When BBC launched their famous BBC Monitors they incorporated an unconventional dip in frequency response nicknamed the BBC Dip. And check out our newest RU-vid channel / @octaverecordsanddsdst... Octave Records.
A little more detail (and slightly different version) of this from one of the founder of Harbeth - at the BBC when this was going on (from the Audyssey room adjustment FAQ) with additional comments from Sean Olive: The well-known and highly regarded English speaker manufacturer, Harbeth (makers of the legendary LS3/5a speaker, itself based on a BBC design) have this to say: "There is much myth, folklore and misunderstanding about this subject. "The 'BBC dip' is (was) a shallow shelf-down in the acoustic output of some BBC-designed speaker system of the 1960s-1980s in the 1kHz to 4kHz region. The LS3/5a does not have this effect, neither in the 15 ohm nor 11 ohm, both of which are in fact slightly lifted in that region. "According to Harbeth's founder, who worked at the BBC during the time that this psychoacoustic effect was being explored, the primary benefit this little dip gave was in masking of defects in the early plastic cone drive units available in the 1960's. A spin-off benefit was that it appeared to move the sound stage backwards away from the studio manager who was sitting rather closer to the speakers in the cramped control room than he would ideally wish for. (See also Designer's Notebook Chapter 7). The depth of this depression was set by 'over-equalisation' in the crossover by about 3dB or so, which is an extreme amount for general home listening. We have never applied this selective dip but have taken care to carefully contour the response right across the frequency spectrum for a correctly balanced sound. Although as numbers, 1kHz and 4kHz sound almost adjacent in an audio spectrum of 20Hz to 20kHz, the way we perceive energy changes at 1kHz or 4kHz has a very different psychoacoustic effect: lifting the 1kHz region adds presence (this is used to good effect in the LS3/5a) to the sound, but the 4kHz region adds 'bite' - a cutting incisiveness which if over-done is very unpleasant and irritating. "You can explore this effect for yourselves by routing your audio signal through a graphic equaliser and applying a mild cut in the approx. 1kHz to 4kHz region and a gradual return to flat either side of that." Sean Olive has commented on the futility of Audyssey's attempting to second-guess if your speakers are poorly designed and thus would benefit from MRC: "Neither I nor Floyd Toole had never heard about the Gundry* dip until about 2 months ago when an audio reviewer used the term in an email to us. Many poorly designed 2-way loudspeakers already have dips in the sound power response in the cross-over range 1-3 kHz where the directivity of the woofer is too high compare to the directivity of the tweeter at those frequencies. As a result, this produces a notch in the sound power response of the loudspeaker, usually followed by a peak. Depending on the bandwidth and depth of the notch, it is the peak that is often heard as sounding objectionable (harshness, hardness or excessive brightness). The extent to which this a problem depends on whether you are sitting on or off axis, and the reflectivity of the room. Some room correction products, by default, have a dip in their target curve in an attempt to compensate for this sound power problem, essentially trying to second guess whether or not the loudspeaker is well-designed. As I've shown in a recent AES paper - this doesn't always lead to good results." (*Just another name for the "BBC Dip".)
Thank you for the very nice explanation, the topic of the BBC dip came up in a class I was teaching a few weeks back for my second year students (introduction to speaker management systems for live sound), studio engineering students crossing over into live sound engineering. All I could tell them from my knowledge about the BBC dip is that it was crossover compensation done for some BBC studio monitors, being a live engineer I have not spent that much time in the studio since I was studying myself twenty plus years ago.
Also, the human ear is five decibels more sensitive in the range from 1k to 5k. Notching that range down makes a more pleasurable speaker, not a more accurate one. Great report. Great question. Thanks.
You said it before I could. That frequency range is where the sibilance in human speech occurs, so it's very important for our ears to be more sensitive to that range so we can really pick apart all the subtlety and nuance of spoken language. Also, as sound gets louder our ears will naturally compress the sound which is why most manufacturers have a passive EQ that takes more and more of that frequency range out the more you crank up the volume to preserve the "BBC dip" even as our ears are compressing the highs, mids, and lows together.
3:54 I think you meant to say “if the woofer continues to go up too high, it’s gonna get into _treble_ . (I immediately would like to apologize to you and your audience for that terribly cheesy joke. First video I’ve seem from you, very interesting and very well explained. But I couldn’t resist. I’ll go away now to shamefully hide in the bassment.)
