ProSoundWeb Community
Sound Reinforcement - Forums for Live Sound Professionals - Your Displayed Name Must Be Your Real Full Name To Post In The Live Sound Forums => Audio Measurement and Testing => Topic started by: Shane Ervin on October 30, 2012, 11:36:48 AM
-
This thread applies to the roll-your-own crossover / DSP / LMS situation for bi-amp wedge owners. In my case, I picked up some used 15" units from a nearby regional provider.
I'm looking for comments on measurement mic placement, and am interested in comparing the relative merits of placing the measurement mic:- where the performer's head would be, as contrasted with
- alternate arrangements whereby the meas. mic is placed on the floor / ground.
Topics which I'm hoping to see in this discussion of wedges as the D.U.T. (Device Under Test), are covered in Mark Frink's article (http://www.prosoundweb.com/article/print/monitor_mixing_tips_tools_of_the_trade_to_have_a_successful_show):- Half-Space coupling of the wedge with the floor. (160 Hz area)
- Baffle loading & directionality of LO driver (300 Hz area)
- Use of wedges in pairs (1000 Hz area)
Also, it'd be great to see comments touching on:
- Quasi-Anechoic technique of time-windowing, as it applies to a D.U.T. so close to a nearby reflecting surface, that it's touching it (namely, the floor), and
- the obvious "weirdness" of elevating a floor wedge to an "above floor" position - done merely to achieve a measurement setup in which the meas. mic is placed on the floor / ground, and/or the D.U.T. is far enough away from a nearby surface such that quasi-anechoic time-windowing makes clearer sense.
So far, I've only conducted T.F. measurements by placing the wedge in the "as-used" position on the floor, and the meas. mic was at performer's head height, aiming down at the grill.
Propagation delay found, based on the HF driver (muting the LO), and used for T.F. measurements, I ran another impulse measurement on the LO to obtain an initial delay value to employ in the MiniDrive LO band.
I proceeded to apply gains, and some filtering, in both bands, to achieve a magnitude response picture suitable for identifying the acoustic crossover point.
Moving to the goal of achieving phase alignment through the crossover region, I captured a new T.F. trace of the (now) EQ'd HF driver, and proceeded to dial-in / fine-tune the LO delay by lining up the phase traces at f=acoustic x-over.
A 0.417 ms delay applied to the LO did the trick.
- Electronic X-Over: f = 1.14 kHz (L-R 24)
- Acoustic X-Over: f = 1.4 kHz
With both bands un-muted, it was very satisfying to sanity-check matters by reversing the polarity on one band, and observe a deep valley in the magnitude response at the crossover point.
P.S. Kudos to Rational Acoustics; I'm enjoying the demo version of Smaart v7.4. (... 3 cheers for Object-Oriented Programming!)
-
You might want to try aligning @ the x-over via all-pass filters .
I would use as little delay as possible on a stage monitor or no delay at all if my processor has all-pass 1st and 2nd order filters .
But a 0,4 something delay would not be a problem so if it works fine......
-
Hi Timo,
That delay was just time-aligning the woofer to the horn, since the acoustic centres are not lined up physically in these boxes. I'm with you, of course, in avoiding the use of overall delay in floor wedges. Indeed, since the horn is not delayed electronically, I wouldn't say that delay is applied, per se, to the wedge.
Back to the thread topic, what are your thoughts on the orientation of the D.U.T. and placement of meas. mic? Are there are compelling reasons to go to the extra trouble of elevating the speaker, say, onto a pedestal and conducting T.F. measurements with the mic on the ground?
In the context of comparing measurement methods (and setups) with the use of an anechoic chamber, my interest is in an exploration of the merits of including the half-space loading effects in the T.F. measurement, since, after all, that's how the wedge is used in practice. This, in preference - perhaps - to full anechoic testing. Building on this idea, could it be that the optimum test method for floor wedge T.F. measurements is to use an anechoic chamber equipped with a conventional (i.e., not mesh) floor?
-
If there are all-pass filters in the processor you do not need to use delay to align the low driver to the mid highs . Via all-pass 1st or 2nd order just delay the frequencies that need delaying and leave the rest untouched .
For mic placement i would go for the fast and easy by just putting the mic roughly where the listener would be or @ the middle point of both drivers really close and then verify at the listeners position if the alignment holds .
Before doing the alignment of the drivers may it be delay or allpass i would take out some of the peaks and maybe some heavy dips (if any are there) on the drivers via eq .
