Guest kaiser_soze Posted March 4, 2006 Report Share Posted March 4, 2006 Hello all,I am new to this forum, but not new to Advent speakers. I've owned and enjoyed a pair of Advent 5012 speakers since '82 or '83. I believe that this was the last variant of the "Large Advent" made prior to the takeover by Jensen.I was motivated to write something after I perused the thread having to do with modifications to the Advent loudspeakers, which thread was begun by Pete B., who reported that he and his family determined certain deficiencies in the sound of the speakers.There are many reasons why subjective listening tests are not always reliable, and one such reason is that the frequency response that you perceive depends substantially on where you sit with respect to the speaker, particularly with respect to the vertical off-axis angle. Small changes in the vertical off-axis angle can bring about enormous changes in the quality of the sound. Owing to the fact that the Advent speaker is a 2-way design using a low-order crossover, it does not have "a" characteristic sound per se. Rather, it has a wide range of characteristic sounds, depending primarily on the vertical placement of the speaker relative to the height of your ears when you listen, and depending on whether you place the speakers with the tweeter on top or the woofer on top. The only way that there could be a single characterstic sound of this speaker, would be if the effect of the direct radiation could somehow be eliminated so that only the average room response were heard, but that never happens in practice, and if it did, the sound would then depend as much on the room as on the speaker.The reason why the sound has such a strong dependency on the vertical relationship of the listener to the speaker, has to do with the lobes in the emitted sound radiation pattern. The lobes have several secondary causes which include the directivity of the woofer at higher frequencies, but the primary cause is the phase cancellation, which itself arises from two sources. One source is when the electromotive forces associated with the same frequency at both voice coils are not in phase, and the other source is the phase difference that results from not being at equal distances from both drivers.I'm not familiar with the crossover that is used in the Advent loudspeakers, but there is at least one 1st-order design that is nearly phase coherent, and because I can still recall the images of how square-wave pulses were preserved by Advent speakers, let's assume that the two drivers are phase-coherent, and concern ourselves only with the effect of being at different distances from the two drivers.Engineers who design crossovers for a living, think of crossovers mainly from the perspective to the lobing pattern that is seen on the vertical axis. Unless you have studied acoustics or possibly antenna theory, it probably won't be immediately apparent what this is. So imagine that you have glued a string to the front of the speaker between the woofer and the tweeter, and then you pull it out straight and centered from side to side, but you vary how high you hold it at the far end. At any given distance along that string, if you measure the SPL for any given frequency, the value will depend on the relative distance to the two drivers, because that relative distance will determine the relative phase between the sound emitted at that frequency by the two drivers. If you change the vertical angle, the relative distance to the two drivers will change, which will change the phase relationship for the two drivers at the given frequency, which will determine the extent to which the output from the two drivers at the frequency will be additive or cancelling, which will affect the SPL at that frequency. For any given frequency, there will be vertical angles where the outputs from the two drivers are 180 degrees apart in phase and will cancel. The overall effect is that for any frequency that is emitted in appreciable measure by both drivers, there will be a vertical lobing pattern. The specific crossover design that is used determines the shape of the lobing pattern, and crossover designers characterize different crossovers according to their specific lobing pattern in the vertical listening axis.Consider, for example, what will happen if you set the speaker so that the approximate center between the woofer and the tweeter is placed at the same height as your ears in your favorite listening position. If you then sit at a typical listening position, the woofer's voice coil will be perhaps 3" further away than the tweeter's voice coil. The fundamental frequency that will experience perfect acoustic cancellation will be the frequency where 3" is one-half of a wavelength. Perfect cancellation occurs when the two separate waves are 180 degrees out of phase, which is the same as being one-half of a wavelength apart. So the wavelength where perfect acoustic