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

OmniDirectional - work in progress
OmniDirectional - work in progress
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 7th August 2017, 06:26 AM   #141
fluid is offline fluid  Australia
diyAudio Member
 
Join Date: Jan 2009
Good work looks like you are closing in For your correction have you used minimum phase EQ and then removed the excess phase, I think from memory that was the way swiss bear went?

From doing a similar thing recently I would say to look how closely the IR peaks are aligned before you average them. For me REW did not do a great job of doing it with the time align function. Your's might be different but it does have an effect on the high frequencies if the impulses aren't within a certain tolerance when they are averaged. Make sure the averages are run on raw data too just to be sure, leave any windows or smoothing for after.

The impulse averaging weights the dips more than the peaks, have a look in the all spl window with all of the individual measurements open to see if there are any areas where the average is not representative of the majority of positions. EQ'ing those up could result in peaks which might be more objectionable than a dip.

Have a look at the frequency response through a Frequency dependant window too, if there are any areas where there are a number of dips close together they will show up quite clearly, those are the trouble spots.

The spectrogram is a good way of looking at the time data too.

Impulses are tricky to interpret sometimes, there is some unusual pre-ringing in yours. That could be the result of phase correction or something funky in your chain. Take a loopback measurement of your soundcard to see and set the processing to flat to see if there is a change there. Some DAC's have some quite strange impulse responses due to their filtering.
  Reply With Quote
Old 8th August 2017, 02:42 PM   #142
DonVK is offline DonVK  Canada
diyAudio Member
 
DonVK's Avatar
 
Join Date: Jan 2017
Location: Ottawa
Quote:
Originally Posted by fluid View Post
Good work looks like you are closing in For your correction have you used minimum phase EQ and then removed the excess phase, I think from memory that was the way swiss bear went?
That's correct, it was min phase and remove excess. My amplitude corrections were manual in RePhase only, I prefer to match the filter [fc, G, Q] to the response defect. I have only corrected (so far) using the standard EQ PK forms

Quote:
Originally Posted by fluid View Post
From doing a similar thing recently I would say to look how closely the IR peaks are aligned before you average them. For me REW did not do a great job of doing it with the time align function. Your's might be different but it does have an effect on the high frequencies if the impulses aren't within a certain tolerance when they are averaged. Make sure the averages are run on raw data too just to be sure, leave any windows or smoothing for after.
Good suggestion. I have not paid much attention to alignment. I just let REW generate min phase and remove offsets. The FR data was raw before averaging.

Quote:
Originally Posted by fluid View Post
The impulse averaging weights the dips more than the peaks, have a look in the all spl window with all of the individual measurements open to see if there are any areas where the average is not representative of the majority of positions. EQ'ing those up could result in peaks which might be more objectionable than a dip.
Looks like you want to remove the "outliers". None of them looked weirder than the others. Maybe I'll try some subset combinations.

Quote:
Originally Posted by fluid View Post
Have a look at the frequency response through a Frequency dependant window too, if there are any areas where there are a number of dips close together they will show up quite clearly, those are the trouble spots.

The spectrogram is a good way of looking at the time data too.
It is possible that my manual corrections, actually corrected things that were not there There was no FDW performed before averaging. I'll see if any weirdness remained via FWD on the avg.


Quote:
Originally Posted by fluid View Post
Impulses are tricky to interpret sometimes, there is some unusual pre-ringing in yours. That could be the result of phase correction or something funky in your chain. Take a loopback measurement of your soundcard to see and set the processing to flat to see if there is a change there. Some DAC's have some quite strange impulse responses due to their filtering.
Lookback may be a problem. My output chain is [PC]->S/PDIF->RVX659->speaker. My input chain is [Mic]->[PC] . I might be able to take a line out from the amp to line in on the PC for loopback.

