[Rudolf]

I have been asked to resurrect the Helsinki thread which had been lost in the latest forum crash. Here we go again:



The Gradient Helsinki 1.5 http://www.gradient.fi/models/helsinki is the successor to the Gradient 1.0-1.4 http://www.gradient.fi/vintage/10-series
The Gradient 1.0 - 1.5 are the results of a continuing effort of finnish designer Jorma Salmi to control loudspeaker directivity beyond the pattern of conventional dipole speakers. Targets and achievements have been discussed in

"Basis of Stereo" www.gradient.fi/file_download/15/Basis_of_stereo_Gradient_design.pdf
"The Gradient 1.3 Revisited" http://www.regonaudio.com/Gradients.html and
a review in "The Absolute Sound" http://www.avguide.com/review/gradient-helsinki-15-loudspeaker

My special interest was in the "laid back" midrange driver, which directs a bigger portion of the midrange output to the ceiling than to the floor. The speaker-to-ceiling-to-ear distance is typically larger than the speaker-to-floor-to-ear distance. This should lead to a minimized floor bounce, substituted by a ceiling bounce at a higher frequency with a narrower response dip. I also wanted to change the midrange distribution from a plane parallel to floor and ceiling to a tilted plane, which might lead to a more evenly distributed far field response:



This is what I came up with:



First achievement was a better alignment of the acoustic centers of woofer, midrange driver and tweeters (red line). For the record: woofer is an 15" A&D Audio 15308, midrange is Monacor SPH 176, tweeters are Monacor DT 25N.

The baffle is tilted 20°, much less than the Helsinki´s 45°. Measurements did not show significantly less floor bounce or a new ceiling bounce, but I like the overall response better than with the upright baffle.

[mtnbob]

Thanks for bringing this back up. I was wondering about the crossover points you used. I know the original Gradient mid was 100-1500 Hz although I don't know what type of crossover it used.

[jimdgoulding]

And how does what you've done overall matter to your end game compared to what you are accustomed to in your room?  Thanks.

[Rudolf]

I was wondering about the crossover points you used.

Crossover is at 300 Hz (18dB/oct) and 2 kHz (24 dB/oct) effectively.

And how does what you've done overall matter to your end game compared to what you are accustomed to in your room?

What I hear is a little bit more pleasant than with the upright baffle. What I measure is worse than before  
Getting the ACs in line did not achieve anything measurable, but the different surrounding for the tweeters at 0° and 180° makes a huge difference. You don't see it at 10 cm, but at 1 m distance it is obvious. It becomes even worse in combination with the midrange driver.

I very well understand why Gradient did not opt for a dipole midrange, but for a cardioid. It is revealing that you see almost nil back side measurements for any dipole loudspeaker with dynamic drivers - Orion included. H frames work perfectly as dipoles up to 300 Hz. Back-to-back tweeters are well behaved too. But midrange drivers are far from perfect when trying to integrate them in a true dipole configuration.

[jimdgoulding]

Rudolf-  Fascinating stuff.  I read the TAS review when it came out and know that Robert Greene is serious about how sound behaves once it departs a loudspeaker.  I'm wondering if using over and under bass drivers with the tweeter placed inbetween in a single enclosure loudspeaker might not make bass more even measuring and listening at the seated position because then the two drivers would be different distances from the floor (and ceiling).  Could that make for a better balance you think?  Or, even stacked dual bass drivers period of which you see a lot of these days.

Mtnbob-  I have to chuckle reading Mark Twain's quote.

[mtnbob]

Jim - yeah, MT is my favorite American theologian. Perfect quote for this forum - aren't we all going against the sealed/BR flow?

I still would like to try the idea of the slant back mid. I've got a bunch of odd material lying around, so I think I'll try a sort of u-frame but with triangular sides - a delta frame? 

But not right now, starting tomorrow we're keeping the grandsons (1yr and 2yr) for a while, so I'll be just a little bit busy!

[Rudolf]

I'm wondering if using over and under bass drivers with the tweeter placed inbetween in a single enclosure loudspeaker might not make bass more even measuring and listening at the seated position because then the two drivers would be different distances from the floor (and ceiling). Could that make for a better balance you think? Or, even stacked dual bass drivers period of which you see a lot of these days.

With a crossover at 300 Hz you can't place the woofers more than 50 cm from the midrange driver (which I'm already violating in this case). Maximal woofer spacing would be 1 m. That would certainly help to fill out the floor bounce dip and distribute the excitation of room modes.

But the bigger problem -as I see it - is with the midrange driver:



I can get equal response for the front and rear tweeters easily. But there is no easy way to EQ the front (0°) and rear (180°) radiation of the midrange cone.

[scorpion]

Rudolf, this is interesting. I think you are approaching the in Sweden famous Stig Carlsson who put up four tweeters in forward and backward directions to produce a diffuse stereo with the help of mainly the back wall. His approach was in fact quite convincing with his speakers and with a pleasent effect. They were all BR-cabinets. Now I think you are trying to achieve something similar with the help of roof reflections. There is a manufacturer left in Sweden producing speakers in his footsteps with angles for bass-midrange very similar to yours. For what I have heard at trade shows these speakers are very nice sounding and with good definition. But I am still to be conviced about the OB uses here. These also are BR-designs. I will not repeat my DCX2496 argument but respond that I don't think JK's argument is among his best and strongest in: http://www.musicanddesign.com/Dipole-offset.html , given the common backward weaknesses of practically all measured units and also that angeling will disrupt his conclusions. So it may very well be that delay insertion will improve stereo in all dimensions. 

/Erling

[Rudolf]

In my last post I got caught in my own prejudices. 
I assigned the difference between 0° and 180° response to the different front and rear geometry of the midrange driver. Today I know better (don't mind the dB scale):



I measured the midrange driver in the tilted baffle - in normal mounting position (cone facing outward / red) and 'inverted' (magnet facing outward / black). The left diagram shows, how both responses are almost identical up to 2 kHz.
Then I turned the midrange baffle 180°, looking at its back side. I measures again with the magnet outward (normal / blue) and the cone outward (inverted / green). See the right diagram.

The 'bad' response to the back side is not caused by the driver geometry, but probably by reflections from the top of the H frame below.
Obviously it wasn't very brilliant to tilt the baffle as shown above.

@ scorpion:
There had been a lengthy discussion about Carlssons direct/indirect radiation pattern on DIYaudio some time ago. From what I learned there, Jorma Salmi surely works in that tradition. But, as you already remarked, it does not seem to work for OB. I was not aiming for roof reflections, but for aligned acoustic centers. Next step will be back to an upright midrange baffle centered above the H frame. I do keep the DCX delay as a last resort.

A follow-up to this can be found at http://www.audiocircle.com/index.php?topic=68685.0

[catapult]

The left diagram shows, how both responses are almost identical up to 2 kHz.

Very interesting, Rudolf. I wonder if we can add another rule of thumb that the backside of the driver is (may be) pretty good up until the cone diameter is about a wavelength. Obviously some drivers are more open than others but it makes sense that obstructions smaller than a wavelength would have less effect.