I've been running a couple of well behaved drivers (Peerless 850146 & SEAS metal cone tweeter) with just a single cap on the tweeter, - last week tried a series 2nd order on one channel, and while there were subtle differences in the sound, they weren't great enough for me to want to wire up the 2nd channel.
I'll probably end up using 2nd order parallel for the tweeter, simply to protect it if I get a bit heavy handed on the vol control one day, no other reason.
Pete McK
I'll probably end up using 2nd order parallel for the tweeter, simply to protect it if I get a bit heavy handed on the vol control one day, no other reason.
Pete McK
I wonder too
But sometimes it´s just it. Especially in diy.
I won´t go for "the simpler is always better" thing.
But if things get complicated, there is more room for errors.
So maybe the high order crossover was just incorrectly implemented.
Besides that, we have done some measurements lately with a first order highpass and a time delayed tweeter. It was mounted few cm beyond the woofer.
Compared to a 4. order L-R filter there was a clear advantage in phase and impulse response.
There seems something more in 6dB filters than just intermodulation and distortion.
Regards
Jürgen
I think that 2nd order is a good compromise between low frequency rejection and complexity (with tweeters in mind). On the other hand, higher order filters are harder to get right because they produce narrower dips or bumps in the frquency (and polar) response when they are not aligned properly. Also, 2nd and higher order passive filters rely more on a flat load impedance in order to get a suitable frequency response, while almost no driver shows such a flat impedance plot without some compensation.
I have built my own mid-high horn loaded system, and I'm very happy with the performance that I got when I finally decided to go custom 24db/oct active (after trying 24dB/oct passive and lower orders). A lot of tweaing in filter response and 6 first order phase shifters (per channel) were required in order to finally achieve phase coherence and proper summing (which causes strong cancellation when tweeter polarity is reversed) between the outputs of the horns in the crossover region (around 1600Hz). It was not an easy task indeed.
I strongly recommend custom active filters because frequency response (poles and Qs for each side) has to be tailored to the requirements of each driver, and generic filters just don't allow to do that. Also, in my experience, when horns or big or complex drivers are involved, phase shifters have proven to be essential in order to match the group delay of both drivers by adding suitable lag to the leading one. Note that the phase shift characteristic produced by an analog all-pass filter is different from that produced by a plain digital delay unit.
I have built my own mid-high horn loaded system, and I'm very happy with the performance that I got when I finally decided to go custom 24db/oct active (after trying 24dB/oct passive and lower orders). A lot of tweaing in filter response and 6 first order phase shifters (per channel) were required in order to finally achieve phase coherence and proper summing (which causes strong cancellation when tweeter polarity is reversed) between the outputs of the horns in the crossover region (around 1600Hz). It was not an easy task indeed.
I strongly recommend custom active filters because frequency response (poles and Qs for each side) has to be tailored to the requirements of each driver, and generic filters just don't allow to do that. Also, in my experience, when horns or big or complex drivers are involved, phase shifters have proven to be essential in order to match the group delay of both drivers by adding suitable lag to the leading one. Note that the phase shift characteristic produced by an analog all-pass filter is different from that produced by a plain digital delay unit.
Reji
Did you measure the frequency response with a your new simple crossover? As Eva suggested in an earlier post you are probably getting break-up of your bass driver which can have a nice sound. As an example of this I was lucky enough to go to a concert of a Dutch group Focus who had international hits with songs Hocus-pocus and Sylvia. I was so impressed I went out and bought their LP record.
When I played the record the sound was too bland. I was using a JVC amp with graphic equaliser at the time and restored the concert hall sound by boosting 1KHz up by 12dB Now Jan Akerman's lead guitar had bite!
What we are after is fidelity however, even if the recording engineer got it wrong! So I suggest your "better" in the title is not "better" just "different"
Did you measure the frequency response with a your new simple crossover? As Eva suggested in an earlier post you are probably getting break-up of your bass driver which can have a nice sound. As an example of this I was lucky enough to go to a concert of a Dutch group Focus who had international hits with songs Hocus-pocus and Sylvia. I was so impressed I went out and bought their LP record.
When I played the record the sound was too bland. I was using a JVC amp with graphic equaliser at the time and restored the concert hall sound by boosting 1KHz up by 12dB Now Jan Akerman's lead guitar had bite!
What we are after is fidelity however, even if the recording engineer got it wrong! So I suggest your "better" in the title is not "better" just "different"
Hi,
Mr Kirchner allowed me to show some pictures from his still unpublished paper "LAUTSPRECHER MESSTECHNIK".
The paper is meant as an introduction to speaker measurement discussing some signals and techniques.
It also shows some "real world" measurements on speakers with first order crossovers and 4. order crossovers.
Although the differences between crossovers are already known, discussion is usually based on theory or simulated filters.
So something "real" maybe of interest.