-
Back to the thread topic, what are your thoughts on the orientation of the D.U.T. and placement of meas. mic? Are there are compelling reasons to go to the extra trouble of elevating the speaker, say, onto a pedestal and conducting T.F. measurements with the mic on the ground?
In the context of comparing measurement methods (and setups) with the use of an anechoic chamber, my interest is in an exploration of the merits of including the half-space loading effects in the T.F. measurement, since, after all, that's how the wedge is used in practice. This, in preference - perhaps - to full anechoic testing. Building on this idea, could it be that the optimum test method for floor wedge T.F. measurements is to use an anechoic chamber equipped with a conventional (i.e., not mesh) floor?
The response of a wedge changes radically (more than +/- 3 dB) from a placement on the floor compared to in free space, unless you also use the wedges on poles, an anechoic response of a wedge is not of much use other than above the frequency of pattern control for the HF horn.
Even on a pole, the floor bounce lacking in an anechoic environment will be a real part of the sound heard, though it will change with distance.
A useful test position would be the wedge on the ground, outdoors, mic approximating normal ear placement, wedge at least 20 feet from any walls or large objects.
If your wedges need shimming (Angular Discrepancy Adjustment Modules, AKA 2"x4", audio logs, etc.) to get to a proper listening angle (like 12AMs need) they should be tested both flat and with the shim, as the baffle to floor distance affects response considerably.
This will give you a "large hall" approximation, as boundaries in a large hall are far enough away so won't affect the monitors all that much, while on small stages you are often EQing as much for the room as the wedge.
As well as room problems, the performer's face, mouth, eyeglasses, and hat all have huge effects on the eq needed to make a wedge work properly, a flat mid-band response is only a starting point, but is a helpful reference.
Art Welter
-
This is an interesting side discussion, and since you've twice raised the notion of using all-pass filters as an alternative to the application of delay in driver time-alignment, I thought I'd pick up on it.
If there are all-pass filters in the processor you do not need to use delay to align the low driver to the mid highs. Via all-pass 1st or 2nd order, just delay the frequencies that need delaying and leave the rest untouched.
Rather than it being a case in which "you do not need to use delay" as a means to align drivers, is not delay considered to be the method preferred over a phase shift introduced via an all-pass filter? That is, at least as the first remedy one would apply prior to contemplating a supplemental phase shift filter to treat the phase relationship at frequencies on either side of the acoustic crossover?
This (older) article (http://sound.westhost.com/pcmm.htm) provides a compelling case that delay is the superior method to correct for physical misalignment of the drivers' acoustic centres.
And here's an old thread (http://zmarchive.com/psw-srf/flat/0035/th0035438-all-pass-filters.html) featuring some of PSW LABsters from years gone by, debating the same topic.
I'll add two more arguments in favour of delay (in preference over a phase shift filter) as a means to achieve driver alignment:
- Not all DSP crossovers provide all-pass filters, whereas most, if not all, will offer a band o/p delay parameter. Case in point: BSS MiniDrive.
- In DSP crossovers where both techniques are available, it takes fewer DSP resources to implement a short time delay, than it does to implement a filter. (i.e., Z-1, and no MAC's).
(minor edit)
-
For the application at hand, aligning the signal arrival of the woofer with a horn that is physically offset behind the woofer, delay is the proper tool to use. Given enough all pass filters it is possible to get an equivalent delay within the pass band of the woofer. However, IMHO, this is the long way to get to the solution. Using delay for this application has absolutely no adverse side effects.
The signal from the horn is arriving 0.417 ms after the signal from the woofer. Delay the woofer by this amount so that their arrivals are synchronized. Only after doing this will it be possible to correctly determine the LP and HP filters which will provide complimentary magnitude & phase response for the LF & HF devices through the crossover region to yield the desired system response.
As to the measurement setup, I'd do exactly as a Art suggested. Place the wedge in a position just as it will be used, but outdoors as far away from any boundaries or objects as possible. Place the mic close to ear height and make some measurements. If multiple mics are available, place them in different positions, but at the same height. This should yield some info about how the response of the floor wedge changes at off-axis positions.
Hope this helps.
-
Hi Art, Charlie, and Timo,
Excellent discussion going here; thanks for your posts!
As I mentioned in the kickoff post, the interesting aspect of a floor wedge as a D.U.T. is the consideration given to placement of the meas. mic placed elsewhere than on the ground - a bit unsettling at first. Then you start to think of the underlying rationale (no pun intended) for the test activity, and you can reasonably justify the departure from standard practice.
It would be a good side discussion to cover saving, in a DSP preset, notch filters for use with certain specific vocal mics, or rooms, or performers. I did that at a gig in late Sept. and left the EQ's at home, knowing I had adequate GBF with a tried and tested custom preset.