cancellation occurs will be 6", and since sound travels at roughly 1000 feet per second, that wavelength corresponds to 2kHz. Hence, if you set the speaker with the tweeter at listening height, the frequency response that you will experince will exhibit a "suck-out" at 2kHz. Looking at it from the standpoint of acoustic radiation lobes, at 2kHz, there will be a lobe above the horizontal, another lobe below the horizontal, and the gap between the two lobes will be aligned with the horizontal line extending outward from a point on the face of the speaker where the tweeter voice coil is 3" closer than the woofer voice coil.All loudspeakers except for those that have their tweeter concentric with the midrange exhibit such lobing patterns, and for any loudspeaker, the lobing pattern changes with frequency. The crossover design influences the lobing pattern. As you move in frequency away from the crossover frequency, one lobe starts to predominate the other, and the line that defines the gap between the lobes shifts toward the weaker lobe. As the stronger lobe pushes the lesser lobe out of the way, the SPL for that frequency becomes more uniform over the vertical axis. The greater the order of the crossover, i.e., the steeper the crossover slope, the more rapidly this effect occurs as you move away from the crossover frequency. In order words, one beneficial effect of higher order crossovers is that the range of frequencies where a gap between the two major lobes can occur within the normal range of vertical listening angles, is reduced.In a 2-way design using a 1st-order crossover, the range of frequencies where both drivers are contributing significantly to the total SPL, covers a span of several octaves through the midrange. If you measure the frequency response of the Advent 5012 speaker at a location that is equi-distant from both voice coils, the frequency response will be smooth and uniform except for a broad, gradual rise that is centered at about 500 Hz but that is spread over the full audible spectrum, and with the rounding becoming more severe at both ends. If you measure the frequency response at any other vertically off-axis location, the frequency response will be highly erratic, exhibiting an orderly sequence of frequencies that are additive for that relative distance from the two drivers, interspersed with frequencies that exhibit phase cancellation for that relative distance from the two drivers. The lobing pattern rotates vertically in one or the other direction according to which direction you move in frequency, and exhibits a repeated cycle as you pass through the sequence of frequencies, but with the stronger lobe pushing the weaker lobe further out of the way with each pass through the cycle.If the two voice coils are vertically aligned, the line along which the two voice coils will be equidistant, will be horizontal. If the two outputs of the crossover are additionally aligned in phase, then the predominant lobe for the crossover frequency will extend horizontally forward from the face of the speaker. If the crossover maintains phase alignment for other frequencies as you move away from the crossover frequency, the predominant lobe for all frequencies will extend horizontally forward, which yields the potential for a uniform frequency response at locations along the forward horizontal extension from the speaker. But one consequence of phase coherency in the crossover is that the lobes and the gaps between the lobes will be vertically aligned throughout the frequencies within the overlap region. If you listen at either of the vertical positions that are at the upper and lower edges of the prodominant lobe, the midrange output of the speaker will practically vanish!When one voice coil is recessed into the speaker cabinet, the effect is to cause a tilting of the lobing pattern toward that voice coil. If the tweeter is located above the woofer and the woofer's voice coil is recessed, the lobing pattern will tilt downward away from the speaker, and the line along which the two voice coils are equidistant from the listener will intersect the floor several feet in front of the speaker.In the Advent loudspeaker, the effect that the lobing patterns have on the frequency response at a typical listening position, is somewhat exacerbated due to the wide separation of the woofer and the tweeter. Yet, there is one way that you can virtually elminate these effects, and that is by arranging the speakers such that at typical listening positions, the two voice coils are as near as possible to the same distance. There are several approaches. One is to set the speaker close to the floor and then tilt it backward, which also improves the bass due to the boundary effect. I use a pair of small tables that match the footprint of the Advent very closely, and that are 16" high. Sitting on those tables, the top of the speaker, which is the end closest to the woofer of course, is