Zilch Posted August 31, 2009 Author Report Share Posted August 31, 2009 I believe that there is an issue with the gating of the signal and the window being used. Just an untested theory.I would argue that continuous pink-noise RTA shows the same result, though it is FFT.It is interesting that his cyan curve here looks to have a reasonably flat passband, why did this one end up passinga sanity check, how was the measurement different? Is it repeatable?Cyan is KLH-17, which has essentially no bass worthy of note other than by those who imagine it to be there, apparently. Read the brochure posted in this site's library.... Link to comment Share on other sites More sharing options...
Pete B Posted August 31, 2009 Report Share Posted August 31, 2009 I would argue that continuous pink-noise RTA shows the same result, though it is FFT.Cyan is KLH-17, which has essentially no bass worthy of note other than by those who imagine it to be there, apparently. Read the brochure posted in this site's library.... It is about -3dB at 70Hz with a flat passband above 100 Hz as one would expectfor a sealed system. the general shape looks reasonable is all that I'm saying,the smallish box probably gives it a high Fc. The question is why does this one follow the theory, assuming Fc is around 70 Hz while the others do not? Link to comment Share on other sites More sharing options...
speaker dave Posted August 31, 2009 Report Share Posted August 31, 2009 I can certainly do a quick draw-away study, and also measure with a Behringer RTA here, but there have been many instances such as the KLH-17 shown in Post #11 where the curves have departed significantly from this pattern.You showed 1/3rd octave curves. Were those done with the Behringer or with the FFT box in a 1/3 octave mode? If you get the same curve with two different measurement "boxes" then you can exclude the instruments or even gating effects, right? If it is the same instrument displaying with different resolutions then the instrument could still be an issue.Pete and I have had this debate previously with respect to Large Advents and whether they already have BSC built in or not, and then, as here, I cited the SPICE transfer function of the lowpass filter. Tell me how it is possible for a woofer that does not have significantly rising response to measure flat once the illustrated lowpass filter is applied.If the combined system is flat and the crossover sags, then, yes, the woofer must have rising response. Most woofers do have a rising response and need the network to flatten them out as well as create the crossover corner.Otherwise, what are we thinking, microphone proximity effect?Proximity effect only pertains to Cardioid mikes as far as I know. I assume you are using some kind of instrumentation omni? Furthermore if a proximity bass boost were happening you would expect the apparent cuttoffs to shift down rather than up, relative to the impedance derived fc. That is whats odd here.Does the response peak necessarily occur at Fc in closed-box alignments?For high Q systems then the peak is at fc by definition. Actually it assymptoticly (?) approaches fc after starting a little higher for Q's just above 0.7. The better way to think of it is to extend the 12dB/Octave final rolloff slope up and draw a second, horizontal, line from the midband level down. Those two lines cross at Fc, nomatter what the Q. Then, incidently, the dB level at that frequency re. the midband level (see caveats below) equals 20 log Q.I note that Atkinson characterizes the Smaller Advent as "slightly underdamped." With a nearfield peak of 5 dB? Further, while Fc = 46 Hz, the response peak occurs at ~80 Hz. I'm not sure we're talking the same language here, and it's the reason I asked if you were correcting to anechoic response before applying the concept of "Q." It all makes sense if that is done. I'm going to have to go back and read Small again, apparently, as well....Atkinson isn't being inconsistent. Understand that there is a distinction between corner bump and bass level. By that, I mean if you were to use a lot of crossover inductance and pull the upper range down so that the overall bass range is 5 dB proud of mid and treble level, this isn't the same as a high Q bump at resonance. Thats why I was talking about curve fitting to the corner only. Again, T/S theory only fully applies to the simple case of a woofer with its upper cuttoff well removed from its lower corner, minimal inductance rise near resonance, no heavy crossover effect, 2pi, not 4pi, etc. Add any of those issues in and you need a more complex model.Another way to state it is that adding broad EQ, even if it raises bass level, isn't the same as changing Q.David Link to comment Share on other sites More sharing options...
