In its presentation, musical instruments are full of distortion and hard clipping. So you may want an amplifying device that is good at presenting these distortion and clipping.
Also, you want an efficient amplifier for this purpose. I think, if power is not an issue (with its associated heat-sink size etc), a high power class-A will achieve a better result than use of 4558 opamps or 12AX7 tubes in the front end.
Speakerman, is it possible that you see series crossovers as somehow different to parallel? Above the crossover frequency the woofer itself has high impedance and is out of circuit. Put your hand over it and look at the circuit.
It is returned to the load through the source. At the risk of splitting hairs a capacitor will not store energy as such so much that it will delay a signal causing a phase shift. This issue is often given more concern than I think it deserves.
I'm not sure it is. The point of a crossover is to alter from flat, the applied voltage to change the response. Using resistance to balance a driver's impedance is somewhat irrelevant as there is no compulsion that a set of drivers should receive a symmetrical contribution from the crossover. It is not even a matter of matching drivers precisely. Woofers and tweeters are subjected to different acoustic conditions and the acoustic needs of the upper and lower bands are different, beyond electrical and driver issues.
This statement needs to be more specific. Not all components contribute significant colourations, and some of the problems they are fixing were significant.
Fair enough. They distort gracefully when pushed. I like them in hifi despite their distortion. I'm not sure I can see how this relates to the topic?
Hi,
Series / parallel x/o's don't have the same transfer
function with real drivers, that is the whole point.
rgds, sreten.
Series / parallel x/o's don't have the same transfer
function with real drivers, that is the whole point.
rgds, sreten.
See my earlier post- yes, if you just simply make a conjugate, they won't be the same. But for any arbitrary pair of transfer functions, you can design a parallel or series network to hit that target.
My thoughts
- the idea that less components = less colorations is not at all true.... quite the contrary in the real world with real hearing. There's a lot of claims that are consistently made here that are just unscientific nonsense. It takes a complex crossover to have a coherent sounding multi-way speaker, as mind blowing as that may sound. All these "1st order slopes" are just a mess that do more harm than good, all for the preservation of things that aren't necessarily audible.
- The electrical topology isn't what matters, it's the results. Something no one in this thread has shown. Where are the phase or reverse null plots? Where are the frequency response plots? Where are the Z-charts? Just a bunch of opinions based on half-baked experimenting.
- the idea that less components = less colorations is not at all true.... quite the contrary in the real world with real hearing. There's a lot of claims that are consistently made here that are just unscientific nonsense. It takes a complex crossover to have a coherent sounding multi-way speaker, as mind blowing as that may sound. All these "1st order slopes" are just a mess that do more harm than good, all for the preservation of things that aren't necessarily audible.
- The electrical topology isn't what matters, it's the results. Something no one in this thread has shown. Where are the phase or reverse null plots? Where are the frequency response plots? Where are the Z-charts? Just a bunch of opinions based on half-baked experimenting.
Allen B have you carefully read all my posts on this topic?
The drivers need to have equal impedance at the crossover frequency for it to be a constant voltage circuit.. Woofers impedance increases much higher then a tweeters at the crossover frequency.
.
In order to compensate for the woofers, resistance needs to be added to the tweeter. Since most tweeters have a higher efficiency then a woofer they need the voltage reduced.
The goal is to have the spl be equal from both drivers to achieve a flat frequency response above and below the crossover frequency.. It all comes down to the drivers used for the Zoebel. I know one company that claims it is not needed. I have yet to see a driver that does need one.
Choosing the right Zeta values are critical .
I am sure you know the many examples.
The drivers need to have equal impedance at the crossover frequency for it to be a constant voltage circuit.. Woofers impedance increases much higher then a tweeters at the crossover frequency.
.
In order to compensate for the woofers, resistance needs to be added to the tweeter. Since most tweeters have a higher efficiency then a woofer they need the voltage reduced.
The goal is to have the spl be equal from both drivers to achieve a flat frequency response above and below the crossover frequency.. It all comes down to the drivers used for the Zoebel. I know one company that claims it is not needed. I have yet to see a driver that does need one.
Choosing the right Zeta values are critical .
I am sure you know the many examples.
I think many people are not reading the entire thread before they post.
The final result is listening which is subjective. An Audiologist can interpret an audiogram at what spl a person hears test signals.
A person could have ruler flat hearing and still be a poor judge of speakers
How many have done A and B comparisons? Spl are higher in series. Dynamics are more dynamic. What speakers with series circuits have others listened to?
.
I can't say how something is going to taste until I eat it. Something might look good and taste horrible.
The final result is listening which is subjective. An Audiologist can interpret an audiogram at what spl a person hears test signals.
A person could have ruler flat hearing and still be a poor judge of speakers
How many have done A and B comparisons? Spl are higher in series. Dynamics are more dynamic. What speakers with series circuits have others listened to?
.
I can't say how something is going to taste until I eat it. Something might look good and taste horrible.
Less components = Less design hassles
Less components mean less issues to be solved by the designer. "Coloration" is just not the right word here, but I believe the idea is correct.