3000 hurts!!! Around 3000hz is the most infuriating and fatiguing frequency there is. Around this region the human ear is sensitive. It's no coincidence that the resonant frequency of the ear canal is anywhere from 2 to 4khz. Listen to a frequency test at the same sound power level across all frequencies. Your ears will bleed, aka you'll get a headache, from 2 to 4 khz if you started with too much volume.
At the crossover point, the tweeter has a wider dispersion angle than the woofer does. So the bottom range of the tweeter can be reduced in level to compensate. You wouldn't hear it in an anechoic chamber but that wider dispersion of the tweeter definitely has an audible impact in a sound reflective environment.
I design amplifiers and speaker enclosures as a hobby/home business. I've most certainly observed the need to notch the signal to get the best sound. However I would disagree on what necessitates this. From my testing, it appears to improve the tone in one way systems just as much, and if you look at the frequency response curve of a typical 12" woofer, you'll see why: there's often a particular sensitivity to some narrow bands of frequencies in that range. For example, a speaker I often design for is the Eminance Beta 12A. Google it, you will see a broad increase in sensitivity between about 1 kHz and 4 kHz with a notch part way through, and an even higher peak around 2 kHz. The way I EQ this particular speaker for best sound (to my ear) is with a broad notch from 150 and 4k with the crossover at 3.5k, which gives an overall bass flattening. I then superimpose two notches, one from 880 Hz to 1.5 kHz, and the other from 1.8k to 3k. As you can see from the graph if you google that speaker, that essentially flattens it. That being said, I don't tend to find a need for this EQ style with open back enclosures. Those peaks only seem to cause "harshness" when the speaker has some resistance on it, like a closed or ported box, and especially in sealed boxes. That tells me that the cause of those peaks is probably some kind of interference pattern or standing waves. That's why it sounds harsh, not because there's too much of those frequencies, and not because it's in the crossover region.
bought a pair of home made 3wy coax utah celesta speakers from an oak park guy in 71,almost 3ft tall one ft x 1 ft wide & deep.I still have the cabinets,w pyle driver woofers, a hundred for the pair originally..they had such a nice sound curtain,,30 wpc filled a large room to overflow, so next door neighbors could hear breezin, by g benson,,low shakin bass,never boomy, Neighbor remarked how clear the sound was coming accross the yard.still had them in mid 80s,using goodman achromat 2 ways with second speaker connections on the marantz,oohh, they did sound sweet,on a lark one day I wondered how the utahs would sound if I coated the speaker cones with rubber cement,I did,being careful not to get the stuff on any rubber,,seemed to make the sound even tighter & more 'live & in front,, and bit less hiss,while not taking away hi frequencies or definition,
Years ago I would do similar but I have to tell you. Getting a speaker that is flat in the first place or has very small Peaks and valleys I should say. If it's a material that is pleasing to my ear I would much prefer that. For example if you listen to the dynaudio esotar2 or morel supremo. They are pretty amazing without any work. They just produce a sound that you cannot get out of a cheaper driver no matter how much crossover work you do to it. It's a especially noteworthy around 2 khz where many budget speakers really get into trouble with horns and certain voices. When you look at the crossovers on some of the new dynaudio home speakers. They are very simple. A cap and coil only is common on a 2way. The majority of the engineering is in speaker placement and the box design along with the Box material and even the glue. The filters on the other hand are extremely simple.
Eminance makes a much flatter driver for not much more. The only disadvantage being that you're going from a 38 oz magnet to an 80 oz magnet. That seems to flatten the response dramatically. The Beta 12A is about $80, and the Delta Pro 12A with the 80 oz. magnet is about $100. I haven't used them before simply because the driver weighs 20 lbs. (I do mostly instrument amps). However, next time I do a hi-fi build, I will see if I can work in the deltas. I bet it will make a massive difference.
Eliminating awkward midrange/tweeter crossover phase distortion might be one reason that speakers like electrostatics which avoid the crossover in that region sound so smooth. Of course even electrostatic drivers have different sized regions with slightly different resonances but they share the same deflection voltage from the same "side" of a crossover filter. Phase distortion in the low-mid to bass region isn't harsh but simply detracts from the purity of bass notes (far less objectionable than upper midrange distortion).
