[Rudolf]

There have been considerations about the best position of a dipole full- or midrange driver on top of a dipole woofer:

http://www.musicanddesign.com/Dipole-offset.html
http://www.audiocircle.com/index.php?topic=59619.msg530336#msg530336

Today I was able to delve deeper into the situation. Let's first have a look at the test baffle:



Xover frequencies are 300 Hz and 2 kHz. I have shown some time ago that the response of the woofer in my H frame is almost identical at 0° and 180°. Same for the midrange driver Monacor SPH 176 in it's project passband:



The OB was placed centered on the H frame as shown above.
What happens if I move the midrange OB to the front or the back of the H frame?



First lesson: It does matter. In the right diagram I took the measurements apart to better show the differences. Positioning the midrange baffle at the front edge of the H frame appears to be best, positioning in the center looks to be worst.
I need to exclude the possibility of changing room influence (the baffle was moved 40 cm in the process) to be responsible for the shown difference. Therefore I took the same measurements at another position in the room and with 1 m distance instead of 2.5 m:



Again we see IMO: Front looks best, centered looks worst.
As I see it, most reasonably the changes in response are due to reflections from the top of the H frame. My next steps will be to 'hide' the top of the H frame from the midrange driver.
FYI: Of course I did the same measurements for the tweeters too.  They also show changes, when moving the baffle, but much less than for the midrange.

[Telstar]

You tried already to incline the midrange baffle, right (like in the helsinki)? Is that better to the front straight positioning or not?

[Rudolf]

I'm going for a vacation now and I wont be thinking about my speakers everyday.

Now that you don't allow yourself to think about YOUR speakers, you think about MY speakers instead Relax, man, and enjoy your holidays 

You tried already to incline the midrange baffle, right (like in the helsinki)? Is that better to the front straight positioning or not?

Yeah, that set the ball rolling. The response (measured nearfield) was better to the front than with the upright baffle, but MUCH worse to the back. Next experiment will be a sort of gable roof below the midrange driver, covering the H frame top. Just padding the H frame with layers of damping material did not help much. May be I try a sort of felt wedge instead.

[ttan98]

Rudolf,

Looks interesting, if the baffle is best mounted in the front, then how difficult is it to align your mid-range with the woofer when you design the x-over?

[Rudolf]

Rudolf,
Looks interesting, if the baffle is best mounted in the front, then how difficult is it to align your mid-range with the woofer when you design the x-over?

Actually it is not best to mount the baffle in the front, but to leave the baffle in the middle and take care of the reflections from the H frame top. Only if you can't do the latter - for whatever reasons - you should move the baffle to the front.
If you sit at least 2 m away from the baffle and cross over in the 200-300 Hz range, the misalignment with the woofer should be less pronounced in measurements than the reflections. That is my experience. As JohnK has shown, you can not align the drivers afterwards by electronic means.

[Telstar]

Now that you don't allow yourself to think about YOUR speakers, you think about MY speakers instead Relax, man, and enjoy your holidays 

I have to write to the carpenter...
I have to order the woofers and the dcx...
No, i have to go out and have a walk in the park instead 

[ttan98]

If the mid is mounted in the middle where does the FR dip comes from? you mean caused by woofer directly below?

[Rudolf]

If the mid is mounted in the middle where does the FR dip comes from? you mean caused by woofer directly below?

All measurements were done with the woofer and tweeters switched off. The dips are due to path length differences between the direct response (red) and the response reflected by the H frame top (pink):

[scorpion]

For sure, Rudolf, we look forward to see your next measurement when you have succeded to get rid of H-top reflections.
And at that point your sonic impressions comparing the mid (or perhaps the in plane placement of bass and mid units) with the angled baffle from your 'Ode to Helsinki' thread. 

/Erling

[Russell Dawkins]

So I wonder what the effect would be of attaching the bottom of the baffle near the front and angling it back so the mid is almost vertically over the woof. Some degree of vertical alignment of acoustic centers would result.

[D OB G]

Are we getting closer to a U frame?

[JoshK]

You tried already to incline the midrange baffle, right (like in the helsinki)? Is that better to the front straight positioning or not?

Yeah, that set the ball rolling. The response (measured nearfield) was better to the front than with the upright baffle, but MUCH worse to the back. Next experiment will be a sort of gable roof below the midrange driver, covering the H frame top. Just padding the H frame with layers of damping material did not help much. May be I try a sort of felt wedge instead.

