Thanks Jack
Watch, read and learn! I might try to get into HR again. I agree, that loading is second and 1/4WL is first priority.
Also I have begun to think, that there maybe are different ways/requirements to design 1/4 WL for crossover!? Also the volume under the cone to control the low-pass frequency of the bandpass is maybe a consideration (i.e. "natural roll-off").
A: Mark with his 96 dB/okt FIR-filters needs some overlap for linearization befor FIR-filtering. So maybe he needs less volume under the cone to reach higher?
B: Chris´s approach in his K402MEH (as I have understood) with PEQ in IIR-DSP, zeroth-order-filters, utilizes the natural roll-off of the drivers in the horn, which might require less overlap? Here maybe more volume under the cone is needed to pull down the roll-off-frequency?
Does what I write make sense?
Steffen
Watch, read and learn! I might try to get into HR again. I agree, that loading is second and 1/4WL is first priority.
Also I have begun to think, that there maybe are different ways/requirements to design 1/4 WL for crossover!? Also the volume under the cone to control the low-pass frequency of the bandpass is maybe a consideration (i.e. "natural roll-off").
A: Mark with his 96 dB/okt FIR-filters needs some overlap for linearization befor FIR-filtering. So maybe he needs less volume under the cone to reach higher?
B: Chris´s approach in his K402MEH (as I have understood) with PEQ in IIR-DSP, zeroth-order-filters, utilizes the natural roll-off of the drivers in the horn, which might require less overlap? Here maybe more volume under the cone is needed to pull down the roll-off-frequency?
Does what I write make sense?
Steffen
Last edited:
Hi Steffen,
A big part of the reason i use 96 dB/oct filters is because they require much less overlap for summing bands together properly.
Remember we are talking net acoustic overlap (after electrical xovers are added).
Let's say you combine two drivers that both have shallow first order acoustic rollofs at 1000Hz, with no electrical xover.
The lower driver would need to have near perfect first order rollof all the way up to about 4kHz.
The upper driver would need near perfect low end rolloff down to about 260Hz.
And this is just to have good summation of the drivers through the range of contribution to the -12dB level .
For the same window of good summation to -12dB contributions ....
96 dB/oct LR's need:
flat response of the lower driver extending up to only 1070Hz.
and flat response of the upper driver down to 940Hz.
Those narrow overlaps are a heck of a lot easier than 260Hz and 4kHz, huh?
A big part of the reason i use 96 dB/oct filters is because they require much less overlap for summing bands together properly.
Remember we are talking net acoustic overlap (after electrical xovers are added).
Let's say you combine two drivers that both have shallow first order acoustic rollofs at 1000Hz, with no electrical xover.
The lower driver would need to have near perfect first order rollof all the way up to about 4kHz.
The upper driver would need near perfect low end rolloff down to about 260Hz.
And this is just to have good summation of the drivers through the range of contribution to the -12dB level .
For the same window of good summation to -12dB contributions ....
96 dB/oct LR's need:
flat response of the lower driver extending up to only 1070Hz.
and flat response of the upper driver down to 940Hz.
Those narrow overlaps are a heck of a lot easier than 260Hz and 4kHz, huh?
I sometimes think things, and if they are stupid ideas then I'm prepared to look a bit stupid but..........
Can we not model the woofers rear port in hornresp by modelling the driver 'in reverse' ie: model the driver at the (rear) ports location as if that is the front port. Use the port dimensions as the front port, the cabinet volume as the front chamber volume, and the space between the cone and the (higher frequency port) as the rear volume, and the (higher frequency port) as the 'rear port'
Hope that makes sense, only just thought of it so haven't tried it yet.
Cheers,
Rob.
Can we not model the woofers rear port in hornresp by modelling the driver 'in reverse' ie: model the driver at the (rear) ports location as if that is the front port. Use the port dimensions as the front port, the cabinet volume as the front chamber volume, and the space between the cone and the (higher frequency port) as the rear volume, and the (higher frequency port) as the 'rear port'
Hope that makes sense, only just thought of it so haven't tried it yet.