Zilch Posted August 31, 2009 Author Report Share Posted August 31, 2009 Here's KLH-17 measured 10 months ago. It is subsequently re-measured another four or five times in that thread, months apart. There is no lowpass filter in the product:http://www.audiokarma.org/forums/showthrea...848#post2275848And here's two JBLs and a Goldwood in the same box, measured the same way:http://www.audiokarma.org/forums/showthrea...366#post2214366********CLIO's RTA is a subset under FFT, and I measured at its highest display resolution, 1/6 octave, using 1/24 octave measurement resolution. CLIO measurements are made using the CLIO mic, the longest one, whatever that model is.I just measured driver #4 in the study AR3a, no lowpass filter, using the Behringer DEQ2496 RTA, which is also FFT, 1/6 octave, using its ECM8000 mic, the orange bar display, below. I am providing the CLIO sinusoidal measurements of woofer #4 from the second study presented here for comparison. The grey curve is without the lowpass filter: Link to comment Share on other sites More sharing options...
Zilch Posted August 31, 2009 Author Report Share Posted August 31, 2009 Compare also to draw-away, 0.25" - 48" (black); right-click to open second image above and first one here in separate tabs, then click between them: Link to comment Share on other sites More sharing options...
Zilch Posted August 31, 2009 Author Report Share Posted August 31, 2009 For high Q systems then the peak is at fc by definition. Actually it assymptoticly (?) approaches fc after starting a little higher for Q's just above 0.7. The better way to think of it is to extend the 12dB/Octave final rolloff slope up and draw a second, horizontal, line from the midband level down. Those two lines cross at Fc, nomatter what the Q. Then, incidently, the dB level at that frequency re. the midband level (see caveats below) equals 20 log Q.O.K., reverse that: draw a vertical line at the mean Fc of the unfiltered response which is 43.57 Hz, and, as I read it, it intersects at ~92 dB. Calling that the midband level, we have ~5 dB of bump.With the filter in place, Fc and the peak occur at lower frequencies, and though the peak is also lower SPL, the bump is more like 8 dB. You're saying that the Q should be constant, but I believe Ken suggested that the filter could shift it.... http://www.classicspeakerpages.net/IP.Boar...-1251613951.jpg Link to comment Share on other sites More sharing options...
kkantor Posted August 31, 2009 Report Share Posted August 31, 2009 O.K., reverse that: draw a vertical line at the mean Fc of the unfiltered response which is 43.57 Hz, and, as I read it, it intersects at ~92 dB. Calling that the midband level, we have ~5 dB of bump.With the filter in place, Fc and the peak occur at lower frequencies, and though the peak is also lower SPL, the bump is more like 8 dB. You're saying that the Q should be constant, but I believe Ken suggested that the filter could shift it.... http://www.classicspeakerpages.net/IP.Boar...-1251613951.jpgWhat I meant to say was that the crossover can alter the curve shape greatly, and the effective -3dB point, (or -6dB or ...), and that very little of this comes from an alteration of the Fc or Qtc. I agree with David that one must differentiate between EQ effects and resonance effects. All this is academic anyway, since bass is about moving air. It's easy to build a 4.5" woofer box with a bass cutoff of 20Hz, if that's what matters. At the bottom of their range, AS woofers tend to be as close as you can get to a reliably model-able transducer. They almost always work just as expected. But, that doesn't mean all that much in the real world, where factors like the room, 60 Hz to 120 Hz performance, etc, are really the dominant issues. At any rate, I have some simple sealed boxes around... I can easily characterize one in ways I understand well, and drop it by for you to test. That would help me establish a reference point for further discussion. Also, I can use a calibrated mic, so that variable can be eliminated.-k(As an aside: if anyone knew what "correct" bass response was, it would take about 10 minutes to achieve it with today's DSP, using any number of enclosure types. Unfortunately, it's not a trivial issue, as anyone who has tried to come up with a "room correction system" quickly learns. There is a great tendency in audio to apply the most rigorous possible techniques in quest of utterly ill-defined goals. In other words, "what would an ideal woofer measure like?" Anyone who thinks that answer can be written in a sentence or a paragraph or 10 pages, hasn't really thought it through. In my un-humble opinion....) Link to comment Share on other sites More sharing options...
