diyAudio (
-   Multi-Way (
-   -   Adventures in cardioid (

MBK 21st April 2009 06:23 PM

Adventures in cardioid
2 Attachment(s)
Well, after a long hiatus I am taking to experimenting again. This time I intend to modify my existing OB setup to something more truly constant directivity and power response.

The existing system has 15" 18Sound 15ND930 woofer up to 320 Hz, then a ScanSpeak 8543 up to 1600 Hz, and finally a Seas 27 series dome tweeter. All this on a 48"h, 24"w, and 7" shallow U-baffle without damping, a large panel really.

The plan is to get a CD horn and to convert the midrange to cardioid, probably leaving the woofer in OB. I should point out that after several years of listening to my current setup, I find really nothing much wrong with it, on and off axis balance, imaging, intelligibility are subjectively great, except for one thing: even with toe-in the imaging changes substantially if you move around in the listening position, and sitting off center happens often when watching movies. Plus we moved to a house with a smaller living room so the speakers moved too close to the wall and just look too big there.

Now Earl Geddes in his Summa whitepaper, describes his approach with a fairly narrow CD waveguide with substantial toe-in, to get wide field imaging and optimal small room reverberation, and it all sounded very convincing to me. So I wanted to try this concept with the aim of getting better imaging when off center, and even better power response, and more tolerance for closeness to back wall, as by-products.

I could have bought a Geddes product but I like OB bass, plus, we are DIY here right... so initially I wanted to have an OS CD wave guide a la Geddes CNC machined in wood. But then I thought, let's not vary all parameters at the same time and proceed piece meal. So for the horn I will go for now with the 18Sound XT1086 with an ND1020, both cheap off the shelf items.

That leaves the midrange to fix. It will have to behave like a cardioid on top to mate with the horn, and go towards dipole at the bottom.

Her is a pic of the existing system, to be modified.

MBK 21st April 2009 06:27 PM

Existing mid polar response
2 Attachment(s)
Measurement in the yard, some reflections from plants unavoidable but at least I get the driver 2 m or more off the ground and can use a 2 m measurement distance. Windowed this goes down to 150-200 Hz but unwindowed actually looks better. Arta with Linkwitz mod Panasonic capsule, Rod Elliot's mic preamp modified for the Linkwitz mod mic (...), M-audio Mobile Pre, rather old, and run off the mill laptop.

Attached is the ScanSpeak 6.5" 8543 midrange, raw polar response on a mockup 48hx24wx7deep U-frame - basically my existing system. Surprisingly ugly for sounding so good (of course in the actual system I am listening to it is EQ'd reasonably flat, but the polars are no good).

MBK 21st April 2009 06:29 PM

The 18Sound XT1086 with ND1020
2 Attachment(s)
And next, the raw polar response of the ND1020 compression driver on the XT1086, mounted on the same mockup 48hx24wx7deep U-frame.

The XT1086 response looks quite respectable to me (must be EQ-d of course). So this is what I have to match with the SS8543 6.5" midrange at say 1200-1800 Hz somewhere.

Edit: the spacing here is 7.5 degrees, all in all 0-90 degrees fontal horizontal polars. On this particular measurement the aberrant 7.5 degree curve is an artefact due to a soundcard timing glitch.

MBK 21st April 2009 06:44 PM

How to do a cardioid?
2 Attachment(s)
Well, one way to make a driver's response more cardioid like is described here by Siegfried Linkwitz. Basically it consists in putting a flow resistor on the rear of the driver, that acts as a low pass but more to the point, slows down air flow. Linkwitz cites materials that do just that, sintered metal meshes used for exhaust mufflers and filters.

I wanted to try this, but without exotic materials. So I took raw cotton material sold for pillow stuffing (extremely light and fluffy, totally unprocessed) and built something that literally looks like an air filter or a large inside part of a muffler. Front and back to hold the driver, are just routered out of MDF, the cotton is held in place with aluminium mesh. Incidentally I saw a picture of one late model Gradient speaker that seemed to have a construction vaguely like mine here. Looked prettier actually.

MBK 21st April 2009 06:45 PM

2 Attachment(s)
with cap

MBK 21st April 2009 06:49 PM

2 Attachment(s)
This muffler was the first mounted inside the familiar baffle, basically like a 'damped backwave' baffle. Elegant masking tape holding it all together.

MBK 21st April 2009 06:52 PM

And the polar response is...
2 Attachment(s)
... not so bad. But still, the response is first widening towards 800 Hz, then narrowing again, plus crossing lines etc.

MBK 21st April 2009 06:57 PM

So let's eliminate the baffle
2 Attachment(s)
... so I put the muffler assembly on top of the baffle, instead of in it, and measured again. And lo and behold - stunningly regular polars, without crossing lines, without lumpy resonant peaks, none of this stuff. Only drawback is rapid falloff in the bass, due to insufficient separation distance with the rear wave (and low absorption of low frequencies of my damping).

I was quite impressed when I stumbled on this: I really did not do anything very sophisticated here, just added light resistive damping and counterintuitively, a circular, regular, small baffle did the least damage.

Edit: now is this cardioid? Actually polar graphs show, reasonably well above ca. 800 Hz, below that the rear response is less and less damped. I'll post more results tomorrow on the full polars including rear, and FR changes with baffle shapes and sizes.

Fast1one 21st April 2009 08:17 PM

Interesting experiment. I am curious to hear your final impressions. So essentially the cylindrical muffler is mounted in front of the original mounting hole?

MBK 22nd April 2009 02:35 AM

2 Attachment(s)
No, it's just an assembly all by itself now, later to be mounted on top of any supporting structure. Below, on top of the old baffle for measurement purposes.

All times are GMT. The time now is 03:32 AM.

vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2016 DragonByte Technologies Ltd.
Copyright 1999-2016 diyAudio

Content Relevant URLs by vBSEO 3.3.2