Cheers,
Rob.
Hi M,
Interesting you said that, because that is what i did to the target woofer ports' distance to throat i wanted in the latest build.
Knowing i wanted the ports as close to horn walls as possible, i chose the horizontal angle that was tangent to the woofers being shoved into the horn as far as possible. IOW, a narrower horn moves the ports further from the throat (since the woofers have to fit in the top and bottom cavities)
It's how i chose a 75x60 vs a 80x60....
Of course, that ignores the "loading" aspect and was only concerning with 1/4 spacing. But it appears the loading aspect is trivial for anything other than a supersized build.
Kinda glad to be done with woofer loading concerns....
I sometimes think things, and if they are stupid ideas then I'm prepared to look a bit stupid but..........
Can we not model the woofers rear port in hornresp by modelling the driver 'in reverse' ie: model the driver at the (rear) ports location as if that is the front port. Use the port dimensions as the front port, the cabinet volume as the front chamber volume, and the space between the cone and the (higher frequency port) as the rear volume, and the (higher frequency port) as the 'rear port'
Hope that makes sense, only just thought of it so haven't tried it yet.
Cheers,
Rob.
Hi Rob, i haven't been able to figure out what you mean, and are trying to do...
Just to make sure i'm not starting with the wrong foot, by 'woofers rear port', do you mean bass reflex ports? (That's what i've thought so far.)
Hi mark100,
Regarding the italicized portion of your quote, maybe my geometry is lacking, but it seems to me that since the distance is constant, the drivers must be at the same distance from the throat. What am I missing?
Or are you talking about the entry-to-entry distance between the different drivers? I attempted to solve it using a round horn, which allows for (relative, depending on the size of drivers) freedom regarding the number of drivers.
Kindest regards,
M
Regarding the italicized portion of your quote, maybe my geometry is lacking, but it seems to me that since the distance is constant, the drivers must be at the same distance from the throat. What am I missing?
Or are you talking about the entry-to-entry distance between the different drivers? I attempted to solve it using a round horn, which allows for (relative, depending on the size of drivers) freedom regarding the number of drivers.
Kindest regards,
M
Attachments
Hi M, sorry for a lack of clarity. What i meant was particular to my method of mounting the 12"s on the top and bottom, where it has to fit in between the side flares.
Like in this pict.
When i'm trying to keep its ports as close to the side walls as possible, a narrower horiz pattern pushes the cone and hence ports out further from the throat.
When I use a single round port centered in the horn, it obviously doesn't matter then.
I really need to figure out an easy way to make vertical polar measurements.
The single round port centered in the horn doesn't mess with horizontal polars any more than ports towards the corners as far as i can tell.
But i figure it has to hurt vertical more. Must test, and find out. If verticals don't die and go to the minors, to heck with woofer corner ports !
The big syns without a box are very awkward to rotate, to rest on a side for measuring verticals.
But their saving grace is weight. This 75x60, with a 33.5" x 26.5" primary mouth weighs just 62 lbs using the 8 lb B&C CD.
Still needs secondary flares, and boxing in the small mids. I'm estimating 70 lbs total.
Your round horn project looks really interesting. Have you got a version of it up and running yet?
How will it transition to adding larger drivers?
Like in this pict.
When i'm trying to keep its ports as close to the side walls as possible, a narrower horiz pattern pushes the cone and hence ports out further from the throat.
When I use a single round port centered in the horn, it obviously doesn't matter then.
I really need to figure out an easy way to make vertical polar measurements.
The single round port centered in the horn doesn't mess with horizontal polars any more than ports towards the corners as far as i can tell.
But i figure it has to hurt vertical more. Must test, and find out. If verticals don't die and go to the minors, to heck with woofer corner ports !
The big syns without a box are very awkward to rotate, to rest on a side for measuring verticals.
But their saving grace is weight. This 75x60, with a 33.5" x 26.5" primary mouth weighs just 62 lbs using the 8 lb B&C CD.