This is why final test is with my ears. I did not know about the BBC Dip, but that range, even down to 800Hz now that Paul mentioned it, generally from 1k-4k with even the expensive drivers is where I spend the most time afjusting, trying to keep from vomiting. Thank you!
As a music and sound producer, I tend to tame this frequency band when mixing vocals to reduce potential harshness. Thanks to this video, now I know why.
I take it these days it's more of a parametric thing done with software rather than something built into the hardware. So you just re-adjust the EQ mixdown to sound good depending on whatever speaker or headphone setup you have.
Wow never knew of the compromises made where the sound level hits the crossover and I thought I knew a thing or two about HiFi. Many thanks for making sharing this. Alan.
This was really an interesting video, and I think it gives a good insight into the complex world of building speakers, as well as the tricky task of balancing measurements vs. what our ears are telling us.
Another reason for a dip at those crossover frequencies 1.5-3 Khz is that the enclosure box adds at least 1 db and maybe up to 3 db in additional loudness due to diffraction based on the front baffle width. I'm surprised Paul did not mention this. So if you add 1 db ( Crossover bump) plus a 2 db baffle bump that can make things quite unruly at the crossover region. I've experienced this as I build my own speakers....Other than that this is one of the most informative videos I've seen Paul make for the average consumer.
I had a pair of Rogers LS3/5A and they were fantastic for vocals.. they nailed it with good drivers, excellent crossovers, build quality and serial matched speaker pairs. I now have SVS Ultras and although for vocals not quite as good for that, they are superb everywhere even at low volumes and worth every penny. I think KEF have nailed imaging though with combined tweeter/ woofer unidriver, imaging sounds so good other speakers appear to severly lack on imaging and it is noticable..
Thanks for explaining things to the world, via RU-vid. I was into audio heavily and have actually never heard of the BBC dip. What I learned was the Fletcher-Munson curve of human hearing. For audio to sound flat you had to make sure speakers (and their environment if possible) needed to produce the inverse of a Fletcher-Munson curve, which is not just a curve. The frequencies that you cited as being the loudest are frequencies the human hearing is most sensitive at, which happens to be in the same range as a baby crying, which explains why that is the most loudest. If you get a RTA and a kids piano or any piano for that matter and hit each key, you will see the ones in that overly sensitive range sound way louder even though the spectrum analyser indicated the volume was the same. I was mostly into car audio, so one had to be careful in the environment because of hard reflective surfaces and resonances while making passive crossovers, subwoofer and midbass box design and placement. Lower order crossovers have the most shallow rolloff but sounded more lively. I've made up to 24 db/ octave passive crossovers and found them to be rather dry sounding. (air core coils, Wonder Caps). Once you start to look at the polar output as you increase the complexity and the order of the crossover, things change yet again. This was very important to minimize reflections, but it wasn't that difficult as I knew where you were going to sit in a car. Sweep the car for modes and resonances. Try to dampen them. Put the crossovers where there was a peak and make sure the points were far enough to be flat. Biggest challenge was time and off axis volumes as people often sat closer to a speaker on one side of the car than others. In the end people went multiple amps on a FFT analyser with auto adjustments on each channel, but they couldn't do it "old school" if you paid them.
Great video! I've been experiencing everything you've been describing. A compromise you didn't mention was figuring out the best way to get the woofer cutting off low enough. I so wanted to go with air core but just didn't have enough coil. Ended up running a much higher value cap than expected with a low value resister to lower the q in parallel which worked out perfectly but i'm sure there's problems with that!
I bought a second Quad 306 amplifier and bi-wired one to the woofer, and one to the tweeter. Big improvement, especially on speech, piano, violin and harpsichord. Search the web and you can find intermodulation distortion graphs that support bi-wiring, particularly of amplifiers. Also I just bought a Humax 400T FreeView box which I use predominantly with Radio 3, driving my quads through a passive pre-amp fed by an optical link from the box. Clarity is superb, especially on live broadcasts as opposed to commercial CDs. Speakers are Mordaunt Short 2-way floor-standing from about 1980, that cost me £50 on ebay. You can record radio, and transfer the recordings to USB sticks. A bang-up-to-date digital system for a few hundred quid.
I think I noticed what Paul is talking about when listening to Dire Straits "Brothers In Arms". No loudness war there. But dynamically accurate. I turn my amps up a lot for that listening experience.
