Go Back   Home > Forums > >
Home Forums Rules Articles diyAudio Store Blogs Gallery Wiki Register Donations FAQ Calendar Search Today's Posts Mark Forums Read

Multi-Way Conventional loudspeakers with crossovers

Cardioid as sum of monopole and dipole speakers:
Cardioid as sum of monopole and dipole speakers:
Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 17th February 2015, 01:34 PM   #1
Barleywater is offline Barleywater  United States
diyAudio Member
 
Join Date: Jan 2008
Default Cardioid as sum of monopole and dipole speakers:

Here a SEAS L16RN-SL using a Linkwitz Pluto type mounting and a Vifa NE180W-04 suspended vertically above the woofer is used for exploring cardioid sound field. The Vifa (Tymphany) is sold as “full range woofer”. As tested by manufacturer, doubtless done in a baffle, on axis performance looks impressive to >10kHz, with rapid roll off beyond 2kHz off axis, and recommended use 40Hz-4kHz.

The SEAS woofer as mounted is highly omni directional somewhat beyond 1kHz; based on prior knowledge and a brief measurement survey the driver is used with steep crossover at 2kHz.

The Vifa driver was measured extensively to get a good grasp of its performance characteristics. A series of measurements from front and back at 27”suggest an acoustic source that starts at neck of driver cone. The driver was rotated in vertical plane through the cone’s neck at 5° intervals. Dual channel sweeps where used with one for loop back timing reference and the other for response measurement. The use of timing reference and use of measuring tape combined together helped in keeping highly fixed measurement distance. Front of driver on axis is 0°. Backside measurements were made from 90-180 degrees and front side measurements from 0-60 degrees(Yes, 75° is missing). Overlays of results indicate highly dipole radiation pattern to 2kHz. Measurements remain well organized to 4kHz. Above 4kHz cone geometry takes over.

The 0° and 15° degree shows some plausible potential use of the driver full range to 14-15kHz
Overlays 0-60 degrees:

vifa 0-60 inc 5 3ms.png

Overlays 90-180 degrees:

vifa 90-180 inc 5 12ms.png

A choice was made to first explore driver alignment with Vifa rim centered above SEAS dust cap. This puts Vifa acoustic center about 1” behind SEAS acoustic center. 1” represents ½ to ¼ wavelengths about an order of magnitude shorter than 2kHz upper range of driver overlap. Driver effective diameters at 1-2kHz and apparent acoustic centers in this range are likely larger factor in polar pattern formed by superposition of each driver’s polar pattern.


Vifa driver connected as microphone with 220Hz signal to SEAS was used in finding driver position minimizing driver interaction. Vifa rim above center of SEAS dust cap produced best null, indicating equal energy delivery to front and back of Vifa from SEAS. Vifa rim is 1” above SEAS driver’s dust cap.

Great results were obtained for Pluto Clone with 2” tweeter when reference measurements from single measurement point located 9” from face of tweeter, and slightly off of tweeter axis. Direct to reflected ratio is sufficient for direct inversion using Kirkeby method in achieving equalization filters. With Pluto Clone development measurements from 2’ contained too many reflected signals, mostly floor/ceiling reflections, leading to great group delay at specific frequencies. Application of Kirkeby to such measurements leads to extreme ringing at certain frequencies, readily heard when inverse filter is convolved with sweep or music signals. This type of filter is only valid at measurement location. Anticipating this and dipole behavior of reinforcement/cancellation as wavelengths become shorter than driver diameter, and tempered by cone geometry influence near/far, a single measurement point for drivers was chosen about 12” from driver centers, and chosen 15° off axis of Vifa so to capture the flatter high frequency response. Direct use of measurements with Kirkeby resulted in ringing behavior. Application of Blackman-Harris 7 term window via REW with 15ms setting prior to Kirkeby produced working results. Windowing of result to 15ms still provides low frequency information within 1dB of large window results whilst suppressing unwanted reflections. Kirkeby filters applied to respective drivers flatten their responses.

Woofer filter is band pass filtered to 50Hz-2kHz, tweeter is filtered 200Hz-15kHz. Results sum over 200Hz-2kHz range. Secondary filter is used to shelve each driver’s response down 6.02dB across 200Hz-2kHz overlap region. Applying Kirkeby inverse transform to sum of 200Hz high pass filter with 2kHz low pass filter generates the secondary filter. Result is flat sum with monopole behavior 50Hz-200Hz, cardioid behavior 200Hz-2kH, and dipole behavior above 2kHz that is shaped by cone geometry.