Rlowe Posted August 31, 2009 Report Share Posted August 31, 2009 Does this help?Another curve provided by Tom Tyson but is no longer accessible in the Forum URL;<< 4pi anechoic chamber curve of AR-3 woofer system >>previously provided by Tom Tyson, Tom's accompanying text read as follows;tysontom Thu Sep-01-05 01:45 PMCharter member545 posts#6983, "RE: Anyone Have AR-3a Measurements, System, Drivers"In response to Reply # 19>>Almost forgot to comment on this plot below, this is in>between, less peaking at 450 Hz than Tom's plot, but flatter>than the one above. It says that the woofer was measured out>doors and all drivers were into 360 degrees? Can anyone>confirm how the woofer was measured, half space or full>space?>" Pete,I forgot to comment on your observations above regarding the measurement solid-angle. AR initially measured all woofers in afree-field, 2-Pi (half-space) environment. This was done by literally burying the speaker in the ground, face-up, flush with thesurface, facing into a 180-degree solid angle. The ideal echo-free environment is outdoors, of course, but it is very inconvenient,and therefore indoor anechoic chambers became a necessity for most measurements. The problem with nearly all commercialanechoic chambers, of course, is that they are not large enough to be “anechoic” at the lowest audio frequencies.Chamber large enough to measure into the deep bass frequencies are huge and enormously expensive, and exist only at suchplaces as Harvard’s Acoustic Laboratory and those belonging to aerospace and automotive companies.AR’s large anechoic chamber (one of AR anechoic chambers was transported in its entirely out to NHT’s Benicia, California siteback in the 1990s) was echo-free above approximately 200 Hz, so AR developed a calibration curve that compared the outdoor 2-Pi woofer curve to the woofer’s curve in the chamber, and thus a meaningful anechoic response down to 30 Hz was possible. Thetweeters were measured in the anechoic chamber facing into a 4-Pi, 360-degree “spherical” environment. Before the calibrationcurves was used, AR took the outdoor, 2-Pi woofer curve and spliced it to anechoic curves made above 200 Hz in the chamber.A 4-Pi low-bass woofer measurement is impractical, of course, since the absolute output of a speaker system below the frequencyof ultimate radiation resistance is influenced by the solid angle of the space into which it radiates. 4-Pi woofer measurement atlow frequencies will be down by 6 dB (halving the solid angle doubles the energy radiated at low frequencies) compared to 2-Pimeasurements. This is the reason for the bass-shy characteristics of most speakers if they are mounted out in the middle of theroom up on a stand, or listened to out doors. Another way of looking at it is that if a speaker is designed to be “flat” down into thelow frequencies when measured into 360 degrees, it will therefore sound intolerably bass-heavy when mounted in the usual 2-Pifashion in a typical listening room. This was the very reason for the AR-10Pi Environmental control -- to enable flexibility inmounting the speaker in various positions to offset the affect of the solid angle on bass frequencies. " Link to comment Share on other sites More sharing options...