Complex filter (high order slopes) means more energy lost in the components (means less sonic) and worse is the problems associated with phase issues.
This is almost the same with using THD to measure amplifier quality. You can make more complex circuit (by adding feedbacks) to achieve lower THD but do they sound better? Is this non-THD thing inaudible?
I will choose the simplest possible filter (or amplifier) that is "possible". The "possible" word here is subjective of course. Most if not all of the time, to me this "1st order slopes" are just a mess that do more harm than good. Then start with a good driver that will work subjectively well even without filter. If a fullrange design is still acceptable for some (they do more good than harm), then how come a first order is not possible?
I have only one speaker that "works" with first order filter (the woofer uses strontium magnet). The reason is that I have a unique tweeter that can match the woofer with first order filter.
The key is the "matching", not "low order filter". The low order filter really do a bit of harm, but the matching do more good. So all in all, the first order is still subjectively better than the second order one (But I believe there is a chance for the second order one to be better).
Less components mean less issues to be solved by the designer. "Coloration" is just not the right word here, but I believe the idea is correct.
Complex filter (high order slopes) means more energy lost in the components (means less sonic) and worse is the problems associated with phase issues.
This is almost the same with using THD to measure amplifier quality. You can make more complex circuit (by adding feedbacks) to achieve lower THD but do they sound better? Is this non-THD thing inaudible?
I will choose the simplest possible filter (or amplifier) that is "possible". The "possible" word here is subjective of course. Most if not all of the time, to me this "1st order slopes" are just a mess that do more harm than good. Then start with a good driver that will work subjectively well even without filter. If a fullrange design is still acceptable for some (they do more good than harm), then how come a first order is not possible?
I have only one speaker that "works" with first order filter (the woofer uses strontium magnet). The reason is that I have a unique tweeter that can match the woofer with first order filter.
The key is the "matching", not "low order filter". The low order filter really do a bit of harm, but the matching do more good. So all in all, the first order is still subjectively better than the second order one (But I believe there is a chance for the second order one to be better).
If SPL is everything, then Fostex design is always better than Dynaudio speakers. Many loudspeakers require special amplifiers to drive them properly.
I believe in simple circuits. For every driver combo that I have used I always try the simpler circuits first, including series filter (but never more than 2nd order with series). Many times I have to go with complex one to achieve an acceptable sound to my own ears.
Yes, something may look good in measurement (especially on simulation software) but sound bad after you build it. For any serious speaker builder, I recommend to have a full set of components for prototyping, because you will probably never know how it will sound before you hear it.
BUT...
You can't say that something you eat tastes good until you know that you have no problems with your tongues. Many people cannot hear, so for those, please be open-minded, instead of thinking that "if I can't hear anything then everyone else can't".
Last edited:
Speakerman, say we did a reasonably thorough job of flattening the impedances. We dealt with the resonances, the voice coil inductances, and we threw in an L-Pad to reduce the tweeter level whilst retaining the desired impedance. We have now used 12 components per channel and the crossover is essentially looking at pure resistance. A first order series crossover still will not necessarily achieve the desired goals.
[I have yet to see a driver that does need one] was incorrect
I incorrectly stated above quote. I have yet to see a driver that does not need a zoebel is what I meant to say. I can't type as fast as I like to..
I can't state the Zeta values being used because of possible legal ramifications.
I incorrectly stated above quote. I have yet to see a driver that does not need a zoebel is what I meant to say. I can't type as fast as I like to..
I can't state the Zeta values being used because of possible legal ramifications.
Last edited:
They do measure flat. No level pad. the number of components is not twelve. L1,C1 R1,R2 and C and R in Zoebel = 6 and L1 andd C1 for subwoofer=8
The sloped baffle pyramid was an extention of what i was doing, which was to pyramid the sides (my suggestion for clients buying and building Model H kits -- or model H clones using B139s from the KEF constructor sequence). He used those in between the rectangular LS3/5A sized boxes in the H, and the later designs that added a sloped baffle (the M had a rectangular box with a sloped baffle).
The possibility exists that Bud got the idea from me, as he was privy to my drawings before he ever came out with this shape sattelite.
dave
BTW these, and the later Q & R boxes had parallel XOs.
Sreten,
The basics of the receipe for the MTM with series XO i mentioned earlier. Drivers were very carefully choosen, baffle width (XO point) carefully considered (including what the drivers impedance looked like, and how close i could get the drivers), and, particularily on the mid-tweeter, suppresion of the resonant peak acoutically. All these considerations made it possible to consider a low order XO with minimal components.
dave
I assume here you are meaning if less components need to be used then there must have been less issues to solve initially with the raw drivers individual behaviour. More components do not create issues, they solve issues.
This is completely untrue a passive low or high-pass crossover works by increasing the impedance of the specific network + driver and thus reduces the current drawn through the circuit and hence reduces the power dissipated. The steepness of the electrical filter simply indicates how quickly the impedance rises and how quickly the power is reduced. A steeper filter does not lose more of anything unless we're talking about the intrinsic loss due to an inductors DC resistance.