Raw Crossover:

raw crossover.png

DSP:

J River Media Player is used for playback of sweep signals, convolution of sweeps with filter kernels, and recording of microphone responses. Previous testing finds software bit perfect. Here are frequency response overlays of measured results 0-180 degrees at intervals of 15 degrees. Microphone is placed 9” from edge of woofer pipe at height of tweeter axis. Tip of microphone is sighted through to protractor ruling around pipe top, and against center of woofer dust cap for achieving rotational alignment. Sweeps are exponential 2Hz-22050Hz with 262144 samples. A series of four sweeps with ½ seconds of silence are used as stereo track. Signal polarity of 2nd sweep of track 1 (woofer) and 3rd sweep of track 2 (tweeter) have polarity inverted. Resultant test sequence is: 1) normal woofer with normal tweeter, 2) inverted woofer with normal tweeter, 3) normal woofer with inverted tweeter, and 4) is repeat of 1). This format allows for sums and differences for verifying integrity of data and extracting images for solo woofer and solo tweeter responses. The four sweep periods are 284194 samples. Each isolated block is convolved with inverse sweep, obtaining impulse response. Resultant IR data are ported into REW for generating overlays of frequency responses.

Using Room EQ Wizard software, frequency response overlay views of 13 measurements 0-180 degrees for normal woofer polarity and for reverse woofer polarity is set up. Flat response region for 0 degree normal woofer polarity measurement across 300Hz-800Hz is offset to 0dB and all other results shifted by same offset. The same plot colors are used for corresponding rotation angles for both overlay views:

normal woofer signal.png

With woofer polarity reversed:

inverted woofer signal.png

Monopole behavior is seen centered at about 55Hz.

Strong cardioid behavior is seen 200Hz-1kHz, indicative both of good monopole driver behavior and of good dipole behavior.

Measurements with microphone at reference point:

cardioid 100hz-2khz 450p.png

Last edited by Barleywater; 18th February 2015 at 02:11 AM.
  Reply With Quote
Old 18th February 2015, 03:24 PM   #2
Davey is offline Davey  United States
diyAudio Member
 
Join Date: Dec 2001
Location: Bremerton, WA.
Andrew,

Very interesting.

I don't think I missed it, but in this case you've suspended (strings from the ceiling?) the Vifa driver completely open-air with no structure attached to it at all?
How transferrable do you think the results would be with other 6.5" drivers of similar construction?

I'm not sure I have the computer horsepower to implement a similar setup. You're running on a desktop machine (maybe Windows 7 64-bit) and incurring quite a bit of latency?

Cheers,

Dave.
  Reply With Quote
Old 18th February 2015, 04:13 PM   #3
architect is offline architect  Netherlands
diyAudio Member
 
Join Date: Nov 2004
Location: Friesland
pictures/photos?
  Reply With Quote
Old 20th February 2015, 09:04 PM   #4
speaker dave is offline speaker dave  United States
diyAudio Member
 
speaker dave's Avatar
 
Join Date: Nov 2009
Location: The Mountain, Framingham
Hi Barleywater,

Impressive work all around!

You seem to be getting ideal performance in your overlapping dipole/monopole region. Well above that region you are doing as well as the full range driver is capable of (it rolls off then comes back to life off axis). I wonder if there is any way to improve the transition between the two regions? Is a hard transition mandatory?

I assume a reasonable phase match of the of the dipole and monopole are required for ideal polar patterns. I know you displaced the drivers to get good time allignment of the two. I wonder if a slow transition could be phase compensated?

Anyhow, interesting stuff.

David
  Reply With Quote
Old 22nd February 2015, 10:59 AM   #5
Barleywater is offline Barleywater  United States
diyAudio Member
 
Join Date: Jan 2008
A 1/4" steel rod was fitted to one of Vifa driver basket arms with small hose clamps and a counter balance. A small hose clamp at pivot point was used with hook formed by clamp's excess tail catch top tube of folding music stand:

Vifa over SEAS 2.jpg

Filters used are about 22k taps. Computational power for convolution is not an issue. I've run 6x 32k tap filters with Pentium III 500MHz laptop.

For music playback latency is non issue, other than lag of volume control. If I am not mistaken, JRiver Media software has facility for delaying video for synching.

In the case of this setup intention is to demonstrate feasibility of this approach. Filters use up a lot of dynamic range. SPL with heavy bass is quite limited. Scale up would likely use 8-10" drivers and second pairing of smaller monopole/dipole pair; the trick is where to put second monopole.

Goal in this case is to maximize cardioid pattern in fundamental range of most music and voice.

SEAS driver has nasty metalic sound above 2kHz. Widening the overlap upwards promotes metalic sound; and driver pair becomes less coincident. Widening overlap downwards degrades cardioid pattern to hyper/super cardioid with backside lobes.

