I was just about to write that the LCR circuits on the input are conjugate load matching. If the amp is a very low impedance source, ideally a voltage source, and you use very heavy cables then they will make virtually no difference. With a higher impedance source such as a tube amp they will make a difference and having them will make the tube amp sound more like a very low impedance amp. Ideally, if you had full input impedance compensation then the amp impedance would make no difference in the FR. KEF started using conjugate load matching some time after they got a reputation for having a difficult load impedance a few or more decades ago. I'm not sure if they were first. They make the load look resistive to the amp and so easier to drive. There is no good reason to remove them, or to use an expensive cap for them.
Last edited:
Thanks Mosquito, I think someone had estimated 1ms delay, which struck me as a figure that might have questionable impact. I am dead certain that a 158ms delay will be audible 😱 though I can not imagine why in the world a 158 ms delay would be needed 🙄 the I'm thinking that 1.58 ms is more likely to be the figure you calculated right? 😎 At any rate, I now have both speaker done the way I outlined and though I have not A/B-ed them yet or measured, 30 years of mixing and mastering tell me I am hearing the same or very, very nearly the same FR from both units which frankly rocks! If that turns out to be the case I will certainly rate these as very, very good mastering references within their bass limitations.
Last edited:
Well Peter, they are *gone* now and are I'm listening to them now on a half assed Sony receiver though the power amp only direct mode, I thought that would be a decent stress test of any problems with any relatively modern SS amp.. They are however going to be driven with a vintage Hafler and decent gauge cables. I am by nature a minimalist, not a purist.. That is I am of the audio school that anything in the signal path has to do something useful in the situation I am using it in or out it goes. Subjectively, these things are now sounding very, very flat pure and clean to me. I do not think that the one I had working "correctly before the "surgery' did at all. This may well be due to the 20uf recap only but as I said, if it ain't doing anything audible, out it goes. The tweeter protection is gone and those two LCR's are out. Now if I fry these fairly rare tweeter diaphragms (I have 3 spares) I'll probably go.....shi.....😀
Last edited:
I really have no preference for what you do just trying to explain the theory behind your crossover for you and others.
This is from a historical KEF paper:
Conjugate Loading Matching
"This technique developed by KEF for implementation into the design of the world famous Reference Series, is designed to make the amplifier’s job easier by simplifying the load it has to drive. In Model 103/4’s crossover network conjugation minimises the impedance variations normally presented to the amplifier in conventional circuits. This reduces the reactive load, ensuring optimum power transfer with minimum distortion. In addition to easier driving, conjugation brings with it another important advantage - a significant reduction in the deleterious effect on the speakers’ frequency response of the interaction between the speaker impedance and the combined effect of cable and amplifier impedance. This is of particular importance where long runs of speaker cable are required, and makes the choice of connecting cable less critical, ensuring a more consistent performance under all conditions. Further, each pair of model 103/4’s is assembled from drivers and crossovers which have been computer selected for identical sensitivity and response. Every pair of 103/4’s is matched to KEF Reference Series standards - ie less than 0.5dB difference in response between each speaker. "
This is from a historical KEF paper:
Conjugate Loading Matching
"This technique developed by KEF for implementation into the design of the world famous Reference Series, is designed to make the amplifier’s job easier by simplifying the load it has to drive. In Model 103/4’s crossover network conjugation minimises the impedance variations normally presented to the amplifier in conventional circuits. This reduces the reactive load, ensuring optimum power transfer with minimum distortion. In addition to easier driving, conjugation brings with it another important advantage - a significant reduction in the deleterious effect on the speakers’ frequency response of the interaction between the speaker impedance and the combined effect of cable and amplifier impedance. This is of particular importance where long runs of speaker cable are required, and makes the choice of connecting cable less critical, ensuring a more consistent performance under all conditions. Further, each pair of model 103/4’s is assembled from drivers and crossovers which have been computer selected for identical sensitivity and response. Every pair of 103/4’s is matched to KEF Reference Series standards - ie less than 0.5dB difference in response between each speaker. "
Last edited:
Hmmm...I wonder if the weird tweeter all pass filters were put in the circuit by B&W to correct phase errors introduced by the two LCR circuits that I have now removed?. If so then I would think it would be beneficial to bypass them now as well. For that matter would such LCR networks like the ones I snipped out introduce such phase errors?. Man! All I really wished to do was fix these things and here we go, I find the need to understand this in detail. So this is how it starts (-:
Given any system that does not have coaxial drivers, such as this one, the delay between the drivers is dependent on the vertical listening axis since the relative path delays change. The delay circuit there is designed to tilt the designed listening axis to front firing or whatever small tilt might have been put in to cover listening at a standing location. If you remove it the same FR will be found on some other, probably well off axis, point along the vertical listening axis.
In which case, I would *certainly* remove it and tilt or raise the speakers to ear level, since below 200 hz the FR is going to be even more room/wall/floor distance dependent anyway. Speakers should not be below ear level for my puposes. Sounds like another golden pickle award needs to go out.................