You can try placing the baffle upright in the center of the H and then make a triangle (when view from the side) by placing a flat board from the bottom of the mid angled down to the front of the H on both the front and the back.  Sort of a "waveguide" to gaurd against reflection.

[Rudolf]

You can try placing the baffle upright in the center of the H and then make a triangle (when view from the side) by placing a flat board from the bottom of the mid angled down to the front of the H on both the front and the back.  Sort of a "waveguide" to gaurd against reflection.

This has already been done. I took some flimsy masonite for a test:


There are differences in measurements, but ambiguous and no real improvement. I need to get a better understanding by further tests.

[Rudolf]

So I wonder what the effect would be of attaching the bottom of the baffle near the front and angling it back so the mid is almost vertically over the woof. Some degree of vertical alignment of acoustic centers would result.

That was step 1:


http://www.audiocircle.com/index.php?topic=68553.0

Now we are already at step 2 

[catapult]

Seems to me the best solution is to eliminate the top surface of the H frame. And the bottom surface isn't really needed either as it sits on the floor. So that just leaves a flat baffle holding the woofer and the midrange with wings on the sides, shaped however works best acoustically and aesthetically. Wings narrower at the top than the bottom maybe, both for looks and to spread out the cavity resonance?

[ga foo 88]

What does removing the top of the h-frame do? I like the idea of having one flat baffle for the bass woofer and the mid range

[Rudolf]

Seems to me the best solution is to eliminate the top surface of the H frame.

Have a look at the first diagram in this post: http://www.audiocircle.com/index.php?topic=49940.msg447911#msg447911

May be you don't see much difference, if you measure very near to the H frame  http://www.audiocircle.com/index.php?topic=34399.msg311489#msg311489, but eliminating the top surface does cost some dB!

[PB]

You can try placing the baffle upright in the center of the H and then make a triangle (when view from the side) by placing a flat board from the bottom of the mid angled down to the front of the H on both the front and the back.  Sort of a "waveguide" to gaurd against reflection.

This has already been done. I took some flimsy masonite for a test:


There are differences in measurements, but ambiguous and no real improvement. I need to get a better understanding by further tests.

Hi Rudolf,
What about to use "Weichfaser" boards on this sloping panel, or foam?

Peter

[Rudolf]

Where do those ripples in the midrange response come from in the first place? It is the floor bounce for a large part:



The green line is a near field measurement at 10 cm. The positions (but not the values!) of the measured peaks and dips on the frequency axis for 0.8 m and 1.6 m correlate quite well with the respective floor bounce simulation in EDGE.

(All following measurements in this post show the complete response of the full dipole - including woofer and tweeters. Don?t care about anything shown above 8 kHz.)

Positioning the baffle at different spots in the room does add another factor:



For both measurements the microphone was placed 1.5 m from the midrange baffle, which was at the front of the H frame.  It looks like the in-room position makes a difference up to 800 Hz in this case.

In a next step I changed the mic distance from 1.5 m (red) to 1.35 m and 1.2 m (white). In the following graphs I spread the measurements on the SPL axis to let the differences show better:



Up to 250Hz there is not that much difference, same above 1 kHz. But in between ?

After that I left the mic at 1.5 m distance and changed the position of the midrange baffle on top of the H frame from front (red) to center to rear (white):



Up to 250 Hz the curves are identical. That?s where the H frame is crossed to the midrange baffle. Above 250 Hz we do see quite some difference, but the changes are not really better or worse than the changes due to the floor bounce.

For a last test I kept the mic at a fixed 1.5 m distance from the midrange baffle - moving the microphone while moving the baffle on top of the H frame. In theory that should eliminate the influence of different floor bounce geometries:



For me it is still like reading tea leaves. Obviously there are differencies and similarities - but what does matter? Probably I would need an anechoic chamber (without floor bounce) to get a clearer picture.

My attempt to summarize: If you don't feel the need to EQ for floor bounce or for the individual speaker position in room, you need not worry about the position of the midrange driver on top of a H frame either. It is all in the same league/magnitude.

[scorpion]

OK Rudolf, let's leave these measurements to our english speaking friends, they usually have some anechoic floors.
But what about the last resort, the DCX2496, why not explore the front position with delay inserted to compare ? 

/Erling