Impulse response:

ir.png

Frequency response and phase:

spl and phase zoom.png
  Reply With Quote
Old 4th June 2015, 06:09 PM   #6
RNMarsh is offline RNMarsh  United States
diyAudio Member
 
RNMarsh's Avatar
 
Join Date: May 2012
Location: 2457 Cascade Trail; Cool, CA. 95614
Are these measurements truely of group-delay or just convolutions to compliment and correct for frequency response to be flat?

Is the group-delay flat as well?


THx-RNMarsh
  Reply With Quote
Old 4th June 2015, 08:34 PM   #7
Barleywater is offline Barleywater  United States
diyAudio Member
 
Join Date: Jan 2008
Convolution is used in JRiver Media player to apply correction filters to source file, so music and test signals are treated the same.

Click the image to open in full size.


This is the measured GD using 110ms Blackman-Harris 7 term window.

Thanks for the interest.

-Andrew
  Reply With Quote
Old 16th June 2015, 08:44 PM   #8
vacuphile is offline vacuphile  Netherlands
diyAudio Member
 
vacuphile's Avatar
 
Join Date: Apr 2011
Location: Seaside
Barleywater, those are outerworldy filterslopes and ditto results. I am digging into FIR filtering, and one of the innate disadvantages appears to be pre-ringing. Is this something to worry about in your experience?
  Reply With Quote
Old 17th June 2015, 04:16 AM   #9
Barleywater is offline Barleywater  United States
diyAudio Member
 
Join Date: Jan 2008
When delving into FIR filters I initially worried a lot based on all the commotion an IR waveform causes when people see waves to the left of where all the high frequency energy is centered. It only drove me to find out first hand what all the flap is about.

I strive for driver alignment including phase and amplitude match across overlap region, and getting driver centers within 1/4 wavelength for XO overlap frequencies. For vertically oriented driver centers worst response is on vertical axis. FR has 3dB dip there, and maximal ringing. For drivers XO at 1kHz ringing is 0.25ms.
Difference in sound for IIR based transients on vertical v horizontal axis are discernible in direct comparison by A/B of recordings or simulations. I don't generally listen on vertical axis.


My initial interest was to get cardioid response from 200Hz-1kHz, and attempted 2kHz to see how the system deviated. The reverse woofer polarity 3dB peaking in the polars at 1.5kHz is do primarily to 1/4 wave behavior of the two drivers.

The dipole driver location above the woofer was chosen by hooking the dipole up as a microphone, playing a low frequency tone through the woofer, and adjusting the dipole location for minimizing picked up signal.


I'm looking forward to what you dig up.
  Reply With Quote
Old 17th June 2015, 08:07 AM   #10
vacuphile is offline vacuphile  Netherlands
diyAudio Member
 
vacuphile's Avatar
 
Join Date: Apr 2011
Location: Seaside
Quote:
Originally Posted by Barleywater View Post
When delving into FIR filters I initially worried a lot based on all the commotion an IR waveform causes when people see waves to the left of where all the high frequency energy is centered. It only drove me to find out first hand what all the flap is about.

....

I'm looking forward to what you dig up.
That is exactly what has been putting me off, but if I translate your post correctly into Australian, what you are saying is "no worries, mate".

I don't think I have any option other than getting into this technology. I can get speakers with an analog active crossover to within +/- 1.5 dB, with only corrections for delay and baffle step. This is not bad at all, but the newest speakers that come out are as flat as a salt plane. It is really amazing to see how, using this technology, you were able to draw a funky contraption of speakers as straight as you did.

An important thing I learned for your post is that I do not need to invest right away in a MiniSharc or another FIR capable DSP, but that I can take my first baby steps using hardware I already have - a discarded XP PC. Since you only use two channels in your experiment, I suppose you are using the stereo outputs on your PC. However, I would like to play in stereo and need three channels per speaker, so that calls for a solution. Perhaps a multi channel sound card will do the trick. Do you, or readers of this thread, have hands on experience with such a setup?

As soon as I get something that works I'd be most happy to share my results.
  Reply With Quote

Reply


Cardioid as sum of monopole and dipole speakers:Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump

Similar Threads
Thread Thread Starter Forum Replies Last Post
Getting Dipole bass out of a monopole subwoofer Hara Subwoofers 19 3rd January 2016 10:00 PM
Cardioid/Dipole woofer drchait Multi-Way 1 15th December 2011 05:10 PM
dipole monopole or? Kjeldsen Multi-Way 38 28th October 2011 11:30 AM
Omnipole, monopole, dipole and...nopole?? terry j Multi-Way 25 1st July 2007 03:28 AM
Dipole vs monopole, balls or not ... ? Jussi Multi-Way 11 4th May 2006 03:38 PM


New To Site? Need Help?

All times are GMT. The time now is 03:58 AM.


Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 15.00%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.
Copyright ©1999-2017 diyAudio
Wiki