Last edited:
If this turns out to be the case, then I think I am beginning to understand the oddly humorous principles of starting a speaker company, kind of disheartening really, but does show the really fantastic value of de-bull-izing DIY groups like this one. Bear in mind, besides one person here, I have no idea who owns a speaker company. This is however how I think this goes:
1. The laws of physics can't be broken.
2. Find some common sense principle like "Point your speakers at your ears"
3. Either rename it and exploit it or find some way around it and........
4. Give that a new "scientific" name, start speaker company based on something like "point speakers toward ears"
I do not know whether this if happy or sad, but I do know it rings true, so I'll strike a major seventh chord here.
1. The laws of physics can't be broken.
2. Find some common sense principle like "Point your speakers at your ears"
3. Either rename it and exploit it or find some way around it and........
4. Give that a new "scientific" name, start speaker company based on something like "point speakers toward ears"
I do not know whether this if happy or sad, but I do know it rings true, so I'll strike a major seventh chord here.
Last edited:
What I'm saying is that if they could have placed the tweeter in a position to provide the correct delay they would not have needed that network, but since they're on a flat baffle it is needed. Remove it and what was the design axis where the proper frequency response is obtained will move from front firing to who knows where, far up, or far down depending on the layout and the amount of delay. The on axis frequency response will not be what was intended for this design.
I think you are getting the basics, but if not draw 2 drivers on a flat baffle, then there will be a point in front where the acoustic centers of the drivers have equal path length, move any angle away from this and one will get longer and the other shorter or vice versa depending on if you move above or below. Next, draw them coaxially and note that the path delay does not change.
Let me just point out to you that generally B&W knows what they are doing, you are not going to out smart them. Trust me, you will not! But have fun, as you are doing.
This design was from a time when the advanced companies were learning to use advanced methods to push their designs much closer to match ideal networks through network synthesis. This is a KEF paper but it covers many similar concepts:
http://www.kef.com/uploads/files/en/kef_topics/KEFTOPICS_vol4no2_crossover filters.pdf
I would often look at some of these computer optimized networks and see components that are not really needed, and you might note that the B&W designs became simpler later in time where they probably noticed the same thing.
I studied and "cloned" or built an improved clone of the B&W 801 in the 1990s starting with the very complex crossover network and later simplifying it.
http://www.kef.com/uploads/files/en/kef_topics/KEFTOPICS_vol4no2_crossover filters.pdf
I would often look at some of these computer optimized networks and see components that are not really needed, and you might note that the B&W designs became simpler later in time where they probably noticed the same thing.
I studied and "cloned" or built an improved clone of the B&W 801 in the 1990s starting with the very complex crossover network and later simplifying it.
If you are equidistant from both drivers, you should need no delay, is this correct? Therefore putting my drivers equidistant from my ears and defeating the delay is not trying to outsmart B&W, who designed my speakers to be nearer the floor and not as I am going to be doing, placing the drivers equidistant from my ears for purposes of mastering. Or am I missing something huge here? It would not be the first time. I am stressing that I will put all drivers equidistant as close as possible from my ears in this case, in lieu of not having, a more perfect and impossible point source, or is this incorrect?
Last edited:
It is generally assumed that the woofer radiates HF from the dust cap, or approximately that point, that is set a few inches back from the tweeter.
"If you are equidistant from both drivers, you should need no delay, is this correct? " That is what might be assumed when people see the delay circuit however it might also be compensating for phase offsets introduced by the acoustical response of the tweeter. I do not know for sure without simulating the design.
The classic B&W paper:
AES E-Library Computer-Aided Design of Loudspeaker Crossover Networks
Another classic:
AES E-Library Design of Optimized Loudspeaker Crossover Networks Using a Personal Computer
AES E-Library Computer-Aided Design of Loudspeaker Crossover Networks
Another classic:
AES E-Library Design of Optimized Loudspeaker Crossover Networks Using a Personal Computer
Well, heck Peter. This is the whole idea of what I'm trying to do, and why I called the complexity of that x-over "horrid". No wonder they simplified it as you did. As you and others have pointed out, when you get a new toy (umm like a computer) you tend to "use" it. Later you learn how to really use it. In my profession effect plugins are great example of that principle.. I was on the original Auto-Tune beta team, and featured in their ad copy too, you know what became of that! Now we are talking real horrors. I am certainly assuming that the guys @ B&W overused the new toys they had just like the rest of us do, then fixed it. Ergo, this thread.......and my now great sounding and possibly in the future even better sounding tertiary reference boxes. (-:
I think at deep inside my "esoteric" B&W tweeter, that is in its deepest heart of hearts, and soul of souls, and when no one else is looking, is is pretty much a black 1" dome tweeter with Ferrofluid damping and holding no secrets whatsoever from the world of physics. Please don't tell it, but is was used in several more pedestrian B&W speakers that lived their in trailers, probably had pit bulls tied up outside sported entirely zero exotic analog delay lines.
Therefore........could some one tell me how to physically bypass that silly delay line they put in so this unit could be a floor-standing speaker and no doubt show off its "unique" and special British royalty phase acumen in the showroom? I think I'm getting this now, thank all of you so much!
- Status
- Not open for further replies.