Whilst on the issue, any resistive energy loss simply = a reduction in output at the driver and this is entirely within the control of any competent designers capabilities. You're not losing any critical fidelity by these losses either, not unless the designer hasn't accounted for them and there is now a depression in the frequency response because the mid driver is now playing quieter then it should be.
This is in no way analogous to amplifier design. By definition a low THD amplifier and an ultra low THD amplifier will both have distortion that is already below the threshold of hearing anyway. So technically speaking you're not achieving anything except an exercise in engineering by improving upon what was already perfectly acceptable. Can you hear it? Most likely not.
In loudspeaker terms though this thread seems more skewed towards lets reduce the component count because it makes it sound better. This will inherently make the loudspeaker perform worse in measurements, but it will sound better because the crossover is simpler. Rubbish. From an engineering point of view, the amplifiers went from good to better, but the loudspeakers in this case went from good to bad because reducing the component count also meant that we ended up with issues, bad issues.
Lets take a standard well designed loudspeaker though, it's well designed but is hard to drive. This makes my amplifiers job much harder then it needs to be and as a result the amplifiers linearity is reduced because it's having to work very hard. Now I decide to add a bunch of components extra to the loudspeakers, these have no impact on the frequency response, but counter balance what makes the loudspeakers difficult to drive and gives my amplifier a much easier task. This makes it sound better because now the amplifier doesn't have to work as hard and it is now operating more like the perfect wire with gain. Extra components doing nothing but good.
Your missing the point. Most drivers do not work well when used with first order crossovers. Your analogy is like saying I don't like getting my feet wet so I'll never go out when it's raining. Rather then saying lets just wear rubber boots and go out side. There are designs where first order crossovers are appropriate but these are few and far between. Lots of designers however would say that even in those situations that you could possibly improve upon the situation by using a more complex filter because it reduces some of the issues that first order crossovers have. Sreten gave a good example of where you can use a first order crossover, in integrating a full range driver to a bass driver in a FAST system. Providing this protects the full range driver adequately against over excursion for your SPL demands, it elegantly handles baffle step compensation.
You are correct here in saying that matching is what is important. One matches the necessary filters to the drivers so that you end up with a well integrated system. Note though that if a more complex filter is necessary by design, that simplifying the filter, just for simplicities sake, will not result in anything sounding better, it will sound worse because you've neglected to correct for things that those extra components were there for originally.
Simpler crossovers are loved though by everybody and every engineer on the planet for one simple reason - they are cheaper. If a simpler crossover can successfully be used and arrive at the target technical goals that the engineer set out the achieve, then you can bet your bottom dollar that they will go with the simpler crossover.
Yes, complexity is in fact a result of an effort to solve issues. But there are always issues that are beyond designer's capability to solve. Such issues may arise from complexity. Whether someone is capable of solving the issues, that is not the point of discussion.
What else can it be? That's what I had in mind. But I don't talk (as usual) based on theory but observation. If I had to think about theory, well, I still can see other things. But because I don't like to talk from theory alone, I don't want to mention it.
Yes, SPL is not everything. Energy loss is not a big issue. It's the phase.
***Oh yes, I can hear it. Do you think those amp guys just wasting their time improving the THD? The THD thing is good. It is just what you do to make it low that creates another compromise. At least, refer to what NP writes on the F5 Turbo, regarding the compromises in the THD.
From engineering point of view, speaker/amp design is about balancing compromises. For certain people it is about knowing how to design, but for somebody else it is about knowing the compromises.
You can say like that because you assumed that I don't know better than you do 😉 But if you put that aside nothing is in disagreement. Fullrangers are okay without filter, why shouldn't it be okay with first order (as in the FAST)?
"There are designs where first order crossovers are appropriate but these are few and far between."
Our job is to find it even if it is few and "far between". That's my message (find out if simpler circuit is "possible" before going with the complex one). This is of course based on an assumption that simple circuit has something to offer. Have you ever come to a decision whether you have to use a notch filter or not? Or you just add it because you know that it will make things better? Me, I try many kinds of notch filters to find one with less horrible effect to sound.
Just to let you know that many expensive speakers (with custom ordered woofers) use simple crossover.
I will eventually get back to everyone. I have this to deal with. I am in a lawsuit over wildlife management. Look at 322c7 [govern the use of recorded calls]
http://ujsportal.pacourts.us/DocketSheets/CommonwealthCourtReport.aspx?docketNumber=389 MD 2006
Powers and duties of commission - 34 Pa. Cons. Stat. § 322 - Pennsylvania Attorney Resources - Pennsylvania Laws
might be off topic, everything is off topic, everyone has to read the whole thread and think about everything before posting
http://ujsportal.pacourts.us/DocketSheets/CommonwealthCourtReport.aspx?docketNumber=389 MD 2006
Powers and duties of commission - 34 Pa. Cons. Stat. § 322 - Pennsylvania Attorney Resources - Pennsylvania Laws
might be off topic, everything is off topic, everyone has to read the whole thread and think about everything before posting
Last edited:
- Status
- Not open for further replies.