speaker dave Posted August 31, 2009 Report Share Posted August 31, 2009 CLIO's RTA is a subset under FFT, and I measured at its highest display resolution, 1/6 octave, using 1/24 octave measurement resolution. CLIO measurements are made using the CLIO mic, the longest one, whatever that model is.I just measured driver #4 in the study AR3a, no lowpass filter, using the Behringer DEQ2496 RTA, which is also FFT, 1/6 octave, using its ECM8000 mic, the orange bar display, below. I am providing the CLIO sinusoidal measurements of woofer #4 from the second study presented here for comparison. The grey curve is without the lowpass filter:First off, I apologize if I'm dragging this off on an academic tangent. Hopefully someone out there is interested and learning a bit.It looks to me like your Clio measurement (either resolution) and your Behringer measurement are the same. Both show significant rolloff around 60Hz and the same midrange characteristics.I'm very comfortable with your second set of draw away curves. The plot is now 50dB tall which is closer to standard. As you get a little distance away the system looks fairly flat from 100 to 800Hz. I'm thinking the sag above 200 is inductance and network, and designed in to make the far field flat, i.e. the power response rolls down to balance out rising d.i.I've attached a sketch over your curves to clarify what I meant about getting Q from a graph. The only tricky part is defining the mid band level. I think that if all inductance (woofer and network) were removed that the curve would continue to rise to about 98 on your graph. As shown you draw a line up the 12db/8ve slope, another down from the mid level, and see the intersection at about 59 Hz. If the level there is -3.5dB then Q is about 0.67. So the system is slightly overdamped. The bump at 150 isn't a "Q bump" at resonance, it is the mound left after the 300Hz and up area is pulled down.Now, if a Q of .67 at 59 Hz doesn't agree with your impedance curve Q mesurements, then that is a discrepency! My experience is that T/S measurements are hard to get within, say 20%, when measured in different ways. Others may do better than that but that is my experience. It appears to me that all your curves imply higher resonances that your T/S numbers. I think Ken has it right to suggest you measure a known system that he provides. Its always good, if you suspect your measurements, to measure something where you know the answer.A second image from the web is what I was refering to as a "Universal resonance curve", showing a 2nd order highpass with a variety of Q's at 3.3 Hz (nice sub!) Note that the Q of 1 curve is exactly 0dB at 3.3 Hz but has the expected 1dB rise at a higher frequency.Ditto with Ken about inductance shifting resonance. I was thinking in terms of the simple 2nd order highpass only, where level at resonance directly equates to Q. Big network inductance not only shifts the apparent corner, it also interacts with the equivalent circuit of the woofer and pushes electrical resonance down (as your T/S numbers show).Regards,David Link to comment Share on other sites More sharing options...
speaker dave Posted August 31, 2009 Report Share Posted August 31, 2009 Another curve provided by Tom Tyson but is no longer accessible in the Forum URL;<< 4pi anechoic chamber curve of AR-3 woofer system >>previously provided by Tom Tyson,Interesting curve Rlowe, thanks for providing it.I think this is without network, so it shows the 2pi to 4pi droop and also additional midrange rise that the crossover network would deal with (assuming the end result is flat in 2pi as all other AR curves show).Regards,David Link to comment Share on other sites More sharing options...