The original CD of Brothers in Arms remains the best sounding music release I have ever heard. The dynamics are insane!! The funny thing is if it didn't sound so great I probably wouldn't like it nearly as much, but it's just such a wonderful experience to listen to that I keep coming back to it, and over time every track has grown on me. I could never get into the '96 remaster on Spotify before I got hold of the CD. It doesn't sound dreadful, but it's changed undeniably for the worse. I don't get the same rush when the drums kick in on So Far Away.
Anyone who doesn't understand dynamic range and why you want it needs to listen to that album on a real stereo. I'll admit, "Full time loud" does have it's place (such as listening to something on a phone or really lame portable speaker so you don't have to keep cranking it up only to have screaming loud parts come in) but it should be an effect you enable on those devices.... Due to that music has lost so much of the detail that makes it enjoyable on a good stereo.
I grew up with Infinity IRS Series V loudspeakers coupled with SAE first gen amps, components and high end turntables. Was interesting to hear vinyl records sound better than the early compact discs of the day as well.
Chris, interesting. I remember back in 1987 at the AudioFile in Bishops Stortford, I went in with the intention of upgrading my CD player. We used Brothers in Arms to compare a top of the range Mission PCM7000 CD/Naim Nait/Monitor Audio setup against Vinyl Linn Sondek LP12/Basik tone arm/AT-OC5/Naim Nait/Monitor Audio, and the LP12 was a Eureka moment for me. I walked out with the LP12!
It is also possible that the midrange dip was a token effort towards compensation for the Fletcher-Munson curve whereby human hearing drops off at bass and treble frequencies (human ear is most sensitive and can perceive the most number of volume levels at midrange, turn music down and the perception is that all the bass and high treble will disappear first... something that loudness compensation tried to correct, but never got right)
Do you think this also has to do with the resonant frequency of the human ear canal which is between 2k-4K hz? Frequencies in this range will sound louder to us even though they measure flat on the spectrum.
Wow. Great question - had never heard of the BBC dip and great answer - even though l'm neither techie nor an audiophile l actually managed to understand your answer and got to learn something cool. Thanks
I stumbled onto this. I am an audio-phile, and a guitarist. All guitarists know the British amps have the best tone. Of course the American Fender gear was standard, but since the 60's rock n roll has been all about the Vox and Marshall amps, and especially the Celestion speakers. This stuff worked with distortion by its own rules, and still sounded pleasing. As you said, they build with their ears, rather than just meters, I guess.
I like your videos but this is one of your best since I am sure that most of us have had experiences with BBC speakers. I loved them when I heard the first time.
Good question and a VERY good answer. I spent a lot of my career dealing with control systems (which are at their heart filters), and your explanation sounds (!) very reasonable. I suspect that the effect you're describing is caused by the fact that the two filters (the low pass filter and the high pass filter) are not tightly coupled together, and their interaction in the crossover range can be smoothed in amplitude or phase, but not both. Hail to the Dip!
Enjoy your videos. Never heard of the BBC dip but did study the Fletcher Munson curve many years ago as a flat sounding speaker through test equipment always sounded terrible to me . Never could afford good audio and had to build my own. Also never had the wealth of information the internet now offers and thank you for taking the time to share some of your wisdom.
Back In the early 70s I used to stack a small 2-way bookshelf speaker-system on top of a large 2-way bookshelf speaker-system, the cabinet on the bottom usually being a 12-in. mid-woofer with a 3in mid-tweeter, and on top of that would be a smaller two-way cabinet with an 8in mid-woofer and a 1-inch phenolic-Dome mid-tweeter, both were 8 ohm systems, but I would wire them together in series to present a 16-ohm load to the amp (which was meant for use with an 8-Ohm load). I had found that doing so would make them sound quite warm and Rich at low to medium volume levels rather than having to push more wattage into them to make them sound Full, which would have required playing them louder than I wanted to. A bonus of wiring the speakers this way was two-fold, not only did the amps run cooler (both tube AND transistor amps of around 20~50 WPC) but the sound was noticeably less harsh. I had found that many of the two-way systems of the time sounded a bit Brash in the upper mids, and this seemed to tame that problem as well.
and i thought the midrange harshness was a result of the cd recording not my original stacked Advent two way monitors. thanks Paul. looking forward to your new speaker product, saving my dollars .
hi, i replaced the woofercones twice in my wonderful advents but since1975 and repairing the crossovers and silver wire, they have been moved to a better location to make room for tekton paradragons, but just incase they ready to rock,thanks.