Our side discussion on all-pass phase shift filters, as compared with straightforward application of delay is also welcome. I still own a BSS FDS-360 with flush-mounted phase-adjustment pots. Remember those old forum posts on the "poor man's technique" for driver alignment? Using a sine tone at the crossover frequency and polarity-reversing cable, you'd adjust for the deepest null - out front - then remove the pol-rev cable. IIRC, BSS described the technique in their product manual, or other literature.
Fun story to relate: One day, years ago at the sound company warehouse, when phase-aligning for a sidefill's 18" to the 10" mid at 250 Hz, I did an ad-hoc demo of beam-steering! We could make loose objects rattle on their shelves in different locations around the warehouse - just by action on the phase-adjustment pot. Ah, the good old days of swept sine testing!
So time rolls on and thanks to advances in technology, and we now have access - at low cost - to so much more in the realm of test gear, and techniques. It's been 21 years since I attended the Vancouver chapter AES meeting where Audio Precision demo'd their (then new) "Porta-One", conducting a quasi-anechoic T.F. measurement on a small speaker. Then came MLSSA from DRA Labs, and others with TDS techniques.
And in all that time, I'd never stopped to consider floor wedge monitors as a special case for consideration w.r.t. test setup. We'll have to add the internet in that list of measurement-related resources we now enjoy.
-
Interesting discussion. I, too, have had this question as to the most relevant way to measure floor monitors.
The measure-it-how-you-use-it approach has logical appeal even though it runs counter to the widely held belief that the frequency response of the first sound to arrive is of the greatest perceptual importance. But in this case the reflection off the floor is so early that it may well be perceptually more part of the the speaker than part of the room.
In my limited experiments, I ended up basing my settings on a pseudo-anechoic measurement (speaker in the air, windowed impulse response), figuring if the uncompensated hole in the response caused by the floor reflection was objectionable, it could be EQ'd in use.
So far musicians have liked the sound, and I haven't had any feedback problems, but then, I haven't been in any very demanding situations either.
I'm very interested to learn from you folks who've done this more than I have.
--Frank
-
The measure-it-how-you-use-it approach has logical appeal even though it runs counter to the widely held belief that the frequency response of the first sound to arrive is of the greatest perceptual importance. But in this case the reflection off the floor is so early that it may well be perceptually more part of the the speaker than part of the room.
I am not aware of this widely held belief. I am aware of a widely held belief that says the apparent source will be the first arrival assuming certain other conditions, like near equal level and a limited time differential.
In my limited experiments, I ended up basing my settings on a pseudo-anechoic measurement (speaker in the air, windowed impulse response), figuring if the uncompensated hole in the response caused by the floor reflection was objectionable, it could be EQ'd in use.
How do you eq the hole caused by reflection back in? All you can do is be aware of it or change the setup so it doesn't happen. You cannot correct a reflection caused suckout with eq.
Mac
-
I am not aware of this widely held belief. I am aware of a widely held belief that says the apparent source will be the first arrival assuming certain other conditions, like near equal level and a limited time differential.
How do you eq the hole caused by reflection back in? All you can do is be aware of it or change the setup so it doesn't happen. You cannot correct a reflection caused suckout with eq.
Mac
+100 This is VERY IMPORTANT!!!!!!!!!!!!!!!
This is a fact that many people seem to want to "overlook". Sure, you can "appear" to fix the suckout in one place-but what you REALLY do is to make it worse every where else.
And to expand a bit, while it may "appear" that you can "fix" a hole caused by a "suckout" or reflection, IN REALITY you are NOT fixing it. You are ONLY fixing the display on the computer screen. If you were to use a finer resolution on the screen (ie no smoothing), you would see that the notch is still there. But with a smoother resolution, it will "appear" to be fixed-because of the averaging of the display.
Think of it this way-let's say you have a notch at 2000 Hz at a particular location. The reflection is causing it. The loudspeaker does not have this notch when away from the boundary. So it appears as a dip-so you boost 2Khz 6dB.
BUT WHAT DOES THAT REALLY DO? It boosts the direct signal 6dB AND it ALSO boosts the reflection 6dB. So the net result between the two is EXACTLY the same. SO you have done NOTHING! However on a display with smoothing, it "appears" that the hole is gone-because the AVERAGE around 2Khz is higher than the rest of the response.
But the hole is still there. All you have really done is put a 2KHz boost in all other listening positions-which can actually make fighting feedback more of a problem.
