higher frequencies. So the first element for modeling would be a lowpass for the rear radiation at 1KHz.

Secondly, the distance to the rear wall will delay the rear soundoutput of the dipole and the added distance will further attenuate the outpuit. So if the rear wall is 50cm behind the Speaker and we measure at 1m distance the SPL ofg the rear radiation arriving at the virtual mike will be 6db attenuated compared to the direct radiation. As long as the distance in pathways is smaller than around 45 degrees or 7/8 wavelength (in our case 0.875cm or 400Hz) the rear radiation will re-inforce the front output with an almost full addition achieved. With 180 degrees or 1/2 of the wavelength (50cm or 690Hz) we actually subtract the sound output of the rear, but as this already 6db down it's not much of a suckout.

So all in all once we are far enough from the rear wall the rear output will merely cause some mild lift in the < 500Hz and put a few more wriggles above that. The LF effect should be easily predictable.

Now to the floor. With the baffle extending to the floor and being nearly airtight connected to it we get out "floor image". This appears a second sound source, equal in level and polarity to the original. This actually effects both rear and front radiation, so if we superimpose this function on the front radiation and then derive the rear radiation with suitable delay and attenuation we acoount for it correctly.

Now the floor image also has some delay. So again we will find that as long as the distance to the floor image (about 2 X hight above floor) is small compared to the wavelength (say 0.875 wavelength) the floor image output will addup, probably closer to 4db than the theoretical 6db, plus we will see the usual dip at 0.5 Wavelength = distance, again, this should theoretically go 0db but will likely be much less deep.

So, all in all there is a material LF boost from the floor and less so from the rear wall.

Sayonara

PS, I had a long day at work, so while I have the logic clear in my brain I may have typed garbage by inverting a sign or making division a multiplication or such, so treat with care and check logic and math...

 

[Nuuk]

But clearly, the wave propagating on the back side has equally large magnitude as the one on front side, so both sides would have to be treated.

This is effectively what I tried when I mounted drivers on a 50mm thick sheet of polystyrene. See here

Even using the cheap drivers, the clarity of this speaker was stunning but I don't have the means of taking meaningful measurements.

And yes, this type of baffle can flex, but with the very low mass involved, it doesn't seem to be detrimental in practice.