Pete B Posted August 31, 2009 Report Share Posted August 31, 2009 Here's KLH-17 measured 10 months ago. It is subsequently re-measured another four or five times in that thread, months apart. There is no lowpass filter in the product:http://www.audiokarma.org/forums/showthrea...848#post2275848And here's two JBLs and a Goldwood in the same box, measured the same way:http://www.audiokarma.org/forums/showthrea...366#post2214366********CLIO's RTA is a subset under FFT, and I measured at its highest display resolution, 1/6 octave, using 1/24 octave measurement resolution. CLIO measurements are made using the CLIO mic, the longest one, whatever that model is.I just measured driver #4 in the study AR3a, no lowpass filter, using the Behringer DEQ2496 RTA, which is also FFT, 1/6 octave, using its ECM8000 mic, the orange bar display, below. I am providing the CLIO sinusoidal measurements of woofer #4 from the second study presented here for comparison. The grey curve is without the lowpass filter:One thing to keep in mind, that we discussed here on this forum many years ago, is that the series DC resistance of the AR woofer XO inductor (about .9 ohms) plays a significant roll insetting the Qtc of the system. Driving the woofer directly provides a Qtc of about .5 (6dB down at Fc)to .6, as compared to .7 (3dB down at Fc) to .8 with the inductor. This is due to the fact that the .9 ohms is a significant percentage of the DC resistance of the driver. Often the inductorDCR can be ignored if it is not significant as compared to the DCR of the VC. A resistor is requiredwhen driving the woofer directly if you want to confirm a 2nd order rolloff with a Qtc of about .7.The DC resistance of the woofer L does alter the system Q and the amplitude at Fc is highlysensitive to Q. The inductance of the L and VC tilts the passband. We don't want to keep theinductance to confirm the measurement, but we do want to keep the DC resistance.Other things to keep in mind is that the near field measurement is not accurate above 200 Hz.Also, the curve from AR spans 3 decades on the FR axis, yours mostly only 1 decade. It canbe interesting to see how just changing the axis can make things look "right".Your amplitude scale is also expanded, might want to try 10 dB per major division or use moredivisions.Interesting that you're using an AR-3a since Allison's paper in the library under Other>Technical Paperis also showing an AR-3a woofer only through the crossover under 2pi, 4pi, back against a singleboundary, etc. This provides another set of curves, and one could obtain a correction curve for2pi to 4pi by recording the difference. Note that those curves are PWL on the amplitude scale, thedB values should be double to convert to SPL. Note that the mid XO frequency is mentioned as575 Hz which might indicate that it is an earlier 3a with the #7 (1.9 mH IIRC) woofer inductor.Another sanity check would be with the system facing up outdoors, say on a driveway with themic duplicating the distance in Allison's paper. There is a curve for this condition in that paper. Link to comment Share on other sites More sharing options...
Zilch Posted August 31, 2009 Author Report Share Posted August 31, 2009 First off, I apologize if I'm dragging this off on an academic tangent. Hopefully someone out there is interested and learning a bit.It's just not a problem; I am in no respect jealous of my findings.We each work toward comfort with using the specific "kit" of tools and methodologies available to us, and I believe it's clear there's considerable rigor in my approach. I'm always pleased to find others getting the same or similar results, but no less so to discover discrepancies such are are apparent here, with the fundamental interest of learning more in the course of resolving them.I'll swap samples with Ken and see what we can find out. This is FUN stuff! Link to comment Share on other sites More sharing options...
kkantor Posted August 31, 2009 Report Share Posted August 31, 2009 First off, I apologize if I'm dragging this off on an academic tangent. Hopefully someone out there is interested and learning a bit.Just for the record, my "academic" comment was not in any way directed at you, (or anyone in particular, for that matter). I think the rigorous approach is great, and I'm learning and re-learning some things in this thread. My "academic" comment was meant to remind us all not to assign absolute figures of merit or rankings to enclosure designs on the basis of data that is highly disconnected from real-world product performance. IOW- understanding the woofer's isolated frequency response is foundational, but it is only the first step in delivering good bass response to the customer. -k Link to comment Share on other sites More sharing options...
kkantor Posted August 31, 2009 Report Share Posted August 31, 2009 It's just not a problem; I am in no respect jealous of my findings.We each work toward comfort with using the specific "kit" of tools and methodologies available to us, and I believe it's clear there's considerable rigor in my approach. I'm always pleased to find others getting the same or similar results, but no less so to discover discrepancies such are are apparent here, with the fundamental interest of learning more in the course of resolving them.I'll swap samples with Ken and see what we can find out. This is FUN stuff! Long term, I have a (not so) hidden agenda here: I'd like to see the measurement quality in audio discussions go up. As would you, I know. But, I would also like the >interpretation< of the measurement data collected to get more sophisticated than it tends to be. Most meter-reading, anti-subjectivist audio techs are hardcore Logical Positivists, and that is a trap of its own, as surely as subjectivism is. It's a narrow path to navigate. One cliff to fall off is: "All that matters is how it sounds to me!" That way lies madness, circular reasoning, technological stasis, the triumph of advertising, etc.But the other cliff is: "We performed many rigorous, objective measurements on this gear and pronounce it to be as accurate as technology allows." That way forgets that our measurement tools are ad hoc, and are being used like screwdrivers hammering nails. The result is, to some extent, what we face now: a loss as to how to provide a better and more consistent experience to the customer; a situation where more and more design resources yield less and less true improvement. -k Link to comment Share on other sites More sharing options...
