Home Theater Forum and Systems banner

1 - 20 of 45 Posts

·
Registered
Joined
·
81 Posts
Discussion Starter #1
Hi
I have built "Xovers" for an 8ch setup using REW's sweep measurements with great success.

I have understood that the crossover phase matchs between Sub, woofer, mids and tweeters are extremelly important. The "IR" response gives the exact time delay to apply for the best phase "XO" matching.

I'd like to know why the number of impulse waves on my 8" mid speakers is higher than from my 15" woofers, for the same frequency range at the "XO" point? This may be obvious for you, but I confess my ignorance. It is the same between my 15"woofers and the 15" Subwoofer. For a fixed frequency I just can imagine the same transductor mouvement.... So Impulse is something else?

jacques
 

·
Registered
Joined
·
239 Posts
How do you verify the success of Xovers, and how do you build them with sweep measurements?

8" mid is band pass with time domain wave structure for rejecting both low frequencies and high frequencies. The low pass stop band is reason for seeing more zero crossings in IR of the mid XO filter.

Andrew
 

·
Registered
Joined
·
81 Posts
Discussion Starter #3
My last question was so embarrassing, I figured out that the IR responses were not at the "XO" point.
How do you verify the success of Xovers, and how do you build them with sweep measurements?
I use sweep measurements to verify phase between overlapping speakers (phase plot overlays), by adjusting time delays (IR plot overlays), "XO" slopes and frequency cuts to get the best fits.
I have built 8ch "XOs" within Jriver, folowing;
EQ1 is only for frequency EQ
EQ2;
1- copy right to subwoofer
2- copy left to subwoofer (1 and 2 brings the signals to 1 subwoofer)
3- copy left to center
4- copy right to center (3 and 4 brings signals to the time reference physical loopback "one at the time" for REW measurements)
5- copy left to RL
6- copy right to RR (5 and 6 brings the signals to woofers)
7- copy left to SL
8- copy right to SR (7 and 8 brings the signals to tweeters)
9- high-pass at 15hz (sub) 24db/oct (low cut for subwoofer)
10- low-pass at 60hz (sub) 48db/oct (high cut for subwoofer "XO")
11- high-pass at 50hz (RL-RR) 48db/oct (low cut for woofers "XO")
12- low-pass at 150hz (RL-RR) 48db/oct (high cut for woofers "XO")
13- high-pass at 120hz (right-left) 36db/oct (low cut for mids "XO")
14- low-pass at 4000hz (left-right) 36db/oct (high cut for mids "XO")
15- high-pass at 3500hz SL-SR) 36db/oct (low cut for tweeters "XO")
16- low-pass at 20000 (SL-SR) 36db/oct (high cut for tweeters "XO")
17- delay 17.5ms (RR) right woofer
18- delay 17.5ms (RL) left woofer
19- delay 26.8ms (right) right mids
20- delay 26.8ms (left) left mids
21- delay 0ms (SL) left tweeter
22- delay 0.2ms (SR) right tweeter

I hope these samples of woof-mid phase and SPL plots are speaking by themselfs?

8" mid is band pass with time domain wave structure for rejecting both low frequencies and high frequencies. The low pass stop band is reason for seeing more zero crossings in IR of the mid XO filter.

Andrew
Can you comment on the these graphs?

jacques
 

Attachments

·
Registered
Joined
·
239 Posts
Quite the recipe.

I see:

driver: Band: HP slope: LP slope: *slope in dB/octave
sub 15Hz-60Hz 24 48
woofer 50Hz-150Hz 48 48
mid 120Hz 4kHz 36 36
tweeter 3.5kHz-20kHz 36 36

So, filters are Linkwitz-Riley, Butterworth?

I see overlaps in bands? How did you arrive at this? And mixed filter slopes from woofer to mid?

Impulse response view and phase plot view not very revealing, yes phases track roughly parallel, but scale of view doesn't resolve well.

Frequency response overlays of woof and mid look like a cross, but how do they perform?

Is sum (real response) flat, and with leads to one driver reversed is response at crossover point nice deep narrow notch?

I too do active speakers with convolution engine, and get very good measured results. Crossovers built in Cool Edit, and drivers responses corrected with room correction techniques.