So I found a discussion on a forum that one member thought there was a drop of 1-2 db from 300 to 3khz. He said he would even like it better if there was a 2-3 db drop from 100 to 10khz, which is kind of how I set my EQ with my old speakers. Since that time I can't find a reciever with a tape loop to use the EQ with. But it would be interesting to try!
I have designed active crossover circuits for professional speaker systems. In these systems, the woofer has its own amplifier, so do the midrange (if any) and the tweeter. So a speaker enclosure is fed by 2 or 3 (or more) power amplifiers. The frequencies are separated at low power level (~1V RMS in a 602 ohm balanced line) then sent to the separate dedicated amplifiers. Butterworth filters were implemented using small op-amps with either first-order active filters or Sallen-Key second order. Something is very obvious at crossover frequency : even though the modules of the frequencies add up to unity if the filters are well-balanced, the phasing at this frequency is 45° per order, so this causes a peak in the compounded module of square root 2 (+3dB) per order. Hence the "harsh" peak at that frequency. We corrected the problem by spacing off the cutoff frequencies until the COMPLEX (imaginary plus real) waves add up to a module of 1, or by introducing an all-pass filter (that changes the phase but not the module) in one channel. And we got a real flat response, no dip, no peak, no harsh. Now, this is not so easy with passive components. Large capacitors and coils used in enclosures have wider tolerances than the precision resistors and small polypropylene capacitors used in small active filters. Furthermore, we could trim the frequency and phase response by adding a small value multi-turn potentiometer in series with the resistor either in the input or feedback network.
There are far too many factors involved to simply suggest that scooping out the midrange response around 2kHz will always offer an improvement in perceived quality. Would this be a good idea with a set of vintage Quad ESL 63's?
I have three pairs of speakers by the Swedish speaker designer Stig Carlsson, the SONAB OA6-type 1, the Oa-14 and the OA-116. Do you happen to know if these speakers also uses the BBC dip?
That's why 3-way speaker systems, with a genuine mid-range speaker between the tweeter and woofer, is the best. The crossover points are much less likely to cause issues like this -- you can slide off the woofer at the same rate you slide onto the midrange, and keep the resonances to a minimum.
@Shoegaze forever that's a pretty meaningless statement. _Any_ speaker design can sound bad in some frequency range, if it's badly built. The point is, our ears are most sensitive in the midrange, hence it's actually a bit silly to put a crossover right there. A dedicated midrange speaker means that you don't need to worry about those most critical frequencies. The price to pay is that you then need two cross-overs - at less critical spots, but that still doesn't mean they can be neglected. The good news is that they can both be less agressive than the single crossover in a 2-way speaker, because you never need to put any of the drivers close to its frequency limits. If you say 3-way speakers sound harsh then what you mean could be either that the upper cross-over has some problem, or perhaps that the speaker is _too linear_: if a mix was made on a 2-way monitor with a cheat like the BBC dip, it's possible that it actually overemphasizes these mids. That goes unnoticed on a typical hifi design, but would be revealed mercilessly by a good 3-way. That just means the mix is bad though, not the speaker. Taking out mids a bit can always be done with an equalizer; fixing a bad crossover is not so easy.
+leftaroundabout For two speakers of the same output a 3-way speaker and can sound hasher and more penetrating than a 2-way speaker tweeter (although, of course, this will depend on how well the speaker has been filtered or EQ'd) . This may be why many people find 2-way speakers more comfortable to listen to even though 3-way speakers should be superior. Another example of listening comfort vs audio accuracy is that many studio engineers prefer to listen to music for pleasure through audiophile speakers rather than monitors even though most studio monitors are technically superior.
Most 3 way systems have far more time alignment and phase problems than a good 2 way system. Most mass market speaker companies just use more speakers as a way to charge more money. Small high end companies that listen to their products can do amazing things with many driver speakers. Just an 6.5 " mid bas driver matched to the right tweeter in the best engineered enclosure can give you life like dynamic music.