just a little below eye level. With the speaker turned over, the woofer is a little higher, which reduces the effect of the woofer's directivity. Morevover, due to the tilt of the lobing pattern toward the woofer's recessed voice coil, instead of the line along which the two drivers are equidistant intersecting the floor, that line, which is the only line along which the gaps at the edges of the predominant lobes for all frequencies in the overlap region are avoided, is now tilted upward. In fact, if I now set the speaker directly on the floor, the predominate lobe for all frequencies in the overlap would be placed even more optimally in the room, being at listening height at typical distances, instead of at listening height just a few feet in front of the speaker as is now the case.The difference in sound quality is readily apparent. On a few occasions I have been bothered by the fact that with the grill panels in place, the ADVENT plaques were upside-down, and have tried them for a while turned over the usual way. I've not left them that way for long, because the degradation in sound quality, while not exactly overt, is easily discernible. When they are right-side up, unless you sit real close and a little lower than the speaker in accordance with the distance, you will be much closer to the tweeter than the woofer, and in addition to this causing a highly irregular frequency response, at typical listing positions, a significant amount of the midrange will be cancelled, with the effect being a perceived emphasis on the higher frequencies in the 5kHz - 10kHz range, which gives the speaker a sound that is, well, perhaps it could be described at "thin", or maybe slightly "tinny", almost as though it had a cheap metallic tweeter.I am not inclined to mess around with the crossover. I would only go that route if I could have an exact understanding on the effect that I was having on the speaker sound, and that would require an analysis of the effect that I was having on the lobing patterns throughout the midrange of the speaker, where both drivers contribute significantly to the output of the speaker. I would start by first experimenting with different speaker height placements relative to my ears, and I would be inclined to use pink noise to improve my understanding of how the sound is different at different vertically off-axis listening positions. Only after I was confident that I had an intimate understanding of just how the sound quality varies with the vertically off-axis listening position, would I consider messing around with the crossover. I also would not replace any of the crossover components for the alleged reason of component aging unless I was able to confirm using test equipment that the component that I was replacing was genuinely faulty. Link to comment Share on other sites More sharing options...
Guest Dino1956 Posted March 6, 2006 Report Share Posted March 6, 2006 Mr. Soze,I found your Post quite interesting and very informative. I have the Large Advents and the top of the Speaker is the Tweeter section. I just yesterday refoamed my Woofers and although the Speakers sound excellent, I noticed because there height is about 6 inches below ear level, I'm hearing more high frequency. I've switched the High Frequency to switch to decrease which seems to help. Thanks for the imformation and observation.Dino Link to comment Share on other sites More sharing options...
Guest kaiser_soze Posted March 6, 2006 Report Share Posted March 6, 2006 The reason that the tweeter sounded loud probably has more to do with phase cancellation and lobing than with the more obvious effect of the steady decline in acoustic energy as you move further away from a driver. If you are much closer to the tweeter than the woofer, the tweeter will sound louder of course, but that effect will vanish rather quickly as you move further away from the speaker, whereas the effect of lobing will remain the same as you move away as long you maintain the same differential in the distance to the two drivers, until you reach the point where the room response subdues the lobing pattern in the direct radiation.My 5012 speakers don't have the tweeter level control, but I would prefer to think of the tweeter level control as being useful for correcting the room response, after I had first tried to improve the orientation of the speaker so that the predominant lobe is projected into the middle region of the room at typical listener ear level. If they are sitting with the tweeter above the woofer and still below listener ear level, if it is possible to tilt them back a little, that will almost certainly improve the sound. That will also have some improvement for the effect of woofer directivity. It is a small thing and it seems silly that anyone would give something so minor as a recommendation, but it could improve the sound significantly. Link to comment Share on other sites More sharing options...