Thanks again for the help.
  Reply With Quote
Old 8th August 2017, 03:06 PM   #143
fluid is offline fluid  Australia
diyAudio Member
 
Join Date: Jan 2009
Quote:
Originally Posted by DonVK View Post
Looks like you want to remove the "outliers". None of them looked weirder than the others. Maybe I'll try some subset combinations.
Not quite what I meant, it might be that in the area you measured and with your speakers being omni you didn't get anything that was not representative in the average.

Because the dips are weighted more heavily it is possible when impulse averaging for a dip to appear in the average which is worse than reality if that makes sense. Correcting that might be an overcorrection in reality.

Quote:
It is possible that my manual corrections, actually corrected things that were not there There was no FDW performed before averaging. I'll see if any weirdness remained via FWD on the avg.
You can check afterwards in the same positions to make sure you didn't overdo anything.

Only use the FDW to view the average afterwards don't use FDW before averaging that will give weird results I would think.


Quote:
Lookback may be a problem. My output chain is [PC]->S/PDIF->RVX659->speaker. My input chain is [Mic]->[PC] . I might be able to take a line out from the amp to line in on the PC for loopback.

Thanks again for the help.
It is very hard to get a good impulse when the input and output cards are run on different clocks. For measurement it is much better to use the same soundcard as input and output device if possible.

I had the same issue when I used my USB connected DSP board as output and then measured via my Focusrite soundcard. I got a much cleaner impulse when using the Focusrite for playback and recording.

Happy to help where I can
  Reply With Quote
Old 8th August 2017, 03:45 PM   #144
Juhazi is offline Juhazi  Finland
diyAudio Member
 
Juhazi's Avatar
 
Join Date: Mar 2007
Location: Jyväskylä, Finland
Quote:
Originally Posted by DonVK View Post
.... These are all measured at my listening position 2.4m from the speakers with no room treatment and no subwoofer.
Don, I just noticed this thread. As per my own experience with measurements, you have a major basic fault now. You show measurements done at farfied in a normal room, obviously with standard 500ms gating. That is totally wrong type of measurement to equalize and rephase a speakers output/phase! You see just summation of hundreds of in-room reflections and modes. This method is ok for low frequency eq or tonal balance adjustment with 1/2 octave smoothing.


To analyze and adjust a loudspeakers output (frequency response at various angles, phase response, step response, distortion, mechanical resonances etc.) you must do nearfield measurements at say 0.6 to 1m distance, with the speaker in the middle of the room and analyze the measurement with IR gating (right window) set to say between 3-9ms. First boundary reflections usually start showing already at 2.5-3ms!


A good link to how to do it right https://www.minidsp.com/applications...r-measurements
__________________
AINOgradient speaker project

Last edited by Juhazi; 8th August 2017 at 03:47 PM.
  Reply With Quote
Old 8th August 2017, 05:33 PM   #145
DonVK is offline DonVK  Canada
diyAudio Member
 
DonVK's Avatar
 
Join Date: Jan 2017
Location: Ottawa
Quote:
Originally Posted by Juhazi View Post
Don, I just noticed this thread. As per my own experience with measurements, you have a major basic fault now. You show measurements done at farfied in a normal room, obviously with standard 500ms gating. That is totally wrong type of measurement to equalize and rephase a speakers output/phase! You see just summation of hundreds of in-room reflections and modes. This method is ok for low frequency eq or tonal balance adjustment with 1/2 octave smoothing.


To analyze and adjust a loudspeakers output (frequency response at various angles, phase response, step response, distortion, mechanical resonances etc.) you must do nearfield measurements at say 0.6 to 1m distance, with the speaker in the middle of the room and analyze the measurement with IR gating (right window) set to say between 3-9ms. First boundary reflections usually start showing already at 2.5-3ms!


A good link to how to do it right https://www.minidsp.com/applications...r-measurements

There may be a misunderstanding in what is being attempted here.

My earlier measurements have been near field (1m) as you've suggested, and they were used to correct speaker issues. I have a preferred location for these near field measurements that has minimal room reflections.