3-way isn't "the best", it just has a better chance to manage directionality than 2-way does because a mid range driver is smaller than a typical bass driver. Changing the crossover to fix directionality problems (which is what the "BBC dip" kinda fixes) is the wrong way because it only works in a relatively hard room. The right way is to use suitably flexible drivers or use directional horns. Incidentally, the best speaker I've ever heard was either a 4 or a 5 way.
Anyone know what happened to the Phillips idea I read about ages ago? If i remember rightly...... It was based on negative feedback. Measure the physical displacement of the cone. If it doesn't correspond to the (digital) signal level drive it harder or less hard as appropriate. I forget their sampling rate, but 100kHz should do.
Very interesting. I had never thought about the crossover between woofer and tweeter giving rise to that problem, before. When you think about it, it's obvious, I suppose... Thanks.
The crossover region is always difficult as in nature there are no crossovers. Because one driver has a leading phase shift and the other a lagging shift at the crossover point it is very hard to get it right. When in the 2000s when I worked for a high end professional speaker manufacturer in LA, we did a lot of work on this but not having anything to do with a BBC dip. Rather we found several issues and one of them was helped by time alignment. Aligning the positions of the LF and HF and where applicable the MF drivers was a partial solution. The mid/high crossover was especially critical as the source of sound from the MF and LF drivers were not at the same point, this caused a big phase shift at the crossover point that especially would mess up the female singing voice. Eventually we developed a coaxial mid/high driver with a mechanical mixer for the sound from each driver. As well, the horn throat and flare we developed has so straight sides or square corners. There is no natural sound source with square corners and computer modelling of the power available by 1995 showed us the distortion inherent in horns with sharp angles. Going to all curved horns was a huge help. We also had to tweak every crossover design through extensive listening tests. We had a crossover peg board and boxes of inductors, resistors and capacitors and tweaked the crossovers in a large free field listening area. Then do full 360 degree acoustic dispersion bubbles with out speaker robot. Our reward for this was the top share in the US in the church market as our speakers were the very best there were for large venues and gospel music. Especially with female singers. We found that once the female singers came through well, everything else too care of itself. I cannot tell you how much Diana Krall and kd lang I listened to. with 24/95 SACD players. Babylon Sister was another track that I heard too many times. But while I was there, we had the best sounding large speakers and went off then to do digitally steerable columns.
I run a 4 way p.a., the cross over point to the top end is about 3.5-4khz, it can sound harsh, I lower those frequencies on the graphic, the result is a smoother sound, in keeping with the rest of the frequeny range, which I run flat, unless the acoustics of the venue demand it.
Thanks Paul. I've got a question. The Marantz pm6006 SE doesn't use a toroidal transformer , yet it has won a lot of awards for excellence, does it matter that it uses the t shape transformer ie would a toroidal transformer make it better?
Sounds a lot like the “V Curve” we used set our graphic equalizers to before they went extinct (leaving audiophiles to tune their systems with silly-money cables).
If you get the chance, would it be possible to see your new building and how you utilized it after moving into that large space you showed us? With the crossovers, my Dahlquist DQ10s were upgraded, but the interaction between the drivers and the room acoustics just did not translate as well as I had hoped. Another Stereophile with lots of experience in time and with all kinds of speakers and equipment showed me the value of the DBX DriveRack PA2, which is used for professionals setting up speakers for concerts, etc. He had his system set up as an active system with the DBX as the controller between the preamp and the amps. I got one and the difference and control is phenomenal . . . not only can I get the right balance between the drivers in their level, but I can fine tune even passive crossovers and tune the speakers to the room acoustics, which all work to bring these vintage speakers into the 21st Century -- one more step towards what my ears like and enjoy, which when it comes down to it is exactly what matters for each individual person.
Hi Paul, how about choosing a woofer with tweeter that match one another, naturally so they don’t require a crossover= clearer sound. I remember a British speaker, Epos 14, which had no crossover, just a small transistor for the tweeter.
That sounds familiar. I've been playing with 2 and 3 inch full range drivers and finding the resulting sound too harsh. I added a -6db notch filter at 1.5khz but I'm still not happy with the sound.