I suggest measuring tweeter/mid performance from 9"-24" microphone on tweeter axis with small window applied to results for mid/woofer if speaker face isn't too big you can still measure fairly close, but 24"-30" works well. If possible elevate speaker to help increase time for floor reflection, and gate window to no more than 10ms. For woofer/sub 3'-5' speaker back on floor, microphone 6"-24" off floor compare big window to gated window of about 20ms. Awesome if you can measure outside away from structures, then its just ground plane reflection, which exists in most situations.


Andrew
 

·
Registered
Joined
·
81 Posts
Discussion Starter #5 (Edited)
Hi Andrew,
Many interesting questions
Quite the recipe.

I see:

driver: Band: HP slope: LP slope: *slope in dB/octave
sub 15Hz-60Hz 24 48
woofer 50Hz-150Hz 48 48
mid 120Hz 4kHz 36 36
tweeter 3.5kHz-20kHz 36 36

So, filters are Linkwitz-Riley, Butterworth?
The high-pass and low-pass filters are Butterworths

I see overlaps in bands? How did you arrive at this? And mixed filter slopes from woofer to mid??
I first did several tests and at the end (temporary) I want to use speakers at their best spectral performances. So I wanted to have an "XO" around 130hz (mid 8"-woof 15"), from there I tried different frequency cuts and slopes to get around this to be in phase without a dip combining them. I had to compensate on the left side gain because of diffrent wall reflections.
Impulse response view and phase plot view not very revealing, yes phases track roughly parallel, but scale of view doesn't resolve well.
For this post, I used a shorter sweep frequency lenght zoomed on the "XO". A larger sweep response is more reveiling of a nice fit.
Frequency response overlays of woof and mid look like a cross, but how do they perform?.
This setup is one of the best I've been able to make. The Supravox 215 RTF64 (50hz-10Khz) gives a very tight low mid response for my acoustic jazz music. I didn't want the large 15" woofers to go further on HF.

Is sum (real response) flat, and with leads to one driver reversed is response at crossover point nice deep narrow notch?
No, at real response, I paid attention to avoid that dip at the sum region. This is part of the freq. cuts-slope-gain tuning.

I too do active speakers with convolution engine, and get very good measured results. Crossovers built in Cool Edit, and drivers responses corrected with room correction techniques.

I suggest measuring tweeter/mid performance from 9"-24" microphone on tweeter axis with small window applied to results for mid/woofer if speaker face isn't too big you can still measure fairly close, but 24"-30" works well. If possible elevate speaker to help increase time for floor reflection, and gate window to no more than 10ms. For woofer/sub 3'-5' speaker back on floor, microphone 6"-24" off floor compare big window to gated window of about 20ms. Awesome if you can measure outside away from structures, then its just ground plane reflection, which exists in most situations.


Andrew
When I started this project, I was testing as you suggest at the speakers. At the end I was doing so many EQ room corrections to acheive flat curve that the "XO" setings where to modified.
Now I'm looking for the acoustic "XO" to be at my listening position. Maybe I should do the two positions?
All setting are at their best from the chair position to include all reflections, acoustic treatments have been extensivelly worked out and it is still a work in progress.
.

I found your questions highly appropriate. I'd like to see some of your measurement for comparison with my work. It could be helpfull for the two of us.

+
jacques
 

·
Registered
Joined
·
239 Posts
It is difficult not to get sucked into complex effects of room acoustics at the preferred listening position, making it easy to overlook the behavior of the recording/playback system.

For example, it is widely excepted that when recording voice, that the principle microphone is located inches from the mouth. We don't typically listen to singers at that distance. If the recording is done in acoustically dead studio, the result is dry and typically needs reverberation added back in to get a palatable recording. In a not so dead recording studio of suitable size, often not much larger than a fair size living room, a situation often encountered and often preferred by vocalists and recordists alike, a close microphone is still used, but also secondary microphone for room sound. Even without the secondary microphone, the principle microphone picks up a fair amount of ambient sound in the room. An observer sitting in on the recording session will hear the live unamplified voice, and it will sound natural from many points in the studio. The odd wall, floor, and ceiling reflections don't even register in the observers mind.

But sound measurement system leads to conclusion that response of speaker isn't flat because of floor, wall and ceiling reflections. A vicious endeavor is undertaken to get back to flat sound by eliminating, or some how compensating for what are perceptual system has grown up with, and evolved with.