Pete B Posted March 16, 2006 Report Share Posted March 16, 2006 I notice from your profile that your name is Tom Barber, and as I recall you mentioned physics; is this your profile: http://www.seed.slb.com/en/scictr/experts/barber.htmPete B.>Hello all,>>I am new to this forum, but not new to Advent speakers. I've>owned and enjoyed a pair of Advent 5012 speakers since '82 or>'83. I believe that this was the last variant of the>"Large Advent" made prior to the takeover by>Jensen.>>I was motivated to write something after I perused the thread>having to do with modifications to the Advent loudspeakers,>which thread was begun by Pete B., who reported that he and>his family determined certain deficiencies in the sound of the>speakers.>>There are many reasons why subjective listening tests are not>always reliable, and one such reason is that the frequency>response that you perceive depends substantially on where you>sit with respect to the speaker, particularly with respect to>the vertical off-axis angle. Small changes in the vertical>off-axis angle can bring about enormous changes in the quality>of the sound. Owing to the fact that the Advent speaker is a>2-way design using a low-order crossover, it does not have>"a" characteristic sound per se. Rather, it has a>wide range of characteristic sounds, depending primarily on>the vertical placement of the speaker relative to the height>of your ears when you listen, and depending on whether you>place the speakers with the tweeter on top or the woofer on>top. The only way that there could be a single characterstic>sound of this speaker, would be if the effect of the direct>radiation could somehow be eliminated so that only the average>room response were heard, but that never happens in practice,>and if it did, the sound would then depend as much on the room>as on the speaker.>>The reason why the sound has such a strong dependency on the>vertical relationship of the listener to the speaker, has to>do with the lobes in the emitted sound radiation pattern. The>lobes have several secondary causes which include the>directivity of the woofer at higher frequencies, but the>primary cause is the phase cancellation, which itself arises>from two sources. One source is when the electromotive forces>associated with the same frequency at both voice coils are not>in phase, and the other source is the phase difference that>results from not being at equal distances from both drivers.>>I'm not familiar with the crossover that is used in the Advent>loudspeakers, but there is at least one 1st-order design that>is nearly phase coherent, and because I can still recall the>images of how square-wave pulses were preserved by Advent>speakers, let's assume that the two drivers are>phase-coherent, and concern ourselves only with the effect of>being at different distances from the two drivers.>>Engineers who design crossovers for a living, think of>crossovers mainly from the perspective to the lobing pattern>that is seen on the vertical axis. Unless you have studied>acoustics or possibly antenna theory, it probably won't be>immediately apparent what this is. So imagine that you have>glued a string to the front of the speaker between the woofer>and the tweeter, and then you pull it out straight and>centered from side to side, but you vary how high you hold it>at the far end. At any given distance along that string, if>you measure the SPL for any given frequency, the value will>depend on the relative distance to the two drivers, because>that relative distance will determine the relative phase>between the sound emitted at that frequency by the two>drivers. If you change the vertical angle, the relative>distance to the two drivers will change, which will change the>phase relationship for the two drivers at the given frequency,>which will determine the extent to which the output from the>two drivers at the frequency will be additive or cancelling,>which will affect the SPL at that frequency. For any given>frequency, there will be vertical angles where the outputs>from the two drivers are 180 degrees apart in phase and will>cancel. The overall effect is that for any frequency that is>emitted in appreciable measure by both drivers, there will be>a vertical lobing pattern. The specific crossover design that>is used determines the shape of the lobing pattern, and>crossover designers characterize different crossovers>according to their specific lobing pattern in the vertical>listening axis.>>Consider, for example, what will happen if you set the speaker>so that the approximate center between the woofer and the>tweeter is placed at the same height as your ears in your>favorite listening position. If you then sit at a typical>listening position, the woofer's voice coil will be perhaps>3" further away than the tweeter's voice coil. The>fundamental frequency that will experience perfect acoustic>cancellation will be the frequency where 3" is one-half>of a wavelength. Perfect cancellation occurs when the two>separate waves are 180 degrees out of phase, which is the same>as being one-half of a wavelength apart. So the wavelength>where perfect acoustic cancellation occurs will be 6",>and since sound travels at roughly 1000 feet per second, that>wavelength corresponds to 2kHz. Hence, if you set the speaker>with the tweeter at listening height, the frequency response>that you will experince will exhibit a "suck-out" at>2kHz. Looking at it from the standpoint of acoustic radiation>lobes, at 2kHz, there will be a lobe above the horizontal,>another lobe below the horizontal, and the gap between the two>lobes will be aligned with the horizontal line extending>outward from a point on the face of the speaker where the>tweeter voice coil is 3" closer than the woofer voice>coil.>>All loudspeakers except for those that have their tweeter>concentric with the midrange exhibit such lobing patterns, and>for any loudspeaker, the lobing pattern changes with>frequency. The crossover design influences the lobing>pattern. As you move in frequency away from the crossover>frequency, one lobe starts to predominate the other, and the>line that defines the gap between the lobes shifts toward the>weaker lobe. As the stronger lobe pushes the lesser lobe out>of the way, the SPL for that frequency becomes more uniform>over the vertical axis. The greater the order of the>crossover, i.e., the steeper the crossover slope, the more>rapidly this effect occurs as you move away from the crossover>frequency. In order words, one beneficial effect of higher>order crossovers is that the range of frequencies where a gap>between the two major lobes can occur within the normal range>of vertical listening angles, is reduced.