This latest set of measurements treats the speaker and room as a system and takes multiple measurements around the listening area. In my case its a smallish sweet spot around my listening couch. It's an attempt to get corrected sound to the ears. It excludes LF (<100Hz) as these are room modes. Overall this system correction sounds better than the speaker specific near field corrections.
  Reply With Quote
Old 8th August 2017, 07:54 PM   #146
Juhazi is offline Juhazi  Finland
diyAudio Member
 
Juhazi's Avatar
 
Join Date: Mar 2007
Location: Jyväskylä, Finland
Ok sorry I didn't read through the whole story.

But please do many measurements with varying speaker an mic location to learn what reflections cause which dips and peaks. Also for farfield eq use smoothing 1/2 or 1/3 and don't do any eg operations that are narrower either!

Moving Microphone Measurement technique (MMM) with RTA or averaging of several measurements is also helpful
http://www.ohl.to/audio/downloads/MM...easurement.pdf
Moving Mic Measurement
__________________
AINOgradient speaker project
  Reply With Quote
Old 10th August 2017, 01:16 PM   #147
fluid is offline fluid  Australia
diyAudio Member
 
Join Date: Jan 2009
Quote:
Originally Posted by Juhazi View Post
But please do many measurements with varying speaker an mic location to learn what reflections cause which dips and peaks. Also for farfield eq use smoothing 1/2 or 1/3 and don't do any eg operations that are narrower either!
If you read the whole thread you would see that Don has taken a lot of close up measurements and measurements around a full 360 to verify that his speaker was acting as an omni.

I would agree that getting as close to anechoic measurements as you can is a good idea for setting up crossovers, looking for resonances etc. Don't be too quick to think that you it is not possible to use more detailed EQ when measured from a distance. It just needs some thought and understanding of what you are doing.

Don is using a spatial averaging technique which helps to avoid overcorrection and includes the most persistent room issues. While many people have done room EQ very badly in the past it doesn't mean that some of the software and techniques cannot be used in a better way and achieve good results.

Moving microphone is a form of spatial averaging that discards phase. That can be useful but so can impulse averaging. If you haven't seen it before have a read of the paper on the beamforming measurement method. They use a straight line but the principle is the same.

By using enough averages you can really suppress the room contribution and are only left with significant modal and non position dependant room issues in the measurement. Those are actually useful if you want to try and improve them through EQ and processing rather than acoustic means.

It's always better to deal with a problem at the source but not everyone has that luxury.
  Reply With Quote
Old 12th August 2017, 05:08 AM   #148
DonVK is offline DonVK  Canada
diyAudio Member
 
DonVK's Avatar
 
Join Date: Jan 2017
Location: Ottawa
Default Impulse response improvements via DRC

In an earlier test, RePhase was used to correct the avg room FR room response but resulted in a poor impulse response OmniDirectional - work in progress .

With some generous help from @Fluid, who used another tool (DRC) to create a filter than cleaned up the impulse response significantly.

Both the RePhase corrections and the DRC corrections significantly cleaned up the measured responses and improved the sound quality. This is the best Omni solution to date and its time to repackage it into something more finished looking.
Attached Images
File Type: jpg measured impulse left.jpg (83.9 KB, 187 views)
  Reply With Quote
Old 12th August 2017, 05:29 AM   #149
DonVK is offline DonVK  Canada
diyAudio Member
 
DonVK's Avatar
 
Join Date: Jan 2017
Location: Ottawa
Default Compared to what ?

This is a comparison to other speaker solutions. The OmniV4 has compensation as previously described. The listening room I use is setup as per the attached pic. The signal path is [PC] -> S/PDIF -> RXV659 -> Speakers. Stereo upmixing is in the RXV659 but Yamaha does not disclose the algorithms. The comparison set are Paradigm [monitor 7v5 for LR, mini monitors for SR+SL, PDR10 sub, CC370 center]