You just can't blindly "whack away" with an eq-thinking you are fixing the "issue".
-
I am not aware of this widely held belief. I am aware of a widely held belief that says the apparent source will be the first arrival assuming certain other conditions, like near equal level and a limited time differential.
How do you eq the hole caused by reflection back in? All you can do is be aware of it or change the setup so it doesn't happen. You cannot correct a reflection caused suckout with eq.
Mac
Mac,
Thanks for the reply. I'm obviously babbling above my pay grade here but I'm trying to learn this stuff.
On your first point, of course you are correct with respect to the time scale under consideration here. However, is it not generally accepted that sound arriving later than the time that corresponds to a comb-filter interval smaller than the critical bandwidth is perceived more as spacial than tonal? [McCarthy, Sound Systems: Design & Optimization, pp.164] For example, if we assume a critical bandwidth of 1/6 octave and a frequency of interest of 1kHz, then sound arriving at the listener more than 8ms after the first arrival would not have a large impact on tonality.
On your second point, I agree that you can't fix a suckout with EQ. But this actually supports the idea of basing the settings on an anechoic measurement since there is nothing you can do about a suckout you might see when you put the monitor on the floor and the mic at the listening position. Whether the EQ is in the speaker processor or somewhere else, it is still EQ, and a losing battle. There might be some value in mitigating floor, or other room, effects with EQ to some extent. All I was saying is that since I'm not smart enough to do that in advance I'll leave it for ad hoc EQ in the field.
--Frank
-
For example, if we assume a critical bandwidth of 1/6 octave and a frequency of interest of 1kHz, then sound arriving at the listener more than 8ms after the first arrival would not have a large impact on tonality.
Why would only be interested in 1Khz when talking about the "tonality" of a loudspeaker. There are a lot more freq that should be looked at.
In your example, the 8ms 2nd arrival would produce the first notch at 63Hz and further notches that are spaced at 125Hz apart. So the second notch would be @188hz, the 3rd notch would be at 313Hz etc.
So I would argue that notches at those freq WOULD affect the tonality.
Now when you get to the higher freq, the spacing of 125Hz is going to produce so many notches, that they will tend to "mesh together" in our brain. But at the lower freq-they would still be very much audible.
-
In your example, the 8ms 2nd arrival would produce the first notch at 63Hz and further notches that are spaced at 125Hz apart. So the second notch would be @188hz, the 3rd notch would be at 313Hz etc.
So I would argue that notches at those freq WOULD affect the tonality.
Now when you get to the higher freq, the spacing of 125Hz is going to produce so many notches, that they will tend to "mesh together" in our brain. But at the lower freq-they would still be very much audible.
I agree. I just gave the 1kHz example for clarity (apparently it didn't work). I was not claiming that an equal-level addition of an 8ms copy would not affect tone at lower frequencies.
To get back to the original problem, I see three approaches:
1. Base the settings on an anechoic measurement, averaging over the required range of angles.
2. Base the settings on a "real floor" measurement being careful to ignore uncorrectable artifacts.
3. Build 20dB of boost at 300Hz into the speaker settings in a pathetic attempt to correct for the existence of the floor. Well, maybe not such a good idea.
Signing off for tonight.
--Frank
-
I agree. I just gave the 1kHz example for clarity (apparently it didn't work). I was not claiming that an equal-level addition of an 8ms copy would not affect tone at lower frequencies.
To get back to the original problem, I see three approaches:
1. Base the settings on an anechoic measurement, averaging over the required range of angles.
2. Base the settings on a "real floor" measurement being careful to ignore uncorrectable artifacts.
3. Build 20dB of boost at 300Hz into the speaker settings in a pathetic attempt to correct for the existence of the floor. Well, maybe not such a good idea.
Signing off for tonight.
--Frank
4: Use a monitor that couples well with the floor and doesn't have much of a "bounce" off of the floor.
-
Shane-
Have you taken readings from other biamp wedges, hopefully from a known manufacturer?? We can overlay measurements A. v. B.
I appreciate the quest for hard copy on relevant parameters, but my first test on your fresh purchases would be:
Do they all sound the same? If not, why not? Match them.
Get them to sound good in the environment they will be used in: On floor (half space sorta)..
Get them to sound good with a mic & vocal - on axis. (58 or better).
Then measure.
What we all desire is:
Consistency across power response (1W/1M & 100W/1m).
Smoothest phase response.
Least dramatic impulse response.
Minimal group delay.
Consistent polar response. Esp @ crossover.
Now you are having fun!