speaker dave Posted August 31, 2009 Report Share Posted August 31, 2009 Just for the record, my "academic" comment was not in any way directed at you, (or anyone in particular, for that matter). I think the rigorous approach is great, and I'm learning and re-learning some things in this thread. My "academic" comment was meant to remind us all not to assign absolute figures of merit or rankings to enclosure designs on the basis of data that is highly disconnected from real-world product performance. IOW- understanding the woofer's isolated frequency response is foundational, but it is only the first step in delivering good bass response to the customer. -kHey Ken,I certainly wasn't taking any offence from that. Just genuinly concerned about whether I was putting everyone to sleep with "the graphical analysis of f and Q". Whats interesting to me isn't necessarily of interest to others, but I do enjoy understanding the ins and outs of it all.Regards,David Link to comment Share on other sites More sharing options...
Pete B Posted September 1, 2009 Report Share Posted September 1, 2009 Just for the record, my "academic" comment was not in any way directed at you, (or anyone in particular, for that matter). I think the rigorous approach is great, and I'm learning and re-learning some things in this thread. My "academic" comment was meant to remind us all not to assign absolute figures of merit or rankings to enclosure designs on the basis of data that is highly disconnected from real-world product performance. IOW- understanding the woofer's isolated frequency response is foundational, but it is only the first step in delivering good bass response to the customer. -kI see what we are doing here is confirming that these systems are working as designed,and that the measurements are accurate. It seems that there is a large difference betweenZ's measurements and the published response by AR. In fact, the near field woofer responseis a good way to confirm that a design is working as designed per the T&S and box design spec. It is not useful for showing baffle effects as I'm sure you know.I am also highly interested in a discussion regarding the real world in room performance,perhaps another thread? Link to comment Share on other sites More sharing options...
Pete B Posted September 1, 2009 Report Share Posted September 1, 2009 O.K., reverse that: draw a vertical line at the mean Fc of the unfiltered response which is 43.57 Hz, and, as I read it, it intersects at ~92 dB. Calling that the midband level, we have ~5 dB of bump.With the filter in place, Fc and the peak occur at lower frequencies, and though the peak is also lower SPL, the bump is more like 8 dB. You're saying that the Q should be constant, but I believe Ken suggested that the filter could shift it.... http://www.classicspeakerpages.net/IP.Boar...-1251613951.jpgLet me take another shot at this. People who are really into measurements like to think that they can at least measure some things with a high degree of accuracy and we might thinkthat we get Fs in free air, or Fc in box say to a 1% tolerance since all we have to do ismeasure the peak, or the freq where the phase crosses zero (the definition of resonance).However, it turns out that even at 19 to 40 Hz the VC inductance introduces an error termin that measurement. Crossover inductance and drivers with very high VC inductance introduce even more error, still I do not believe it is significant in this case once the XO isbypassed. We know that Fc should be in the low 40s and that is what you are finding.VC inductance, and the crossover have large effects obviously above 200 Hz, but we shouldkeep in mind that their effects often start at 100 Hz when a large inductor is used to providesome baffle step or FR compensation.As I said before the XO inductor DC resistance will certainly impact the system Qtc in a significantway if it is a good percentage of the VC resistance and less so, or much less so with smaller DCR.It is a tradeoff of more copper for a lower resistance inductor, or more magnet in order to hitthe design target for Qtc. Clearly, the inductor DCR was taken into consideration in designingthe AR woofer. We (I) did Unibox simulations to verify the differences between zero source impedance drive, and with a source impedance between .5 and 1 ohm. Link to comment Share on other sites More sharing options...