1) Paradigm Stereo 2.0
- normal stereo, phantom center slightly behind mid speaker line
- sound stage generally as wide as the speakers
- the room is wide enough that there's little lateral reflections from these
- these are 2.5 way tower speakers, and bass is weak
2) Paradigm Stereo 2.1
- same as above only the bass is improved (XO=80Hz)
- its a narrow slice (30-80Hz) but makes a huge improvement to the sound
3) Paradigm Stereo upmix 3.1
- center speaker (2 way MTM) is on
- stereo center is more solid (less position sensitive) + forward of LR pair
- relative volume w.r.t LR pair can move position forward
- sound stage width the same as 2.0
4) Paradigm Stereo upmix 4.0
- front LR and side LR (2way) are on
- still has the phantom center like 2.0
- the sides can be used to delay sound to mimic a larger room
- just like an acoustic center for the front, there is one for the side now
- sound stage is much wider than 2.0 from lateral sound
- phantom center appeared to move forward in line with speakers
5) Paradigm Stereo upmix 5.1
- very wide soundstage and solid stereo center
- like the 4.0, lateral sound widens the stage
6) OmniV4 Stereo 2.0
- phantom center is aligned to LR pair
- soundstage is wider than Paradigm 2.0 not as wide as Paradigm 5.1
- bass from 8" driver insufficient for this room
7) OmniV4 Stereo 2.1
- same as OmniV4 stereo 2.0 only the bass is improved (XO=80Hz)
Attached Images
File Type: jpg speaker setup reduced.jpg (87.5 KB, 190 views)

Last edited by DonVK; 12th August 2017 at 05:39 AM. Reason: not finished
  Reply With Quote
Old 19th August 2017, 03:08 AM   #150
DonVK is offline DonVK  Canada
diyAudio Member
 
DonVK's Avatar
 
Join Date: Jan 2017
Location: Ottawa
Default Omni V5 - construction - cones

These are some construction pics for V5. Probably the hardest construction detail is making the cone shapes. It can be done in 3D printing or fiberglass molds but there is another easier means. Generating the profile can be done in an excel spreadsheet and cut into a ply sheet as a "slot" . Its fairly easy to make a curve tracing fixture, that will allow a grinder to follow the profile slot to initially bulk machine polystyrene foam. The fixture can then be reused with a variable speed hand drill to spin the workpiece like a wood lathe to sand smooth the surface. These are common hand tools on hand, and the fixture is scrap wood with $20 of bushings, bolts and collars. The foam+sureply backing is about $6 per cone. The tips and final painting will be added later.
Attached Images
File Type: jpg Omni cone mill reduced.jpg (186.6 KB, 151 views)
File Type: jpg DSCN2024 reduced.jpg (734.2 KB, 152 views)
File Type: jpg DSCN2026 reduced.jpg (338.2 KB, 140 views)
File Type: jpg DSCN2027 reduced.jpg (488.8 KB, 46 views)
File Type: jpg DSCN2029 reduced.jpg (429.7 KB, 55 views)
File Type: jpg DSCN2032 reduced.jpg (624.5 KB, 66 views)
File Type: jpg DSCN2033 reduced.jpg (623.8 KB, 65 views)
File Type: jpg DSCN2035 reduced.jpg (389.9 KB, 75 views)
File Type: jpg DSCN2036 reduced.jpg (702.1 KB, 75 views)
File Type: jpg DSCN2038 reduced.jpg (586.5 KB, 72 views)

Last edited by DonVK; 19th August 2017 at 03:09 AM. Reason: grammer
  Reply With Quote

Reply


OmniDirectional - work in progressHide 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
Work in progress... Asio4All Q pcb121055 PC Based 3 25th November 2011 05:59 AM
Omnidirectional midrange driver - will it work? el`Ol Planars & Exotics 3 19th November 2007 09:00 AM
FE167E finally ! work in progress... ow31 Full Range 27 28th June 2006 10:03 AM
BZLS work in progress till Pass Labs 10 26th February 2004 08:34 AM
Aleph 5 - work in progress SuppersReady Pass Labs 23 14th November 2003 10:40 PM


New To Site? Need Help?

All times are GMT. The time now is 01:01 AM.


Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 14.29%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.
Copyright ©1999-2017 diyAudio
Wiki