Monitors function in an acoustic toilet for the most part. A consistent and solid tool, even if it lacks "ideal" parameters for high fidelity, can be a winner.
Keep us posted.
-
We're expecting some warmer weather in the next few days. The pics show the area where I can do some outdoor measurement work. (You'd almost think the previous owners of this place wanted to test speakers in their backyard like I do!).
In reply to a few questions / comments so far in this thread:
- JBL loaded, mirror-image wedge pair
- NOT a CD horn. (Another side discussion might be in order, just on merits / drawbacks of CD horns in wedges)
- horns are oriented vertically; great rejection of "adjacent guy's mix" when deployed straight along front line for two musos. Obviously best when mirror image horns are "out"
- ~ 20 years doing duty on festival stages; met rider for various acts
- I bought the last two left in inventory; only 2 to which I have access to measure
- relative to an equal (i.e., normalized) horn gain setting and same amp/NL-4 cable used, one wedge's woofer is 1 dB less sensitive (w magnitude resp cursor on each unit's trace - themselves very similar)
- more detailed match-oriented Smaart measurements remain TBD (e.g., horn / overall match)
- easy to make sound good without Impulse Response / T.F. / measurement equipment. In their first deployment (under my stewardship), I had them in a rehearsal studio for 6 months as a left/right pair for a hard rock guitarist / vocalist in Montreal. I created a MiniDrive preset by ear, the day I delivered them to their room
- later, still before I got the chance to measure them, I used them, one each for two lead singers in the front line. (w those parametric notches for use with my 57, as described in a previous post)
Happy Remembrance Day
Happy Memorial Day
Lest We Forget
-
- horns are oriented vertically; great rejection of "adjacent guy's mix" when deployed straight along front line for two musos.
This is another common misconception in audio. Let's say the horn is 90x40 and is 12" in the 40° dimension (and that is outside of thorn flare-NOT the outside of the flange).
This would give pattern control down to around 2Kz. If the horn is smaller-then that freq would be higher. Of course if the horn had a wider coverage pattern (say 60°), then that number would go down a bit.
So below 2K or so-the pattern starts to get wide. The front loaded woofer has essentially no pattern control-so what is spilling over into the other musicians area is a lot of low and mid "mud".
Only the upper part of the highs is controlled by the horn. The lower part of the highs still coming out of the horn are still spilling into others areas.
The only way to get pattern control down lower is to use a larger horn or a wider horn.
For the same physical size-a wider horn will have pattern control down lower.
-
Only the upper part of the highs is controlled by the horn. The lower part of the highs still coming out of the horn are still spilling into others areas.
+1 Merely assumed that would be understood, but yes, it's best to remind folks, and avoid feeding any misconception.
-
+1 Merely assumed that would be understood, but yes, it's best to remind folks, and avoid feeding any misconception.
And the same thing goes for ALL speaker cabinets.
When a manufacturer states a particular coverage pattern-it is NOT across the entire operating bandwidth (DESPITE what some manufacturers suggest------).
The physical size of the horn is what determines the lower limit of the control. And having a "rotatable" horn that is small-is an excersize in futility. But it does make the user "feel better" about what they "think" they are doing-while in reality they are only affecting the top octave or two only.
Pattern flip is something that most are unaware of-or simply choose to ignore-even though it IS happening.
Take look at the attachment. This is the manufacturers data on the SAME loudspeaker. Notice how the coverage is very different between 630=1800 & 8Kz.
What IS the correct spec this cabinet should have? It depends on what freq you look at-but most all the others are very different.
ACTUAL Loudspeaker patterns is something most users fail to consider. Yet they talk about it all time as if they do. :(
They take a "simple number" and just "assume" that it is overall true of the device in question.
When you start looking at the "whole picture", it gets a bit more blurry and much more complicated.
-
Take look at the attachment. This is the manufacturers data on the SAME loudspeaker. Notice how the coverage is very different between 630=1800 & 8Kz.
What IS the correct spec this cabinet should have? It depends on what freq you look at-but most all the others are very different.
Ivan, since the graphs are too low res to read, what do they represent? It is a nice illustration that in some way, something is changing vs frequency, but what are we looking at?
Are those polar plots of vertical coverage? Horizontal coverage? An illustration of pattern shape looking down the throat of the horn? For someone who rightly rails for clarity of information, this is sadly lacking.
Mac
-
Ivan-
Regarding 12" giving pattern control down to 2khz: What multiple/submultiple of frequency length can we use as a target range for horn pattern control? 1/4 wavelength is the baseline assumption for horn loading.Yes?