Still needs secondary flares, and boxing in the small mids. I'm estimating 70 lbs total.
Your round horn project looks really interesting. Have you got a version of it up and running yet?
How will it transition to adding larger drivers?
Hi mark100,
thank you for the explanation, it makes sense. The round horn does not have the same problem, but, of course, it has its own tradeoffs.
My project stalled several times for various reasons, but I have now finished the horn mouth, and I am waiting for a measurement microphone to arrive, as I had lost an access to a reference microphone I had used before.
Kindest regards,
M
P.S. It is little crude, but for evaluation purposes . . .
M
thank you for the explanation, it makes sense. The round horn does not have the same problem, but, of course, it has its own tradeoffs.
My project stalled several times for various reasons, but I have now finished the horn mouth, and I am waiting for a measurement microphone to arrive, as I had lost an access to a reference microphone I had used before.
Kindest regards,
M
P.S. It is little crude, but for evaluation purposes . . .
M
Attachments
Hi M, your horn looks great....pls post how it sounds when you get there.
OK...progress report.
Been trying many things...
First step was to make a better outdoor measurement rig.
The rotating boom hung from off a deck overhang that i've been using for years, has always had high susceptibility to wind. And i've never been real happy about house/deck reflections still being picked up.
New location uses a fixed boom going off the side of the deck that reduces house reflections, and greatly reduced wind's effects.
You can make out the mic in top left corner.
I'm very lucky to live on a beautiful lake, but you would not believe the noise.
Sound really caries over water, and between boats, jet skis, waves on the rocks, lawn movers, leaf blowers, cicadas, well, ...SPL meter always reads 80+ dB noise.
Remarkably, time averaged measurements with Smaart using pink noise are barely effected. A few seconds and not very loud, and good data is captured.
I didn't fully appreciate how well time averaging worked with the old test rig.
Wind really messes up measurements with even the slightest mic movements. That was obvious, but i also thought all the lake noise was messing up the measurements too.
With the new, more stable mic rig, I can now see how little the noise messes up the measurements. (and it's easily identifiable as a lack of coherence)
I'm elaborating all this to provide encouragement to any contemplating out-door measurements in noisy environments. Time averaging pink works.
Averaging multiple sine sweeps works too, but it's probably hard to get away with many siren swoops just about anywhere. Pink is inoffensive.
If anybody knows a lower cost dual channel measurement program than Smaart, which can time average pink (or music, etc), please post it.
Anyway, back to syn 9 with adding small mids....
I compared four 4fe35's to 4mdn34's on the same horn.
The 4fe35's had a little better extension both high and low, and overall a little smoother response. The only reason to go for the 4mdn34's is for increased SPL imo.
I've been using these mids from 250Hz up to the CD. With either the bms 4594HE or B&C dcx464, I'm happy with any xover frequency between 650-950Hz.
Currently using 950Hz, because i get a little better polars there, (and it sounds awesome)
Speaking of polars...with the better measurement rig, i decided to try to thoroughly revisit steep vs shallow xovers. Comparisons of 96, 48, 24, and 12dB/oct linear-phase LRs were made at frequencies from 650-950Hz.
Polars got progressively worse as order decreased. So heck, I went the other way and tried steeper, elliptical and brick wall. Brick wall is now the xover between CD and mid.
That got me all excited to try even steeper for low-to-mid xover at 250Hz 12" lows to 4" mids). No go, polars diminished some.
To investigate, I made polars of each band independently.
Found a curious thing with the 12" lows. Using lin-phase LR's of various orders, all of course had the same -6dB level at 250Hz on-axis. At 30 degrees H off-axis, a LR4 had 4dB less response drop at 250Hz than a LR16.
(LR8 was about halfway in between).
No clue why, but shallow worked better, even for a single band!!! ... more homework to do.
One final piece to add to this rambling post....measurements are failing bigtime to describe the sonic improvements of adding the small mids.