So back to the vocalist: The time dependent sound pressure variations of the voice are captured by the microphone, and continue past the microphone into the studio, and into the ears of observer enjoying live performance. The best reconstruction of that performance would be replacing the singer with a speaker capable of reproducing the time dependent sound pressure variations captured by the microphone.

But what happens? The speaker is put in place and the same microphone records the playback, and the two waveforms are lined up for comparison. Significant deviations are seen. The spectrum from corresponding points in the two recordings are not the same, reflecting in part that the speaker frequency response isn't flat. Closer analysis additionally indicates that individual spectral components from the original recording and the recording made from the speaker experience frequency dependent shifting: phase distortion as group delay and energy storage with delayed release. The experiment can be repeated with a wide variety of speakers, each with similar results, some better some worse.

What does the observer in the studio hear, when a speaker is found that when recorded produces a virtually identical waveform to the original vocal recording? A very convincing reproduction!

That speaker when played in another room sounds like the vocalist is singing in that room. If the vocalists recording contains ambient information, and the room for playback isn't too live, the listener has little trouble being engaged by the vocalist and the sound of the recording venue.

Such speakers exist. This is my version: Full DSP Pluto Clone, inspired from Linkwitzlab's Pluto active speaker.

A fundamental flaw in use of digital room correction methods, such as DRC, Audiolense, Acourate, and Dirac Live, along with variations on a theme such as Audyssey type AVR correction systems is simultaneously trying to correct for both room and speaker, instead of starting with just the speaker.

Anyway, you've got a handle on multichannel convolution, basic XO generation, and EQ, along with measurement. You're probably hearing sound on a par with some of the better active monitors. With addition of inverse transfer function (this is core of all room correction) tricks and techniques as applied to drivers, you can take this to a whole new level. Interested?

Andrew
 

·
Registered
Joined
·
81 Posts
Discussion Starter #7
Thats a nice story Andrew, I liked it.

This is a lot more than a simple impulse wave thing :gulp: Your description of the recording studio is surprisingly clear for a quite difficult subject. In fact we are slaves of bad or good recording procedures.
All we can do is trying to reproduce what is on the CD as the source. We know that the source is the performance, but it has to go through so many obsticals "filters", and sound engineers.
Though a neutral "transparent" system is the only way to go back to the source, or by default what is on the CD. Your talking about room ambiances. If it's not recorded I don't want to ear a fake one.
The EQ freq. correction "flat curve" is the last oparation to performed after every possible electronic "XO" and latency corections and acoustic treatments.
I don't beleive much in DSP automatic room correctors. A Mic wont differentiate a good from a bad flor-wall reflection. The soft may discriminate first waves from wall reflections but after that? Does it sound real?

One must be able to judge the "plausibility" of the instrument. That a friend concept to work at the end with our ears, and I agree with him.

So back on the "Full DSP Pluto Clone", frankly I will have to get into it before saying anything wise.

But I do want to learn more.

jacques
 

·
Registered
Joined
·
239 Posts
What are you using for tweeters? How are your drivers mounted, closed box, ported, open baffle? Picture?

What is your measurement setup, in particular microphone (calibration files?) and soundcard(does it have ASIO drivers)?

Are you PC or Mac?

Size of listening space?

What length filters are you using for convolution XO?


Thanks, I am glad you liked my story. The subject is difficult and complex, and I've tried on several occasions to find fairly simple, thought provoking description of role of speaker in recorded music.

Andrew
 

·
Registered
Joined
·
81 Posts
Discussion Starter #9 (Edited)
What are you using for tweeters? How are your drivers mounted, closed box, ported, open baffle? Picture?

What is your measurement setup, in particular microphone (calibration files?) and soundcard(does it have ASIO drivers)?

Are you PC or Mac?

Size of listening space?

What length filters are you using for convolution XO?


Thanks, I am glad you liked my story. The subject is difficult and complex, and I've tried on several occasions to find fairly simple, thought provoking description of role of speaker in recorded music.
Andrew
Andrew
I'm back and I'll put some pictures here.
"What are you using for tweeters? " I use supravox
-TG1 tweeters titane - 98 dB - 2.5 kHz / 20 kHz
-215 RTF 64 97db- 50 / 10kHz
mounted on an "open baffle", 1"1/4 massive plexiglass
-400 GMF - 40 cm - 96 dB - 24 Hz / 4 kHz
-15" subwoofer Oaudio bash 500w
Woofers and subwoofers are mounted as "RJ" baffle, by Frank Robbins and William Joseph,

For measurements I use a Samson MM01 and the calibration curve looks OK?
I run on a stable dedicated PC win xp pro sp3 and a RME HDSP 32AES uses ASIO, java setup for REW via VAC virtual audio cable to be able to loopback the signal within Jriver's MC17.

All my filters are on MC17.

The real chalange for me is to bring the acoustical "XOs" at my listening place in the best phase matching way.

Since I use 8 channels outputs with individual digital contrôl, I was able to EQ each channels L-R then the phase matchs at XO where much easyer. The 8 digital RME output are sent to 4 DACs, 3 bel-canto and one temporary motu dac for subwoofer (pict).
The picture shows 2 poweramps, one 2ch Linn 2250 and one 5ch Linn 5125. The subwoofer has is own 500w bash poweramp.
I've worked on acoustic wall-ceiling reflections (pict)
Though, there is a lot of testing to come.
I have to find out the exact location for each "XO" that will bring the best of every speakers. I go up and down and have to redo all measurements everytime and time evaluation between setups.
I now fixed 50hz (sub-woof), 112 hz (woof-mids) and 3.5Khz (mids-tweeters).

If anyone had experienced supravox spkers in open baffle config, let me know your results.
 

Attachments

·
Registered
Joined
·
239 Posts
Each impulse wave may be simple, but you've gotten way beyond simple here! This is very large format indeed.

The creator/inventor of original Pluto, Siegfried Linkwitz, is creator of Orion speaker, an OB speaker. His professional career was in electronics, and he does all his work in analog domain. He shares wealth of knowledge and tools for speaker design, and in particular OB speakers. Pluto started as an afterthought and he found it to beat all his expectations. Both he and others that listen to the two speaker systems say the two are difficult to tell apart across lots of program material.

There are of course great similarities between your creation and the Orion: Driver sizes. Orion uses similar sized midrange and tweeter. In Orion, total surface area of two smaller woofers is close to surface area of your single larger woofer. Your baffle for midrange and tweeter is much bigger, and must be very heavy. Machining of cutouts appears exceptional.

Midrange and its 215 RTF 64 Bicone twin seem designed in mind with single driver enthusiasts in mind. If Supravox claims of extended range, as in their measurement graphs are true, then very good single driver listening results are possible.

Supravox also designs for very high efficiency, probably targeting tube enthusiasts and the low watt crowd. Achieving this efficiency occurs chiefly through two mechanisms: Tightening up motor gap tolerances, and using lightweight cone. Tight gap requires added detail to suspension performance to avoid rub. Lightweight cones tend to have much more complex breakup modes of vibration. These have numerous impacts on driver performance, including directivity performance and distortion performance.

The tendency when looking at basic frequency response of such a driver is thinking to use its extended performance to raise crossover point to tweeter, thus reducing tweeter demands. Doing this includes more breakup modes. Breakup mode excitation is frequency and amplitude driven, thus makes complex distortion, and unstable behavior in crossover performance.

Then there is vertical lobe formation (for vertical tweeter/midrange axis) in crossover frequency region of tweeter and midrange. This is directly related to path length between acoustic centers of two drivers and wavelength of frequencies partially passed by both drivers in crossover region. The greater the driver separation and the higher the crossover frequency, the greater the number of lobes. Focusing the sweet spot is tough. Reflections from off axis radiation are colored.

Your current crossover point of 3.5kHz is much too high. For similar driver arrangement Orion uses 1.4kHz crossover, and this is with analog 24dB/octave. With DSP, and 48dB/octave or even much higher is possible, allowing safe crossover much lower. This does get into resonance range, where natural phase behavior is problematic in analog crossover, but DSP correction can eliminate this.

TG1 is classic hard dome. Looks very nice. This tweeter is doubtless designed for faceplate flush mounted with front face of baffle. This technically applies to midrange as well. For tweeter, 32mm well cavity is created by baffle cutout, and is form of horn loading that creates resonances, and spherical diffraction, forming axis dependent ripple in frequency response. This is also form of lobe behavior. Even flush mounted tweeter has baffle size dependent diffraction behavior. Linkwitzlab\diffraction page is good reference. So real result is combination of two. Narrow baffle design is result of this behavior, as it is widely recognized as impacting imaging, and size of sweet spot. Thus Linkwitz uses narrowest baffle possible for Orion, without overworking midrange.