>>In a 2-way design using a 1st-order crossover, the range of>frequencies where both drivers are contributing significantly>to the total SPL, covers a span of several octaves through the>midrange. If you measure the frequency response of the Advent>5012 speaker at a location that is equi-distant from both>voice coils, the frequency response will be smooth and uniform>except for a broad, gradual rise that is centered at about 500>Hz but that is spread over the full audible spectrum, and with>the rounding becoming more severe at both ends. If you>measure the frequency response at any other vertically>off-axis location, the frequency response will be highly>erratic, exhibiting an orderly sequence of frequencies that>are additive for that relative distance from the two drivers,>interspersed with frequencies that exhibit phase cancellation>for that relative distance from the two drivers. The lobing>pattern rotates vertically in one or the other direction>according to which direction you move in frequency, and>exhibits a repeated cycle as you pass through the sequence of>frequencies, but with the stronger lobe pushing the weaker>lobe further out of the way with each pass through the cycle.>>If the two voice coils are vertically aligned, the line along>which the two voice coils will be equidistant, will be>horizontal. If the two outputs of the crossover are>additionally aligned in phase, then the predominant lobe for>the crossover frequency will extend horizontally forward from>the face of the speaker. If the crossover maintains phase>alignment for other frequencies as you move away from the>crossover frequency, the predominant lobe for all frequencies>will extend horizontally forward, which yields the potential>for a uniform frequency response at locations along the>forward horizontal extension from the speaker. But one>consequence of phase coherency in the crossover is that the>lobes and the gaps between the lobes will be vertically>aligned throughout the frequencies within the overlap region. >If you listen at either of the vertical positions that are at>the upper and lower edges of the prodominant lobe, the>midrange output of the speaker will practically vanish!>>When one voice coil is recessed into the speaker cabinet, the>effect is to cause a tilting of the lobing pattern toward that>voice coil. If the tweeter is located above the woofer and>the woofer's voice coil is recessed, the lobing pattern will>tilt downward away from the speaker, and the line along which>the two voice coils are equidistant from the listener will>intersect the floor several feet in front of the speaker.>>In the Advent loudspeaker, the effect that the lobing patterns>have on the frequency response at a typical listening>position, is somewhat exacerbated due to the wide separation>of the woofer and the tweeter. Yet, there is one way that you>can virtually elminate these effects, and that is by arranging>the speakers such that at typical listening positions, the two>voice coils are as near as possible to the same distance. >There are several approaches. One is to set the speaker close>to the floor and then tilt it backward, which also improves>the bass due to the boundary effect. I use a pair of small>tables that match the footprint of the Advent very closely,>and that are 16" high. Sitting on those tables, the top>of the speaker, which is the end closest to the woofer of>course, is just a little below eye level. With the speaker>turned over, the woofer is a little higher, which reduces the>effect of the woofer's directivity. Morevover, due to the>tilt of the lobing pattern toward the woofer's recessed voice>coil, instead of the line along which the two drivers are>equidistant intersecting the floor, that line, which is the>only line along which the gaps at the edges of the predominant>lobes for all frequencies in the overlap region are avoided,>is now tilted upward. In fact, if I now set the speaker>directly on the floor, the predominate lobe for all>frequencies in the overlap would be placed even more optimally>in the room, being at listening height at typical distances,>instead of at listening height just a few feet in front of the>speaker as is now the case.>>The difference in sound quality is readily apparent. On a few>occasions I have been bothered by the fact that with the grill>panels in place, the ADVENT plaques were upside-down, and have>tried them for a while turned over the usual way. I've not>left them that way for long, because the degradation in sound>quality, while not exactly overt, is easily discernible. When>they are right-side up, unless you sit real close and a little>lower than the speaker in accordance with the distance, you>will be much closer to the tweeter than the woofer, and in>addition to this causing a highly irregular frequency>response, at typical listing positions, a significant amount>of the midrange will be cancelled, with the effect being a>perceived emphasis on the higher frequencies in the 5kHz ->10kHz range, which gives the speaker a sound that is, well,>perhaps it could be described at "thin", or maybe>slightly "tinny", almost as though it had a cheap>metallic tweeter.>>I am not inclined to mess around with the crossover. I would>only go that route if I could have an exact understanding on>the effect that I was having on the speaker sound, and that>would require an analysis of the effect that I was having on>the lobing patterns throughout the midrange of the speaker,>where both drivers contribute significantly to the output of>the speaker. I would start by first experimenting with>different speaker height placements relative to my ears, and I>would be inclined to use pink noise to improve my>understanding of how the sound is different at different>vertically off-axis listening positions. Only after I was>confident that I had an intimate understanding of just how the>sound quality varies with the vertically off-axis listening>position, would I consider messing around with the crossover. >I also would not replace any of the crossover components for>the alleged reason of component aging unless I was able to>confirm using test equipment that the component that I was>replacing was genuinely faulty. Link to comment Share on other sites More sharing options...
Guest kaiser_soze Posted March 21, 2006 Report Share Posted March 21, 2006 Pete,That isn't me.Your approach of inserting a low-power passive circuit between the pre-out and main-in, offers multiple advantages as compared to direct modifications to the crossover. I'm planning to buy a new Denon receiver very soon, and because it will give me the ability to insert such a circuit for the center channel, I am thinking that I may use that approach to insert a notch filter to correct for the peak in the tweeter response of my MC500 (Cambridge Soundworks). Presently I am using two simple resistors, one in series to reduce the tweeter power level by 3 db, and another in parallel with that series circuit to restore the impedance to the original 4 Ohms of the tweeter. The peak was originally 6db, and now it is 3db, but of course the output of the tweeter is reduced by 3db overall, so there is a trench below and it rolls off sooner. By using a proper notch filter between the pre-out and the main-in, I should be able to get it much flatter. I'll probably use pre-manufactured resistors this time instead of the 1.65 Ohm and 13.6 Ohm resistors that I hastily made from writing pencils on the wintry day that I decided to attenuate that tweeter. Link to comment Share on other sites More sharing options...
Pete B Posted April 1, 2006 Report Share Posted April 1, 2006 I wondered if you've done any EE design work based on your suggestion not to change caps.I A/B compared a pair of what I call Rev1 16uF/16uF early Large Advents, to the New Advents my Rev3. While they were similar, the New Advent was worse in my opinion. The early Rev1 seemed to have much better vertical dispersion the sound character did not change much, contrary to what you claim on the assumption that your much later Advents are comparable to all Large Advents. The new tweeter did seem to have more high end extension probably as a result of lower moving mass. You make many assumptions and jump to conclusions without even trying the mod I suggest.You also make many assumptions about tweeters and vertical dispersion that I just don't have the time to get into. Let me offer just one point, all of the better metal dome tweeters, certainly those used by Paradigm and PSB have their primary dome resonance above 20 kHz, not below as you claim. The rise is from response shaping in the crossover in every decent quality speaker that I've analyzed.I don't really want to get into it, I'd suggest you do a bit more research.Pete B.>Pete,>>That isn't me.>>Your approach of inserting a low-power passive circuit between>the pre-out and main-in, offers multiple advantages as>compared to direct modifications to the crossover. I'm>planning to buy a new Denon receiver very soon, and because it>will give me the ability to insert such a circuit for the>center channel, I am thinking that I may use that approach to>insert a notch filter to correct for the peak in the tweeter>response of my MC500 (Cambridge Soundworks). Presently I am>using two simple resistors, one in series to reduce the>tweeter power level by 3 db, and another in parallel with that>series circuit to restore the impedance to the original 4 Ohms>of the tweeter. The peak was originally 6db, and now it is>3db, but of course the output of the tweeter is reduced by 3db>overall, so there is a trench below and it rolls off sooner. >By using a proper notch filter between the pre-out and the>main-in, I should be able to get it much flatter. I'll>probably use pre-manufactured resistors this time instead of>the 1.65 Ohm and 13.6 Ohm resistors that I hastily made from>writing pencils on the wintry day that I decided to attenuate>that tweeter. Link to comment Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.