The little bit of change one set of polars vs the other may be having doesn't appear to account for much.
The change in clarity continues to be dramatic compared to the sound of single larger cones crossed straight to the CDs at lower frequency.
Oh, i lied...one final, final piece. Enlarged the single round, horn centered, 12" ports on the 60x60 horn from 3" dia to 3.5" dia. No more chuffing, 8.5:1 sd to port area. Tight, well defined sound.
kappalite 3012lf, faital 12pr320, and rcf mb12n351 have all been tried.
Raw small signal response is dang close to identical for all three drivers. Evidently the horn, and way the drivers are enclosed, pretty strongly dominates over which driver.
OK...progress report.
Been trying many things...
First step was to make a better outdoor measurement rig.
The rotating boom hung from off a deck overhang that i've been using for years, has always had high susceptibility to wind. And i've never been real happy about house/deck reflections still being picked up.
New location uses a fixed boom going off the side of the deck that reduces house reflections, and greatly reduced wind's effects.
You can make out the mic in top left corner.
I'm very lucky to live on a beautiful lake, but you would not believe the noise.
Sound really caries over water, and between boats, jet skis, waves on the rocks, lawn movers, leaf blowers, cicadas, well, ...SPL meter always reads 80+ dB noise.
Remarkably, time averaged measurements with Smaart using pink noise are barely effected. A few seconds and not very loud, and good data is captured.
I didn't fully appreciate how well time averaging worked with the old test rig.
Wind really messes up measurements with even the slightest mic movements. That was obvious, but i also thought all the lake noise was messing up the measurements too.
With the new, more stable mic rig, I can now see how little the noise messes up the measurements. (and it's easily identifiable as a lack of coherence)
I'm elaborating all this to provide encouragement to any contemplating out-door measurements in noisy environments. Time averaging pink works.
Averaging multiple sine sweeps works too, but it's probably hard to get away with many siren swoops just about anywhere. Pink is inoffensive.
If anybody knows a lower cost dual channel measurement program than Smaart, which can time average pink (or music, etc), please post it.
Anyway, back to syn 9 with adding small mids....
I compared four 4fe35's to 4mdn34's on the same horn.
The 4fe35's had a little better extension both high and low, and overall a little smoother response. The only reason to go for the 4mdn34's is for increased SPL imo.
I've been using these mids from 250Hz up to the CD. With either the bms 4594HE or B&C dcx464, I'm happy with any xover frequency between 650-950Hz.
Currently using 950Hz, because i get a little better polars there, (and it sounds awesome)
Speaking of polars...with the better measurement rig, i decided to try to thoroughly revisit steep vs shallow xovers. Comparisons of 96, 48, 24, and 12dB/oct linear-phase LRs were made at frequencies from 650-950Hz.
Polars got progressively worse as order decreased. So heck, I went the other way and tried steeper, elliptical and brick wall. Brick wall is now the xover between CD and mid.
That got me all excited to try even steeper for low-to-mid xover at 250Hz 12" lows to 4" mids). No go, polars diminished some.
To investigate, I made polars of each band independently.
Found a curious thing with the 12" lows. Using lin-phase LR's of various orders, all of course had the same -6dB level at 250Hz on-axis. At 30 degrees H off-axis, a LR4 had 4dB less response drop at 250Hz than a LR16.
(LR8 was about halfway in between).
No clue why, but shallow worked better, even for a single band!!! ... more homework to do.
One final piece to add to this rambling post....measurements are failing bigtime to describe the sonic improvements of adding the small mids.
The little bit of change one set of polars vs the other may be having doesn't appear to account for much.
The change in clarity continues to be dramatic compared to the sound of single larger cones crossed straight to the CDs at lower frequency.
Oh, i lied...one final, final piece. Enlarged the single round, horn centered, 12" ports on the 60x60 horn from 3" dia to 3.5" dia. No more chuffing, 8.5:1 sd to port area. Tight, well defined sound.
kappalite 3012lf, faital 12pr320, and rcf mb12n351 have all been tried.