It looks like it may not be too difficult to move tweeter to front of baffle, at least in rough fashion for comparison measurements and listening.

Outdoor measurements look highly unpractical. Since looking for behavior above 500Hz (with moderate/low drive level this may be safely done), gated measurement may be used effectively to remove room behavior from results. From pictures, measurements at 100cm look possible for use with gating. It will be floor or wall distance that likely limits. Anyway easy to see first reflection in impulse response and work accordingly. For each measurement microphone distance should be kept to within +/- 5mm. With patience and loopback timing reference +/- 1mm is not too hard. Start measurements on axis (microphone pointing at tweeter). I would cover +/- 5 degrees from listening axis for current speaker position. I'm sure you have worked extensively with speaker toe in/out. With microphone at 100cm (this should be easy gate) 1 degree is roughly angular displacement of 17mm of microphone between measurements.

Just with tweeter loose listening to it free v with hands cupped around faceplate to form 32mm horn should be quite revealing...I imagine you've got some manner of spare tweeter about.

Samson MM01 appears to be discontinued, but looks very similar to Behringer ECM8000 and slew of similar knockoffs. Looks may be deceiving. Most of these microphones have surprisingly flat response 100Hz to 3kHz, then most show individual character, and most must have individual calibration data for peace of mind results. Highly recommend getting microphone calibrated, or buy microphone with calibration such as through Cross-Spectrum or similar. Cross-Spectrum is on this forum as "Anechoic". Here is link to a posting of some ECM8000 calibration results he has performed: ECM8000 Cal Results

At real high end are Earthworks measurement microphones. Manufacturer calibrates all microphones, and microphones released for sale are typically, really so flat that correction with calibration data is not really needed. They are not inexpensive. I own a matched pair of older Earthworks OM-1, and although not billed as measurement microphones, are fantastic as such.

I'm not certain of your posted calibration plot. Is this raw loopback measurement of soundcard for potential use as soundcard calibration? Behavior with smoothing is suspect, and plots suggest signal chain is fully DC coupled. Loopback calibration must include analog out (for tweeter midrange I assume bel-canto) looped back to microphone preamplifier (almost certainly AC coupled). This result to be valid should be exceptionally smooth (with no smoothing), and flat to +/- 0.2db or better across most of pass band.

RJ baffle, how did you ever find this? I've never seen this before. I can only imagine that big box panel resonance is present. Linkwitz gets a lot of low end from much smaller "W" baffle with two 10" drivers. Given scale of main OB, and 15" woofer, going with straight open baffle all the way down will likely have cleaner sound. Sub with built in amplifier could be much smaller sealed box for intended range. These would be my opinions only. Visually as audio geek I find the whole setup as is, very stunning.

I would very much like to see some mdat files for various measurements.

One again, you have crossed the technological hurtle of multiple channel convolution, and possibilities are near limitless.

Regards,

Andrew
 

·
Registered
Joined
·
81 Posts
Discussion Starter #11
Tank you Andrew for this great reply. I appraciate it. It is obvious that you have enormous knowledge on the subject. For me, It started just 5 years ago and I evolved rapidly from an old NAD-BW to a multiamps Linn Aktif card system. From there I realised that only DIY would be affordable for high end audio. This came with a lot of learning to catch-up.

TG1 is classic hard dome. Looks very nice. This tweeter is doubtless designed for faceplate flush mounted with front face of baffle. This technically applies to midrange as well. For tweeter, 32mm well cavity is created by baffle cutout, and is form of horn loading that creates resonances, and spherical diffraction, forming axis dependent ripple in frequency response. This is also form of lobe behavior. Even flush mounted tweeter has baffle size dependent diffraction behavior. Linkwitzlab\diffraction page is good reference. So real result is combination of two. Narrow baffle design is result of this behavior, as it is widely recognized as impacting imaging, and size of sweet spot. Thus Linkwitz uses narrowest baffle possible for Orion, without overworking midrange.

It looks like it may not be too difficult to move tweeter to front of baffle, at least in rough fashion for comparison measurements and listening.
Folowing your advice I moved TG1s in front. They where at the back before I got DSP delays options and I forgot that I could bring them in front face.
Just did it, I'll judge results with time.
Outdoor measurements look highly unpractical. Since looking for behavior above 500Hz (with moderate/low drive level this may be safely done), gated measurement may be used effectively to remove room behavior from results. From pictures, measurements at 100cm look possible for use with gating. It will be floor or wall distance that likely limits. Anyway easy to see first reflection in impulse response and work accordingly. For each measurement microphone distance should be kept to within +/- 5mm. With patience and loopback timing reference +/- 1mm is not too hard. Start measurements on axis (microphone pointing at tweeter). I would cover +/- 5 degrees from listening axis for current speaker position. I'm sure you have worked extensively with speaker toe in/out. With microphone at 100cm (this should be easy gate) 1 degree is roughly angular displacement of 17mm of microphone between measurements.

Just with tweeter loose listening to it free v with hands cupped around faceplate to form 32mm horn should be quite revealing...I imagine you've got some manner of spare tweeter about.
I'm not sure to understand your question, lost in translation again.
Your right, they are quite heavy. But that didn't spare me of trying every possible positions in the transmission area. It always come back around the same spot.

Samson MM01 appears to be discontinued, but looks very similar to Behringer ECM8000 and slew of similar knockoffs. Looks may be deceiving. Most of these microphones have surprisingly flat response 100Hz to 3kHz, then most show individual character, and most must have individual calibration data for peace of mind results. Highly recommend getting microphone calibrated, or buy microphone with calibration such as through Cross-Spectrum or similar. Cross-Spectrum is on this forum as "Anechoic". Here is link to a posting of some ECM8000 calibration results he has performed: ECM8000 Cal Results

At real high end are Earthworks measurement microphones. Manufacturer calibrates all microphones, and microphones released for sale are typically, really so flat that correction with calibration data is not really needed. They are not inexpensive. I own a matched pair of older Earthworks OM-1, and although not billed as measurement microphones, are fantastic as such..
I'v compared my mic with my friend's Beringer and they have the exact same behavior. Though I will certainly look for a better one. For this stage of development I'm ok with this one but when I'll get closer I'll need one. I found a flat response image for the Samson MM01 on the web but that's all I refered to.

I'm not certain of your posted calibration plot. Is this raw loopback measurement of soundcard for potential use as soundcard calibration? Behavior with smoothing is suspect, and plots suggest signal chain is fully DC coupled. Loopback calibration must include analog out (for tweeter midrange I assume bel-canto) looped back to microphone preamplifier (almost certainly AC coupled). This result to be valid should be exceptionally smooth (with no smoothing), and flat to +/- 0.2db or better across most of pass band.
I have used the left analog output from the Motu's DAC to the realtech PC soundcard left input for time reference loopback. This soundcard is used for the REW microphone input. The RME soundcard right channel goes for the subwoofer Motu's dac, for which I murged left and right in the RME soundcard mixer. No smoothing on the plot. Before, I used to plug/unplug analog output for every measurements but I'm taking so many readings that this could become dangerous for the Bel Canto plugs.

RJ baffle, how did you ever find this? I've never seen this before. I can only imagine that big box panel resonance is present. Linkwitz gets a lot of low end from much smaller "W" baffle with two 10" drivers. Given scale of main OB, and 15" woofer, going with straight open baffle all the way down will likely have cleaner sound. Sub with built in amplifier could be much smaller sealed box for intended range. These would be my opinions only.
This came from my "french connection" Supravox users
"Old process of assembly Supravox, this enclosure is intended for the 215 and 285. Of an effectiveness much higher than the enclosures Bass traditional Reflex, few trainage, not noise of vent and a flexible adjustment aisement according to the room of listening."
The space between cone and front face can be mooved easylly for tight bass reproduction, NO boomy bass overhere.

Visually as audio geek I find the whole setup as is, very stunning.

I would very much like to see some mdat files for various measurements.

One again, you have crossed the technological hurtle of multiple channel convolution, and possibilities are near limitless.

Regards,

Andrew
I'm glad that you have appraciated my work. I can assure you it's sound great for any acoustic EMC recording, and high resolution classical files. Old rock bands are however often badly recorded and flaws are highly underlined with this setup. I refer to natural instrument sounds for fine tuning, a base string should not sound like an electric base.