-
Ivan, since the graphs are too low res to read, what do they represent? It is a nice illustration that in some way, something is changing vs frequency, but what are we looking at?
Are those polar plots of vertical coverage? Horizontal coverage? An illustration of pattern shape looking down the throat of the horn? For someone who rightly rails for clarity of information, this is sadly lacking.
Mac
Yes it is a little fuzzy. I tried to get all the graphs on the same page. Sorry.
The angles are in 10° increments with the outer edge being 80°.
The "view" is looking out the front of the cabinet-not the horn per say-but could kinda be thought of that way.. So the coverage pattern that the "cabinet would project". So horizontal and vertical at the same time.
This was gotten from their EASE data.
-
Ivan-
Regarding 12" giving pattern control down to 2khz: What multiple/submultiple of frequency length can we use as a target range for horn pattern control? 1/4 wavelength is the baseline assumption for horn loading.Yes?
The "target" should be as low as possible. But if only the HF has a horn, then the horn should be at least large enough to provide pattern control down to (or preferably below- by 1/2 octave or so) the ACOUSTICAL crossover of the horn. The electrical crossover is often a good bit higher than the acoustical crossover (for a HF device and some midrange devices)
Horn loading and horn pattern control are very different things.
Horn loading is limited on the upper end by the size of the opening that the driver is "entering" the horn at. So the larger the driver exit -the less horn loading will be happening up high.
The horn pattern control extends all the way up (assuming the driver itself has a wide enough pattern at the exit to "fill" the horn). The limit is the low end-which is a combination of physical size and pattern. If one number stays the same-and the other number gets larger-then the freq to which it exhibits pattern control will go lower. And if either gets smaller-then the freq will go higher.
A "quick and dirty" formula to remember for horn pattern control is 1,000,000/ (pattern coverage angle x Horn size in inches for the intended vertical or horizontal pattern of interest). Who says you can't multiply apples and walnuts?
I'm not sure where your 1/4 wavelength comes into play in the question. There are lots of areas that 1/4 wavelength is of interest, but pattern control and horn loading are generally included in that.
Generally 1/4 wavelength is considered a good goal for distance between drivers to couple. That would be within a single horn or seperate cabinets-doesn't matter. Be it HF devices or subs. The problem on the HF side is as you go higher and higher, that 1/4 wavelength gets very small.
-
[quote.
A "quick and dirty" formula to remember for horn pattern control is 1,000,000/ (pattern coverage angle x Horn size in inches for the intended vertical or horizontal pattern of interest). [/quote]
I found the source of my confusion! When I hear "horn size" I think "path length". Whey you say "horn size" you are talking about "mouth area".
-
A "quick and dirty" formula to remember for horn pattern control is 1,000,000/ (pattern coverage angle x Horn size in inches for the intended vertical or horizontal pattern of interest).
I found the source of my confusion! When I hear "horn size" I think "path length". Whey you say "horn size" you are talking about "mouth area".
The path length will vary depending on the pattern of the horn.
As the pattern gets narrower-the path gets longer (to have the same freq of pattern control)-as the size of the mouth also gets larger.
-
A "quick and dirty" formula [...]
Just out of interest, what is the slow and clean formula?
Or is it just very rule-of-thumbish, and is really one of the questions that "is easily answered by a simple-easy to understand WRONG answer!"
-
Shane- Keep us posted.
Weather was nice a couple of days ago, and I was able to get these traces.
- Left Wedge - Overall + overlaid horn trace
- Right Wedge - Overall + overlaid horn trace
Unplugged at the DUT end (same amp, X-Over channel, etc., in use).
High-pass at 100 Hz in MiniDrive preset.
-
Just out of interest, what is the slow and clean formula?
Or is it just very rule-of-thumbish, and is really one of the questions that "is easily answered by a simple-easy to understand WRONG answer!"
The "correct" way is to MEASURE the product and then examine the results.
The formula that I gave is Don Keele's and gives a rought idea of the pattern control freq.
-
Just out of interest, what is the slow and clean formula?
Or is it just very rule-of-thumbish, and is really one of the questions that "is easily answered by a simple-easy to understand WRONG answer!"
Steve,
There's a slow and clean formula, its called the Kirchoff-Helmholtz integral :)
Keele's original CD horn paper "What's so sacred about the exponential horn" is a useful place to start on this topic.
Also, Keele's followup paper "Optimum horn mouth size" is an interesting read, even if makes the incorrect assumption of plane wave propagation.
The narrowing and subsequent expansion of the coverage pattern of a horn near the low cutoff ultimately results from diffraction effects.