Raw small signal response is dang close to identical for all three drivers. Evidently the horn, and way the drivers are enclosed, pretty strongly dominates over which driver.
Last edited:
Hi mark100,
Thank you. Well, my microphone arrived, I am re-learning how to use HOLM.
Do you think that we could discuss your experience with more details, probably via either p.m. or e-mail because, see attached image. (The railing taken off by snow is back on ).
Kindest regards,
M
Thank you. Well, my microphone arrived, I am re-learning how to use HOLM.
Do you think that we could discuss your experience with more details, probably via either p.m. or e-mail because, see attached image. (The railing taken off by snow is back on
).Kindest regards,
M
Attachments
Unfamiliar with/unsure what may be special about Smaarts time averaging of pink noise but won't a lot measurement suites do this? I don't see it in Holm or REW but:
ARTA will let you generate/record pink noise of a set sample length and then perform an FFT, smoothed or not, on whatever slice of time you like from that recording.
As far as lower cost goes, free Speaker Workshop will let you generate/record pink noise of any duration you wish and then perform an FFT on whatever time slice you like from that recording. It will ALSO let you use a wav file as your signal so you could use music although I don't remember seeing anyone doing that. I always wondered if that would be useful for speaker development. I still think Speaker Workshop was/is underappreciated but I digress...
Other than those won't the likes of Audacity, Audition and even the old Cooledit do this?
Thx Kevmoso, sorry i wasn't clearer about the averaging...
The transfer function averaging in Smaart allows settings from none to infinite.
The averaging is of all the discrete transfer functions made during the time setting.
I'm using 2 - 4 second settings, which are giving surprisingly stable repeatable traces in my (very) noisy environment.
If ARTA allows taking a few seconds of pink noise recording, and then slicing the recording into equal time pieces that each get an FFT applied, and then those FFTs get averaged....i think that would be the same thing as Smaart.
I can't see how that can be done yet, but i'm pretty new to ARTA still.
Speaker Workshop looks pretty cool. I looked for a manual (without downloading the program) to see what it can do. From your description, it sounds like it handles things similar to ARTA with regard to analyzing a recording.
I think being able to average a number of discrete pink noise transfer functions together in real time is the goal. I don't know how many discrete transfers per second Smaart makes. I think it's 24 frames per second.
Like mentioned earlier, REW allows multiple sweeps to be averaged (up to 8), and does the same thing kinda. I just can't get away with all those sweeps outdoors. Pink noise is almost unobtrusive.
(Smaart also can use real time music as the test signal, but i've found pink to be less of a nuisance than music that gets turned on and off.
Hmmm...I guess i need to just leave the music playing all the time, and no one would even know i'm making measurements.)
The transfer function averaging in Smaart allows settings from none to infinite.
The averaging is of all the discrete transfer functions made during the time setting.
I'm using 2 - 4 second settings, which are giving surprisingly stable repeatable traces in my (very) noisy environment.
If ARTA allows taking a few seconds of pink noise recording, and then slicing the recording into equal time pieces that each get an FFT applied, and then those FFTs get averaged....i think that would be the same thing as Smaart.
I can't see how that can be done yet, but i'm pretty new to ARTA still.
Speaker Workshop looks pretty cool. I looked for a manual (without downloading the program) to see what it can do. From your description, it sounds like it handles things similar to ARTA with regard to analyzing a recording.
I think being able to average a number of discrete pink noise transfer functions together in real time is the goal. I don't know how many discrete transfers per second Smaart makes. I think it's 24 frames per second.
Like mentioned earlier, REW allows multiple sweeps to be averaged (up to 8), and does the same thing kinda. I just can't get away with all those sweeps outdoors. Pink noise is almost unobtrusive.
(Smaart also can use real time music as the test signal, but i've found pink to be less of a nuisance than music that gets turned on and off.
Hmmm...I guess i need to just leave the music playing all the time, and no one would even know i'm making measurements.)