I still have some problems fixing "XO" final positions. I've moved back tweters to 3Khz, TG1 band pass begins at 2.5 Khz. But I found results to be less sharps. I feel that the 215 RTF (50hz-10Khz) are brigning more brilliance when I cut them higher around 4.5 Khz even 5.0 Khz.
This is hard evaluation based on "plausibility" of instruments. I will bring some files later on for more advices from your part.

best regards
jacques
 

·
Registered
Joined
·
239 Posts
Is comparison between MM01 and ECM8000 done with measurements? I've measured my matched pair of microphones, and though close are not exact. Plots in Anechoic's post show that no two ECM8000 are exactly the same.

I'm glad to see you are open to moving tweeters, and that it wasn't too difficult.

Soundcard loopback calibration has problems. RME feeds digital clock to both bel-canto and to Motu. Bel-canto clock and Motu clock both lock to RME clock by design. Then Motu sends analog to PC's Realtech soundcard. Realtech clock is independent, and is going to be slightly different. Realtech is junk compared to rest of system.

Mixing Java with ASIO also defeats purpose of ASIO. VAC uses ASIO4all, this is "wrapper" that is built on Windows drivers, so more problems.

Big drag with ASIO is most setups only allow single active ASIO device.

Which Motu do you have? Does it have mixer that you can use to patch microphone input to digital output, then send digital output back to RME? If this is possible, then REW uses ASIO drivers and RME as device.

Prior to REW with ASIO drivers I used other software for sweep generation, convolution, and analysis, Cool Edit Pro, and Cakewalk Sonar LE for playback and recording with ASIO. For music, I use Windows Media Player (I'm ready to get into MC17/18), or Console as VST host for Convlover (Sourceforge)

With Console, computer is dedicated and is fed digital or analog from any source. Second computer is measurement and analysis. So two channel card is used to send digital to Console machine's card thus synching clocks, and is split and looped back for digital loopback timing reference.

Regards,

Andrew
 

·
Registered
Joined
·
81 Posts
Discussion Starter #13 (Edited)
Hi Andrew

Last week modifications for tweeters brought me to do the same for woofers. The OB front face is now flat with the woofer box creating a perfect plan from floor to the upper end.

I performed many other changes, including "XO" position and slopes. According to Bob McCarthy (sound systems design and optimisation, 2rd edition 2010) the driver excurtion rise expotentially as the frequency drops.... then lower efficiency. An inaudible transition between drivers is our goal.
The large overlapping woofers/mids I used is the result of intense testing for the best rendering of sharp percussion snaps around 110hz that only the supravox 215 RTF's 100hz cut was able to performed.

By the way I highly recommand this McCarthy reference book, It is well illustrated and in an extremelly comprehensive language for a non math freak like me.

My new "XO" parameters are
sub- 20hz-60hz, (36-24 db/oct)
woof- 50hz-120hz (24-48 db/oct)
mid- 100hz-4khz (48-48 db/oct)
tweeters- 4khz-20khz (48-48 db/oct)

Your concerns about Mics efficiency is surelly justified for resl last fine tunings. I found the image where I overlayed MM01 and ECM8000 on an RTA full range reading (pict 1). One measure with one, mic change on tripod, measure with the other one and so on two times. The mic location thus as changed a little but the curves are steddy enough for each one.

Your suggestion for the motu 828 mk3 loopback setting could be possible if the REW out signal would not have to go through MC17. I can't use a 2nd computer for this loopback time ref. configuration. From REW I need a way to output sweeps to MC17 where all filters are located. Unless I'm missing something?

At the end I have joined a waterfall plot 20hz-200hz sweep, at listening position. Not sure what can be said? Mainly I don't see much variation in time, but I'm not familiar with this plot at all.
One is just subwoofer with a +5db gain at 37Hz. The other one is subwoofer+woofer
I have problem reordering pictures
The next measurements are near ones at 65cm from right side tweeter and mid at the tweeter level. The woofer measurement is at the upper part level of the 400GMF, the lower the tripod was able to get down.
The sbwoofer is a little closer to the floor.
There are IR windowed plots shows the delays used in Jriver's MC17.
 

Attachments

·
Registered
Joined
·
81 Posts
Discussion Starter #14
There were modification in previous self reply. Here I just want to add an image of sweep SPL of all speakers combined.

Dips at 60hz and 300hz are not easy to remove. The 60hz dip as always been there wathever I've tryed, "XO" location, slopes.... nothing:rolleyesno:. EQing results in heavy distortion "rigging". The 300hz dip is more easy but results in rigging too.