Pete B Posted September 1, 2009 Report Share Posted September 1, 2009 Something important here is to confirm/determine if the AR curves are of a 3a with the early inductoror the later #9, because the larger inductor causes a good amount of droop in the passband response.I believe that the AR marketing literature, and the Allison publications were of 3a's with the #7 basedon the fact that the #9 inductor came into use at a much later date.I know this because I simulated it in CALSOD; I've never had an AR-3a to listen to, reverse engineer,or measure.While this is not the exact case that we are talking about, from memory the blue curve is much closerto the early #7 crossover, and the red closer to the later #9 crossover. I believe that the #7 has some peaking right before the rolloff that explains the slight bump in the AR published response. The obviousthing would be to measure the crossover transfer function with each of the inductors.http://www.classicspeakerpages.net/IP.Boar...ost&p=57472 Link to comment Share on other sites More sharing options...
Pete B Posted September 1, 2009 Report Share Posted September 1, 2009 Here is a link to a text that shows, what we call a prototype response curve for a second order highpassfilter, see Figure 5-6 pg 316 in the text.http://books.google.com/books?id=dunqt1rt4...;q=&f=falseIt turns out that if a system is second order then the shape of the response versus Q is defined, locked into what is shown there. It is made into what we call a prototype by scaling Fc to 1 Hz, andthe passband amplitude to 0 dB. Now, we can use those curves to display the response of any truesecond order system by multiplying the frequency scale by the actual Fc, and by adding the passbandsensitivity to the amplitude scale. Using about 40 for Fc, and about 88 dB should be right for the 3a,then pick off the curve with a Q of .5-.6 for direct drive without the crossover DCR, or .7-.8 when the additional resistance is included. Use a Q of .9 to 1.0 when trying to duplicate the system as shownin Villchur's 1957 paper where the damping factor was 1.If you can remember those curves, then you know every sealed box system that can be built simply byscaling frequency and amplitude. The characteristics of 2nd order systems (really any system describedby 2nd order differential equations) have been studied for many years and certain Q values have particular and interesting characteristics. They are often named after the people who studied them:Butterworth (Qtc = .707) also referred to as Maximally (maximum Q but still with a monotonic roll-off) Flat:http://en.wikipedia.org/wiki/Butterworth_filterhttp://en.wikipedia.org/wiki/Chebyshev_filterOr named after a particular feature of their transient response; a Qtc of .5 is critically damped for example. I wrote a bit more about this here; the table formatting is lost:http://www.classicspeakerpages.net/IP.Boar...ost&p=54898Note the shape of the Q = 2 curve (6 dB of peaking), the shape is nothing like that in the measurements,and I'm certain that those speakers were not designed with Qtc = 2. Link to comment Share on other sites More sharing options...
speaker dave Posted September 1, 2009 Report Share Posted September 1, 2009 Something important here is to confirm/determine if the AR curves are of a 3a with the early inductoror the later #9, because the larger inductor causes a good amount of droop in the passband response.I believe that the AR marketing literature, and the Allison publications were of 3a's with the #7 basedon the fact that the #9 inductor came into use at a much later date.I know this because I simulated it in CALSOD; I've never had an AR-3a to listen to, reverse engineer,or measure.While this is not the exact case that we are talking about, from memory the blue curve is much closerto the early #7 crossover, and the red closer to the later #9 crossover. I believe that the #7 has some peaking right before the rolloff that explains the slight bump in the AR published response. The obviousthing would be to measure the crossover transfer function with each of the inductors.http://www.classicspeakerpages.net/IP.Boar...ost&p=57472Here is a question about the #7 vs. the #9 crossovers. If the first is more of a simple 2nd order corner and the second shelves before rolling off, is that due to different woofer response curves (different network needed for same combined response) or was AR moving away from pure 2pi thinking/measuring more towards achieving flatter response in 4pi anechoic conditions? I do believe they moved in that direction with the 3a Ltd (a UK influence?)David Link to comment Share on other sites More sharing options...