I am watching this with the sound coming from a pair of Tannoy SRM10B's. They have adjustable filters, whether that's a good idea or not. They're always flat here, but I do notice that the llittle frequency graph on the fronts of the speakers shows what the knobs on the filter do. And they seem to be causing a dip starting at about 2khz and up. Or the other way around. Maybe they tried to make the BBC dip an optional thing at Tannoy. I also have a pair of JPW Gold Monitor speakers that actually are ex BBC. They still have the stickers on them. They were probably used in some sort of mobile studio of sorts, or a cameravan. But I can assure you, they do not dip at all. They sound rather un-British. Nice little things though.
I see that martin Logan and B+W are going with smaller long throw woofers, bigger midranges, and super tweeters that pretty much self crossover. Theory is with the bigger midrange, the cross can go lower and not be in the critical listening area, and the super tweeters can use the same cross as the mids do. I know that I call the sound from my Martin Logan Motion 40's Magic.
That driver rotation synergy hypothesis is interesting, but what I posit is far more likely is that, while the monitors themselves were in fact flat in their response, human ears are most decidedly NOT in their pickup. The "dip" frequency range was what tipped me off, especially when you mentioned where Mr. Nudell put his "dip"... the ear itself, given its size, shape, and construction (taut, firm skin over cartilage reinforced by bone) has a peak in its' pickup range, the so-called "nails-on-chalkboard" threshold that occurs anywhere between 1.8-2.4kHz (depending on the person) and would be neatly sorted by a well-placed "dip" either in the cabs or via EQ in pre-amp.
It would be the amps that feed the speakers that regulated most of this problem back then, mics too have changed so much from back then. Speakers haven't changed all that much.
Don't forget phase distortion cancelation and reinforcement for the physical proximity between the woofer and tweeter. A lot of noises and clicky pops can jump from the sticky baffle. We tried three and four way systems. The more components the more phase cancelations. We reverse the phase of the mids to the tweeter/ woofer reversing the wave peaks... worse. Keeping the mid in box un attached to a signal dampened the cross over distortion, anti sound dampening. If you like Bose the baffle should not be a flat surface but faceted with no consistent reflective surface all out facing angles are transducers creating a scatter effect washing the distortions into a chorus effect tricking the ear into liking it. Massive notch outs up to -20db at regular frequencies of 44, 440, 2200, 4400 etc must accompany arrays to stop formant feedback in 4 way systems. with each db of gain you must cut the critical frequencies a bit more. Diminishing returns result as more signal energy is reduced through the EQ transients fall off first deadening the sound. Increasing gain increases formant feedback increasing eq decreased transient response deadening lively percussives. If you want good sound learn to play an instrument. You will never create an artistic sound with recreations.
This is my 1st of your videos. Nice info. I don't build electronics, but I still consider myself to be something of an 'audiophile,' (though I've never heard of the 'BBC dip!) I assumed, I suppose, that 3-way speaker systems - with proper crossover & band passing - would be pretty darn good, without needing to 'back off' on vocal ranges, so that idea is new to me. Say! I just noticed another one of your videos titled, "2-way or 3-way crossovers?" so I guess I'll have to check that one out. Thanks. :) Rikki Tikki.
so the point at which one speaker continues to vibrate and the other speaker starts to move is mysterious louder? hmm... wonder what would cause that... almost like there's more stuff moving.
*_This is why full range drivers always sound more natural, and one of the best, and fortunately still reasonably affordable, is the Tang Band W8-1772, which is an 8" driver with a huge following among many audiophiles._*
I had a friend who worked for AppleSound who did a lot of sound engineering work for the BBC. For many years now, when it comes to speakers, these masters of audio use Bose
One solution to the problem of getting drivers to work together in that critical upper range crossover region (2-3 kHz) is to develop an active speaker system, and then utilize a sharper slope of 4th order (24 dB/octave) or even 8th order (48 dB/octave) as I have done in my active system, which narrows the crossover region so sharply as to render the issues of virtually no consequence. In addition, since the system is an active design, a delay circuit can be placed in the mid range side of the crossover in order to allow for electronically aligning the phase of the output of both drivers to further minimize problems in this region.
High slope has problem, low slope has problem too, something to do with impulse response. High slope will sound harsh on crossover region, low slope will sound a little dull/muddy. I prefer 2nd order 12db/oct butterworth for high frequency and higher slope for low frequency.