Job for acoustic material??

jacques
 

Attachments

·
Registered
Joined
·
81 Posts
Discussion Starter #16
Can you post mdat(s) of measurements?

Andrew
Andrew
I'v made new changes of "XO" positions. It is great to have full control of these parameters but there is a lot of work doing it. Every move means new phase matching, new delays and other fine tuning.

Now I got Mid-woofer "XO" higher at 200Hz. I'm sure it sound little to others but Supravox recommand 100Hz for open baffle.

Mic at 65cm Mid-twt at tweeter hight and woof at woofer upper part (pict)
The attachments are different of previous post. I've joined twt-mid and woofer mdats.
Is there's any special measurements you would like to see? I'll do it. It'll help me to have some of your inputs about getting things right.

regards
jacques
 

Attachments

·
Registered
Joined
·
239 Posts
What height are your ears at the preferred listening position?

And, what are heights for center of tweeter and mid above floor?

I would like to see full range 2Hz-24000kHz for tweeter and mid. No filters. Very low level of course! I do this sort of measurement with similar tweeter using terminal voltage of 0.3V, and for 4ohm driver this is only 0.023 watts.

For both measurements place microphone 21cm from baffle. For tweeter place microphone about 2cm above tweeter axis. For mid, getting 21cm from baffle is a little more difficult, you can hold something flat across driver to help find distance. For mid place microphone 3cm below axis.

In REW preferences, analysis, do not use "sub sample timing adjustment".

I would also like to have copies of your filters.

I'll see what I come up with, and then we can work with the woofer.

Regards,

Andrew
 

·
Registered
Joined
·
81 Posts
Discussion Starter #18 (Edited)
Hi Andrew

Lots of problems here since 2 days ago. I'm able to make REW's Sweep measurements for tweeters but not for mids????. In fact I was able for 2 readings with mids at 21 cm from wich I saved one.
All others gave flat responses and suprisingly normal for tweeters.
I thought it was Jriver's MC17 fault at first, but now I can't tell where the problem come from?
RTA readings are normal showing that MC17 "XO" are OK.

This MC17 setup was stable over several months.
The only thing I can tell for sure is that latelly I pluged my camera to download pictures twice and bad things came just after each time. The other change is that I created a second zone in MC17 for comparison needs between different "XO" settings. I've deleted the second one and for one time I was able to make a good sweep measurement, then the problem came back. So many possibilities!!! Virtual audio cable, asio4all, REW?


Some windows flaws?

Anyway it worked for two readings at 21 cm and I hope there suitable.
"XO" parameters are
sub- 20hz-60hz, (36-24 db/oct)
woof- 50hz-120hz (24-48 db/oct)
mid- 100hz-4khz (48-48 db/oct)
tweeters- 4khz-20khz (48-48 db/oct)



Read more: simple impulse wave question - Page 2 - Home Theater Forum and Systems - HomeTheaterShack.com

angry jacques
 

Attachments

·
Registered
Joined
·
239 Posts
Sorry that you are experiencing problems.

When you've got it sorted:

For 21cm measurements I would like to see raw driver responses. No filters, no crossover.

Sweep 2Hz to 24kHz, using no more than 0.3V signal level for mid and tweeter, this will protect tweeter.

Regards,

Andrew
 

·
Registered
Joined
·
81 Posts
Discussion Starter #20
Thank you for your moral support.

I've uninstalled/reinstalled MC17 and REW for nothing. but tonight at last, I found a problem running the Motu 828mkIII. From there, I reconfigured it and had just time to retry one sweep measurement before living and it was good.

I think the other nigt thunderstorm created some sort of statics. Tomorrow morning I'll be able to confirm this findiing.

Then I'll make your measurements. I wasn't sure that you wanted raw signals.

You know that REW limits the frequency spectrum in the measurement settings. Cannot do 2hz-24khz, try it, it will change the 24khz for 10khz automatically. Unless you have a plan?

I'm pretty scary to process those low frequencies by the tweeter.... How low it will be safe? I use -20db for regular measurements. I can reduce it to -60 and more on the dac volume. Give me an idea of how loud it should sound my ears at the Mic position, 21cm. Or any other solution.

jacques
 
1 - 20 of 45 Posts
Top