-
There's a slow and clean formula, its called the Kirchoff-Helmholtz integral :)
I felt smarter just knowing that!
But then I googled it, and now I don't :-[
I will attempt to partly absorb it when I am more awake, but not at 2am (Syd, Aust)
-
Here is another example of a horn that has a "spec", but yet the actual data is not even close to it. Yes it is a real product-not a simulation.
The spec is accurate for the 8Kz response, but look at some of the lower freq-they are WAY WAY off.
How would you array this speaker? As if you were using the 8KHz coverage or the 4Kz coverage. I would argue that both are to high for "normal intelligibility". So if you tried to array "per spec" you would be way off. My guess is that around 6Khz the pattern is pretty much omni.
The numbers under the freq are the coverage of the outer band-so the actual coverage is twice where the band is shown.
It is a bit hard to see-but the scale changes on the different graphs-so it is actually WORSE than it appears (as if the scale was the same)
You REALLY need to look a bit closer to get the real "picture" of what is happening.
-
Just out of interest, what is the slow and clean formula?
This might be of interest, http://www.excelsior-audio.com/Publications/Understanding_Horn_Directivity.pdf (http://www.excelsior-audio.com/Publications/Understanding_Horn_Directivity.pdf).
-
This might be of interest, http://www.excelsior-audio.com/Publications/Understanding_Horn_Directivity.pdf (http://www.excelsior-audio.com/Publications/Understanding_Horn_Directivity.pdf).
Nice paper!
And it "should" be readily apparent that you CANNOT simply have a rotatable horn-that has different coverage patterns-yet the dimensions are the same.
The end result is NOT going to be what the user "thinks" it is.
However that does not stop manufacturers from making them and users from buying them-and then "spouting of that you can "rotate the horn on XYZ speaker" to get the coverage pattern you (think) you need to "array" them.
Oh how blindly people follow along------without actually verifying to testing to see if the product actually comes close to doing what the specs say.
Manufacturers wouldn't lie now would they??????????????????????????????????????
-
Manufacturers wouldn't lie now would they??????????????????????????????????????
That reminds me of the saying "all fishermen are liars except me and you... and I'm a little concerned about you." ;)
-
And it "should" be readily apparent that you CANNOT simply have a rotatable horn-that has different coverage patterns-yet the dimensions are the same.
Ivan's excellent point about rotatable horns is worthy of a +1
-
This might be of interest, http://www.excelsior-audio.com/Publications/Understanding_Horn_Directivity.pdf (http://www.excelsior-audio.com/Publications/Understanding_Horn_Directivity.pdf).
FWIW,
Also visible in charlie's papers' directivity balloons are the first order diffraction lobes off the main coverage axis.
-
Just started reading this thread after I did some measurements of some of my wedges today :(.
I'm interested in if the phase response shown here is a result of incorrect measurements taken or if they are "ok" for what they are.
3 speakers were measured.
EV SxA250
Peavey PV 15pm
Yamaha MS400
Not a lot of filters were needed to keep them in a +/- 3db range.
Looking at them the Yamaha looks like the horn is not in correct polarity to the woofer but there is no major drop in response at the crossover point. There was an acoustic adjustment device used to raise the front up so I wonder if that is part of the dips in phase and response in the woofer range?
The peavey phase seems to be about what I'd expect to see.
The EV I'm unsure of as to if this is normal for this speaker?
Both the Peavey and EV have a peak around 200hz. I took this to floor / mic placement issue although I didn't see it in coh.
Coh was fine for this measurement and I was in a good sized room. I used the setup like in the Mic/speaker picture.
Thanks for any input on these. Sorry for the cell phone pictures..
Douglas R. Allen
-
Just started reading this thread after I did some measurements of some of my wedges today :(.
I'm interested in if the phase response shown here is a result of incorrect measurements taken or if they are "ok" for what they are.
3 speakers were measured.
EV SxA250
Peavey PV 15pm
Yamaha MS400
Not a lot of filters were needed to keep them in a +/- 3db range.
Looking at them the Yamaha looks like the horn is not in correct polarity to the woofer but there is no major drop in response at the crossover point. There was an acoustic adjustment device used to raise the front up so I wonder if that is part of the dips in phase and response in the woofer range?
The peavey phase seems to be about what I'd expect to see.
The EV I'm unsure of as to if this is normal for this speaker?
Both the Peavey and EV have a peak around 200hz. I took this to floor / mic placement issue although I didn't see it in coh.
Coh was fine for this measurement and I was in a good sized room. I used the setup like in the Mic/speaker picture.