They do work similarly and is a bit tedious compared to how it sounds Smaart does this in real time. At $1000 USD for Smaart I'll just bug the neighbors with a few hundred sine sweeps to capture a whole project. Lol
I don't really recommend Workshop for much these days as REW for the same price has become a much better software.
I don't really recommend Workshop for much these days as REW for the same price has become a much better software.
Yeah, i agree. Even for the lesser priced Di2 version of Smaart, that's a whole lot of cost compared to donateware REW and lower cost ARTA.
Savings pays for a bunch of drinks for disgruntled neighbors
If i hadn't owned Smaart for so many years, i'd be hard pressed to spring for it today, being just a DIY hobbyist.
I do think though, that it would be awfully helpful for folks if real-time transfer averaging can be found in any more affordable software.
Savings pays for a bunch of drinks for disgruntled neighbors
If i hadn't owned Smaart for so many years, i'd be hard pressed to spring for it today, being just a DIY hobbyist.
I do think though, that it would be awfully helpful for folks if real-time transfer averaging can be found in any more affordable software.
Do you mean averaging measurements? ARTA can do that with pink periodic noise, with sweeps, with MLS or with external excitation.
Attachments
Watching a video about Smaarts measurements I can say that is not what he is doing with Smaart.
It is also not the same thing I was referring to in ARTA. I was using the signal time record, ctrl+T to capture pink and then gating the recording but that doesn't appear to offer phase data.
Interestingly Speaker Workshop does produce phase with basically the same method.
This will be something I explore a bit more when I have more time.
Smaart is looking really nice for its ability to show this data real time. $1k argh....
It is also not the same thing I was referring to in ARTA. I was using the signal time record, ctrl+T to capture pink and then gating the recording but that doesn't appear to offer phase data.
Interestingly Speaker Workshop does produce phase with basically the same method.
This will be something I explore a bit more when I have more time.
Smaart is looking really nice for its ability to show this data real time. $1k argh....
Thanks Mark, yep, I've seen ARTA is very capable at making averaged impulse measurements.
The particular type averaging I'm finding extra useful, is real-time transfer function averaging.
With a mic in a steady position, Smaart runs transfer functions repeatedly, and for the averaging period specified, continually displays the ongoing moving average.
So random noise events fall out through averaging, and the mag and phase traces stay continually on the screen (changing only as the real time average changes).
Also, (something i haven't mentioned yet), for moving mic averages Smaart can automatically readjust its delay finder (loopback timing), so that each transfer function captured is with excess delay (TOF) removed (24 transfers per second). Iow, phase is also correct for averages, with moving mic .
It's kinda amazing how noise resistant it all works. And how relatively low level the pink noise needs to be.
"And now to something completely different"
I still have to learn to use REW and a lot of other things, but I have been thinking about how to measure my synergy´s, when time comes. I do often read about measurement-noise in respect to neighbors (Mark refers to that issue) and reflections from buildings/walls.
I have been contemplating to have the Synergy placed, so that the loudspeaker fires up in the sky. And then construct some kind of a movable/rotating frame (rotating around the throat of the horn) to hold the microphone for polar-measurement.
This arrangement should prevent most neighbor-complaints.
Would it be a god idea in respect to reflections also? Given that I place the measurement-stand some 25-30 feet away from the house.
Well I thought this was the time and place to ask.
best regards
Steffen
I still have to learn to use REW and a lot of other things, but I have been thinking about how to measure my synergy´s, when time comes. I do often read about measurement-noise in respect to neighbors (Mark refers to that issue) and reflections from buildings/walls.
I have been contemplating to have the Synergy placed, so that the loudspeaker fires up in the sky. And then construct some kind of a movable/rotating frame (rotating around the throat of the horn) to hold the microphone for polar-measurement.
This arrangement should prevent most neighbor-complaints.
Would it be a god idea in respect to reflections also? Given that I place the measurement-stand some 25-30 feet away from the house.
Well I thought this was the time and place to ask.
best regards
Steffen
Last edited: