While fundamentally true (linear systems theory), this is not the whole story. For truly identical conditions you further need identical transfer functions (pulse responses) of the terminating impedance that each driver sees. Sure, these are second order effects but they can make a difference sometimes, even when individual and summed SPL don't show any significant difference.
Speakerman that last link still didn't come through.
I've done some preliminary sims on series v parallel of the type mentioned here, using both resistance and real data. The results between the two types are significantly similar in every way.
Now, Fried makes some points on his website in defense of series. Dissecting them from the way he has constructed them, his points seem to be...
1. That series will present a more resistive load and achieve more power transfer from an amp for better dynamic range.
2. That the above somehow causes transient response issues.
3. That the series network has a self correcting nature that causes drivers to behave.
Assuming my sims were not conclusive, I think it needs to be shown...
A. Why I can't just choose to use efficient drivers,
B. That phase distortion of this magnitude is an issue.
C. Why my amplifier can't hold its own, or at the least I might use Zobel correction across the system for the amp's sake, and
D. Why the 'correcting' nature of this circuit, under a driver error condition, doesn't amount to two wrongs trying to make a right,
before I commit to broadening the responsibility of the crossover.
I've done some preliminary sims on series v parallel of the type mentioned here, using both resistance and real data. The results between the two types are significantly similar in every way.
Now, Fried makes some points on his website in defense of series. Dissecting them from the way he has constructed them, his points seem to be...
1. That series will present a more resistive load and achieve more power transfer from an amp for better dynamic range.
2. That the above somehow causes transient response issues.
3. That the series network has a self correcting nature that causes drivers to behave.
Assuming my sims were not conclusive, I think it needs to be shown...
A. Why I can't just choose to use efficient drivers,
B. That phase distortion of this magnitude is an issue.
C. Why my amplifier can't hold its own, or at the least I might use Zobel correction across the system for the amp's sake, and
D. Why the 'correcting' nature of this circuit, under a driver error condition, doesn't amount to two wrongs trying to make a right,
before I commit to broadening the responsibility of the crossover.
This isn't true. Even the most uneducated, inexperienced listener is capable of hearing good vs bad sound.
I wasn't talking about extremes. I can't say that I 've managed to listen to
all the speakers out there, but I have never ever come across to hear really
bad sounding speakers, not a brand name product though.
AllenB the crossover plan I PM you shows the zoebel on the midbass. It is an interesting design concept.
When it comes to measuring speakers everyone is concerned about on axis. Better off axis response in series . ON axis is 2 dimensional. Off axis is 3 dimensional.
How many people listen on axis.
The same can be said of the sound of a live performance. Every listener is not hearing exactly the same frequencies being at different locations.
When it comes to measuring speakers everyone is concerned about on axis. Better off axis response in series . ON axis is 2 dimensional. Off axis is 3 dimensional.
How many people listen on axis.
The same can be said of the sound of a live performance. Every listener is not hearing exactly the same frequencies being at different locations.
Hi,
Simulator results often depend on what you are looking for.
They can be similar but very different in some subtle respects.
You can't see the acoustic differences between a series
and parallel 1st order x/o unless you know what to look
for and configure the simulator to illustrate them.
Consider series and parallel 1st order with resistances.
Now add the bass drivers inductance as an external component
to the bass unit resistor. Add a series notch filter to the tweeter
resistor representing its Fs impedance peak.
The bass series section filter immediately looks like 2nd order
series on the resistor*, the bass parallel section looks very
different. The tweeter peak will interact differently with the
two types, being less of a problem with the series arrangement.
The tweeter peak issue is also appropriate regarding x/o bass
units to midranges / fullrangers Ist order, series vs parallel,
and first order passive x/o's for subwoofers and satellites.
The differences are significant for some practical cases.
They should be easy to see in a simulator set up.
rgds, sreten.
*Why 1st order series is also known a quasi second order for bass.
The driver inductance helps the roll-off slope, whilst for parallel
it hinders the slope, forming a fixed attenuator with the inductor
at high frequencies, i.e. parallel doesn't roll-off continually.
Hi,
In some cases they don't, full stop, period, see post #107.
In fact for 1st order they hardly ever will sound the same.
IMO series is interesting for 1st order electrical and pretty much
nonsense at any higher electrical order. 2nd order parallel is
generally better than 2nd order series, for the same reasons
1st order series is generally better than 1st order parallel.
rgds, sreten.
Nope, that's bull.
This is the same drivel you've been pushing for at least the last few years.
Later,
Wolf
Well in a sense the voice coil inductance does determine the transient response in a round about kind of way. Increasing the inductance will roll off the drivers top end and this will inherently limit the extension of the driver. This in turn will prevent it from responding in any meaningful way to higher frequency transients. Now higher frequency transients naturally have a faster rise time then lower ones, so I suppose you could just about stretch this to mean 'the voice coil inductance determines the transient response'. I mean the extra inductance isn't actually slowing any part of the signal down as per say, but it will have the overall effect of reducing the drivers ability to respond to higher frequencies and there reduce its ability to respond to signals with faster transient rise time.
This is of course not a particularly useful way of looking at it, but you can see how easy it is to put the wrong spin on something.
This is of course not a particularly useful way of looking at it, but you can see how easy it is to put the wrong spin on something.
I'm sorry I haven't seen it as yet.
A crossover should be designed considering all axes.
Trying to follow, relative on/off axis sound cannot be modified with a crossover, the difference between them is fixed acoustically. As for focussing a crossover towards some directionally balanced target, this is possible with different crossover types and virtually any target.
Yes. I thought that the idea of presenting a resistive load to an amp was not subtle, and had to look significant to be treated as such if at all.
On the other hand the idea of the crossover 'correcting' dynamic errors won't show up on a conventional crossover sim, but would you leave that to a crossover? What about the varying of Le dependent on excursion? Or the non-linearity of ferrite? I'd rather isolate the driver and treat these issues than to burden the tweeter with them.
I made a comparison by starting with dummy flat impedance data and driver data, then substituting real woofer data in both the series and parallel cases.
First plot is parallel. Blue is woofer mag/phase before, orange after using real data. Pink shows Ztotal mag/phase for later comparison.
Second plot is series. Blue is woofer mag/phase before, and orange after, then red is tweeter mag/phase before, and cyan after substituting only real woofer impedance data. Ztotal doesn't appear very different.
Not that I see any of this as particularly relevant. The responses shown could be achieved with either type and any interaction is incidental. The point really is....is it the desired response?
This makes me think of another thing Fried wrote regarding the reduction of Doppler distortion due to a greater driver overlap. Such an overlap could be achieved either way and I wondered whether it had been left to chance?
Attachments
Hey wolf-teeth I have crossover plans and driver specs to prove it from the company ha ha.
Why don't you tell Ken Kantor a thing or two on driver design Driver bandwith also critical factor. The wider the bandwith the faster the transient response. Voice coil mh also involved.
Why don't you argue with Dr.Raines the new owner of the company and tell him a thing or two.
Why don't you tell Ken Kantor a thing or two on driver design Driver bandwith also critical factor. The wider the bandwith the faster the transient response. Voice coil mh also involved.
Why don't you argue with Dr.Raines the new owner of the company and tell him a thing or two.
Last edited:
http://www.diy-audio.narod.ru/litr/WooferSpeed.pdf
Read it wolfteeth. It might be to complicated for you figure out how to open the link.
Read it wolfteeth. It might be to complicated for you figure out how to open the link.
Last edited:
I value the advice of people who have their PHD. in Electro Acoustic Engineering over the forum users on this site..
Allen B driver separation distance of voice coil centers relative to crossover frequency.
Actually designing a acoustically correct loudspeaker is far easier then harvesting a whitetail deer for crop damage. It would take a big game hunter to understand this.
Actually designing a acoustically correct loudspeaker is far easier then harvesting a whitetail deer for crop damage. It would take a big game hunter to understand this.
I respect AllenB . He is willing to have theoretical debate as a gentleman. Doing series parallel listening comparisons . Series sound more musical.
I think very little of people who insult others behind their computer.
If this is the best they can do they probably have problems in everyday life with other people.
I feel sorry for them.
I think very little of people who insult others behind their computer.
If this is the best they can do they probably have problems in everyday life with other people.
I feel sorry for them.
The problem is, why does it sound more musical.
Show me some measurements. The step response, the frequency response, acoustic phase tracking, etc. Show me something that series does, that parallel doesn't.
Otherwise it's just two different designs, one of which just happens to seem to work better.
speakerman19422 you're missing the point about a drivers transient response. Yes, in absolute terms the voice coil inductance or band width of the driver affects its transient response, but once you've put a low-pass on it you've altered this at a fundamental level. The transient response, or the impulse response if you like, follows what the driver + electrical filter now dictates, however, the driver will still respond just as 'quickly' as it did to frequencies below where the filter operates.
In absolute terms the driver will still respond just as quickly, even into the stop band, with the filter in place, it's just that with the filter there it's response will be much lower in amplitude. In other words the signal is still being reproduced accurately and just as 'quickly' it's just at a much reduced level.
In absolute terms the driver will still respond just as quickly, even into the stop band, with the filter in place, it's just that with the filter there it's response will be much lower in amplitude. In other words the signal is still being reproduced accurately and just as 'quickly' it's just at a much reduced level.
Why does certain design sound musical and/or have less fatigue, that's the "problem" that I am aware of, but can never solve till this moment.
Using amplifier as an analogy, we know how even order distortion differ from odd order distortion. We also know about distortion spectrum pattern. I guess the exact same thing exist in loudspeaker.
I can make a crossover that with the drivers "measures" good and sounds subjectively good. And I can make many other versions that measure and sound subjectively equal. But only experienced ears that can detect if one is "superior" than the others.
The more complex the crossover, the more difficult to achieve this "sweet spot". That's the reason why I prefer to avoid complex circuit as much as I can. This is simply about how much time I want to spend in looking into simple circuit possibility, to see if the basic problems can be solved without introducing complexity.
Using amplifier as an analogy (again), we know about IMD. In amplifier design we can feed front stage signal back to earlier amplifying stage. Most of the time these signals have different phase when "superimposed". I believe this phase issues have serious ill effects to sound.
With amplifier, I avoid global feedback as much as I can. With crossover, I don't bypass the main HPF with RC in order to level up the higher end frequencies. There are many don'ts in amplifier or crossover design that I follow.
I have to say that in those basic measurements I cannot see a hint or something that correlate with what ears can perceive. Acoustic phase tracking is probably the closest. But still cannot explain.
To my knowledge this is 100% the case.
- Status
- Not open for further replies.