Thanks for any input on these. Sorry for the cell phone pictures..
Douglas R. Allen
Douglas,
There really are no "incorrect measurements", but interpretation of measurements can be difficult.
The proximity of the monitor's baffle to the floor will affect FR and phase response somewhat, "audio logs" , "acoustic adjustment devices" or Angular Discrepancy Adjustment Modules (ADAM) all will get the HF pointing towards your ears, but change response due to differences in diffraction and woofer wrap/bounce.
The PV 15pm has the least phase wrap of the three monitors you tested, about 130 degree from 250 Hz up, the Yamaha MS 400 a bit more at around 145 degree, the EV the most at around 270 degrees.
Considering they all use active electronics, none of the phase responses look great, but none look "incorrect".
Below is a screen shot of three speakers with flatter phase response, a Mackie 1502HD, an active 2 way using Gunness focusing, and a DSL SH100, a passive 8" coaxial two way, and a Welter Systems 8H, a passive 8"& horn floor monitor.
The 1502HD and the SH100 were measured on a 6 foot pole, the WS8H, measured as in your picture.
I'd bet if you took another set of measurements with the PV, Yamaha & EV you could not reproduce the phase charts. Each of these monitors is designed with the HF horn placed horizontally adjacent to the LF driver and will have a different response depending on what is chosen as "on axis", and the response off axis to the left will be different than to the right.
The 1502HD, SH100 and WS8H are all quite consistent in response from left to right.
The Yamaha MS 400 manual claims a 30 dB per octave crossover at 1600 Hz, the phase looks consistent in that range, what makes you think the "horn" is not in correct polarity to the woofer?
Art
-
Douglas,
Using the arrow key to move the phase trace to center of the display will give you a better look at the phase trace.
Instead of having to think through the confusion of the wraps, moving the trace to a spot where there are no wraps lets you do more accurate math.
PT
-
Thanks Art and Paul;
The phase of the yamaha in the horn range is of course at the very top or in the + area for most of the horns bandpass area. Where as the woofer section is all in the -side of the screen or bottom for the most part. The phase trace seems to be linear for the most part as well in the horn range. Not going behind in phase as the frequency drops as in the EV speaker.
I wondered if I reversed the polarity would this phase move to the bottom or remain the same just with a problem in the crossover area now. ( I feel this would happen but not sure ) I've measured the yamaha speakers from time to time through the years and they always more or less looked like that.
When I'm building a 2 or 3 way system I'll shoot to have the phase angles to lay on top of each other at the crossover point (s) using delay and using different filter types and slopes to get them to match up. When unmuted they more or less look like the Peavey monitors phase.
Looking at the EV or Yamaha it seems like there should be a problem at the crossover point. Yet there is not. I'm guessing had EV or Yamaha used an all pass filter this could be corrected some what?
A larger part of me thinks that the phase is lagging in time more as the frequency goes lower and it just happens to, for the loss of the correct words, wrap around in the crossover frequency general area with the view I had selected. This seems to be easier to see in the EV trace but not so much with the Yamaha one. Thanks for the tip Paul on changing the view. I'll try this next time
To sum it up I'm trying to get a better handle reading measurements like these as they come up.
The way I see it now the EV and Yamaha just happen to have not as good phase response as the peavey though out thier frequency range but wanted to see if there is another way to look at it.
Thanks for any input;
Douglas R. Allen
-
Thanks Art and Paul;
The phase of the yamaha in the horn range is of course at the very top or in the + area for most of the horns bandpass area. Where as the woofer section is all in the -side of the screen or bottom for the most part. The phase trace seems to be linear for the most part as well in the horn range. Not going behind in phase as the frequency drops as in the EV speaker.
I wondered if I reversed the polarity would this phase move to the bottom or remain the same just with a problem in the crossover area now. ( I feel this would happen but not sure ) I've measured the yamaha speakers from time to time through the years and they always more or less looked like that.
Douglas,
Paul's tip is helpful, I have reposted the Mackie 1502HD, DSL SH100, and WS8H so you can see the phase response a bit better.
You can see that it is typical for all ported speakers phase to invert towards the port frequency.
It is also typical for phase to generally trend down from low to high, the Peavey is a bit odd in the "up and down" trace.
If you reverse the polarity of a loudspeaker, the phase trace will still look the same, but will be rotated 180 degrees.
The phase traces you posted all look to be OK in the crossover region, but all use different filters to achieve those results.
The phase traces of the speakers I posted all look a bit better, and again, each used very different filters and transducers- there are many different ways to skin the phase cat.
Art