Paul said it in the video there is no perfect solution. I think many people would argue that that closest we have come is with "Dual Concentric" driver configurations where the high and low frequency drivers are almost entirely physically aligned. Tannoy would argue that they are but even though what I saw was small it was a difference and needed to be corrected to eliminate that difference. The other thing mentioned is that due to their physical size, weight and shape differences the transients do not follow the same envelope even though each driver is trying to generate precisely the same thing. Larger/heavier drivers take longer to react. The other thing that I have seen is that some drivers sound horrible at certain frequencies. I love to use my Focal KRX2's as examples. Their cones are a aramid fibre (close cousin to "Kevlar") sandwhich design. Meaning they use layers of substances in order to create the final diaphram design. Kevlar and Aramid fibre both have a known issue of generating a lot of distortion at certain frequencies. This is actually common with all materials however where the distortion occurs and the type of distortion chances depending on what material is used. We have yet to invent a perfect one. When I set up the active crossover system in my car (Focal KRX are hi fi car audio speakers) I chose cross-over points well away from the points where the drivers would distort. PS there are no holes in the frequency response in my car. I time aligned for phase at the crossover points specifically. Most manufacturers only consider the nominal phase response of the drivers to determine phase settings. As most drivers have vastly different phase throughout their operating range I adjusted considering the phase response at the point where the drivers spl/frequency curves intersected. I also used a very gentle slope (I think 12 dB/Oct) which also generated less distortion making the listeners position smooth and ruler flat right up.
Have you done it yourself and it worked? I tried about 1980 to make two band IC stereo amplifier and it was mess comparing to good matched speaker with filters driven by exceptional mono PP tube amp . I gave it up. By the way - constant delay may be corrected by geometry while delay lines in my opinion may affect timbre of sound by altering the speakers input impedance. Of course it is in theory and who knows how it works.
Do it all of the time actually and it works exceptionally well. I am not sure if the fact that we are doing it in the digital realm matters but by eliminating that one source of distortion it creates a much cleaner and far more accurate sound. Of course the next issue is that the drivers can't actually function with the same attack times so there are going to be issues caused by that that can't be fixed that way.
These problems are exacerbated by small drivers with small wavefront baffles. That's why very large sound systems sound OK for acoustic material, and perfect for rock or jazz: any live performance of rock or jazz is heard through similar gear with similar characteristics. Want a perfect violin quartet in your living room? Hire a quartet. :o)
Thank you for that very interesting talk, Paul. It does raise the issue, of course, of why two-way speakers which inevitably have a crossover in that critical frequency range have become so popular? There is so much musical information in that critical 2 to 3 kHz band; why choose the crossover within it?. Three way speakers used to be more prevalent ‘back in the day’ but have largely gone though there are the Wharfedale EVOs that have a mid-range unit and, I believe, ATC. Thanks again for your very clear explanation of the origin of the BBC dip.
Also I do believe I've rolled off the point where the woofer and subwoofer crossover. I have an old car stereo crossover powered by a pc UPC running the sub here.
I tri-amp and have decent crossovers, but they do change phase at crossover. I bought an FPGA board (Snickerdoodle black) and am planning on trying some experimentation. Four-thousand-pole FIR filters should give me a pretty linear phase response between low and mid. I can reduce the number of poles on the upper filters between mid and high. The gotcha is that I will need 6 DACs, 12 if I go full balanced. I'm looking at some instrumentation R2R chips that are about $30 in quantity. ESS has some nice DACs, but I think they would be to difficult for me to program. Buffering the DAC output signal is the most difficult part of the system to design. If I actually manage to get it all working, I'll make it open source.
With amp you have I would expect very small problem with phase shifts at crossover. I would rather adjust individual distances from listener to each driver to equal their phase. .
If you want to know which R2R chips will go together with a nice BBC dip this may be of help: thetakeout.com/the-takeout-investigates-which-chip-for-which-dip-1822428459
I use to be into Car audio in the 90's and knew a guy that would always do real well in SQ competition. I found out his secret was to go listen to a orchestra and then go tune the system while ignoring all the writing on the crossovers. Ever since then I have noticed how a crappy sounding system can turn into a awesome system with the more you listen to it. I personally enjoy the colorful sound of a rock concert and IMO thats ok it's all about re-living the moment. The hardest issue I have is being able to recreate the rumble in the chest without being loud and it seems the Bass guitar is played thru concert speakers than any home system can duplicate.