kkantor Posted September 1, 2009 Report Share Posted September 1, 2009 Pete,I think 99.9% of serious audio hobbyists, and 100% of audio professionals understand this stuff by now. http://www.kenkantor.com/publications/speakers_by_design/.In fact, writing box design applets seems to have displaced eq applets and sweep generator applets as "My First Undergrad Audio Lab Project." Seriously I find two or three new ones on line every week. (Still like good, old Unibox best.) Aside from the whole issue of target response in the room, the 2nd-order model of an AS woofer has a couple of known weaknesses:1- It does not account for the wide range of possible effects due to cabinet stuffing, and, more importantly,2- Is is a >small signal< model only, and is of limited utility and applicability because of this. -kPS- can you identify the guy in the attached photo? Hint, I'm short but I'm not small.... Link to comment Share on other sites More sharing options...
speaker dave Posted September 1, 2009 Report Share Posted September 1, 2009 PS- can you identify the guy in the attached photo? Hint, I'm short but I'm not small....A famous Australian television engineer??David Link to comment Share on other sites More sharing options...
Carlspeak Posted September 2, 2009 Report Share Posted September 2, 2009 Long term, I have a (not so) hidden agenda here: I'd like to see the measurement quality in audio discussions go up. As would you, I know. But, I would also like the >interpretation< of the measurement data collected to get more sophisticated than it tends to be. Most meter-reading, anti-subjectivist audio techs are hardcore Logical Positivists, and that is a trap of its own, as surely as subjectivism is. It's a narrow path to navigate. One cliff to fall off is: "All that matters is how it sounds to me!" That way lies madness, circular reasoning, technological stasis, the triumph of advertising, etc.But the other cliff is: "We performed many rigorous, objective measurements on this gear and pronounce it to be as accurate as technology allows." That way forgets that our measurement tools are ad hoc, and are being used like screwdrivers hammering nails. The result is, to some extent, what we face now: a loss as to how to provide a better and more consistent experience to the customer; a situation where more and more design resources yield less and less true improvement. -kNOW AIN'T THAT THE TRUTH!!!! Link to comment Share on other sites More sharing options...
speaker dave Posted September 2, 2009 Report Share Posted September 2, 2009 I think 99.9% of serious audio hobbyists, and 100% of audio professionals understand this stuff by now.Understand it? Certainly they are aware of it but understanding is another matter. I find I need to re-read most of the landmark papers every couple of years and still pull out new concepts or at least re-familiarize myself with them. Like you probably, I go all the way back to typing in Program Woof in Fortran, into a terminal, to have it spit out a crude response curve: marvelous stuff at the time.In the last couple of years I've learned more about woofer/box science while trying to create high output very compact subwoofers. Certainly its the large signal, not the small signal issues that create the challenges.The first step along that path was viewing the woofer and enclosure as a system rather than just assuming that "more magnet is good". The AR1 was one of the first steps there, but Jenson and RCA were in there too. (And don't make me prove that RCA invented the acoustic suspension system, imagine the flames!)Second guess: is it Burl Ives?David Link to comment Share on other sites More sharing options...
genek Posted September 2, 2009 Report Share Posted September 2, 2009 I think 99.9% of serious audio hobbyists, and 100% of audio professionals understand this stuff by now.I guess that makes me a non-serious hobbyist, because 90% of this stuff goes right over my head. I've worked enough years in engineering testing and instrumentation to be able to ask some semi-intelligent questions about taking measurements, but understanding what measurements mean is another matter. Link to comment Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.