He compares one of his own designs with a displaced tweeter and 6dB filter against one with flat baffle and 24dB filter. The drivers are equal.
Regards
Jürgen
Mr Kirchner allowed me to show some pictures from his still unpublished paper "LAUTSPRECHER MESSTECHNIK".
The paper is meant as an introduction to speaker measurement discussing some signals and techniques.
It also shows some "real world" measurements on speakers with first order crossovers and 4. order crossovers.
Although the differences between crossovers are already known, discussion is usually based on theory or simulated filters.
So something "real" maybe of interest.
He compares one of his own designs with a displaced tweeter and 6dB filter against one with flat baffle and 24dB filter. The drivers are equal.
Regards
Jürgen
first: frequency response shows nothing special
phase response is different:
the speaker with high order crossover showed a step in phase, just inside a very sensible region
An externally hosted image should be here but it was not working when we last tested it.
phase response is different:
An externally hosted image should be here but it was not working when we last tested it.
the speaker with high order crossover showed a step in phase, just inside a very sensible region
square wave response:
step response:
while the box with first order network showed a somewhat clean step response, the other let us distuinguish two responses, first the tweeter and after some delay the woofer.
An externally hosted image should be here but it was not working when we last tested it.
step response:
An externally hosted image should be here but it was not working when we last tested it.
while the box with first order network showed a somewhat clean step response, the other let us distuinguish two responses, first the tweeter and after some delay the woofer.
responding to the OP ... that 1st order x-overs are minimum phase / transient perfect, they pass a square wave, plays a big part in their performance.
Have you read either directly from [or in re-print form from another source] that the co-author to the L-R x-over paper that started it all off, himself, states that if you use a L-R x-over without proper time alignment of the drivers, you defeat the purpose of the x-over?
Enormous amount of useful info on his site. Not easy to find it all as it's not all linked from the left menu tho.
Have you read either directly from [or in re-print form from another source] that the co-author to the L-R x-over paper that started it all off, himself, states that if you use a L-R x-over without proper time alignment of the drivers, you defeat the purpose of the x-over?
4- Delay Correction http://www.linkwitzlab.com/filters.htm#4
Enormous amount of useful info on his site. Not easy to find it all as it's not all linked from the left menu tho.
electrical measurements with sine burst
the input signal is black, the blue line is measured inside the forth order lowpass
the red one is at the output
there is some delay between input and output and one could see some overshoot. The delay could be compensated by speaker displacement , the overshoot could not. It adds some additional distortion.
The highpass section showed a similar behaviour:
At the first order crossover these effects are hardly measurable.
An externally hosted image should be here but it was not working when we last tested it.
the input signal is black, the blue line is measured inside the forth order lowpass
the red one is at the output
there is some delay between input and output and one could see some overshoot. The delay could be compensated by speaker displacement , the overshoot could not. It adds some additional distortion.
The highpass section showed a similar behaviour:
An externally hosted image should be here but it was not working when we last tested it.
At the first order crossover these effects are hardly measurable.
juergenk said:
Indeed, the measurements of the 24dB/oct filter clearly show a sharp +3dB peak and a -3dB dip right in the middle of the highly sensitive crossover region, which means that the drivers are desperately screaming "PLEASE TIME ALIGN US!!". I had similar issues in my horn system and I had to use several all-pass filters to match the group delay of both drivers. First order filters are much less sensitive to misalignment (more fool-proof) due to the gentle overlapping and the much wider phase transitions, but they have their own set of drawbacks.
juergenk said:
Those effects are a consequence of the own group delay of the 4th order filter and have been demonstrated to be completely unnoticeable to our hearing system. If you don't believe me do the *blind* test by yourself, ask some friend to use a switch to let you compare the *electrical* sum of the outputs of the 4th filter against the input signal. You will never be able to tell which one is which unless you go to very high order filters (well above 24dB/oct so that the group delay becomes huge enough to defeat the short-delay-ignoring system of out brains). The filter has obviously to be set up in such a way that the electrical sum of the outpus produces back a flat frequency response with respect to the input (state variable topology allows to achieve that in stereo with just 5 dual op-amps).
BTW: The hardest part of all is to achieve optimum acoustical sum, because each driver exhibits its own unexpected group delays associated with every peak and dip in its frequency response and with both out of band roll-offs. Each driver has to be effectively considered as if it had several high-order filters already built-in. Also, that's why some offset is added to the tweeter of the 6dB/oct filter system discussed in your figures, despite the low order.
> I won´t go for "the simpler is always better" thing.
Me too. That’s why I am trying to believe that complex crossover circuits should have their advantages afterall..
> But if things get complicated, there is more room for errors.
> higher order filters are harder to get right because they produce narrower dips or bumps in the frquency (and polar) response when they are not aligned properly.
> A lot of tweaking in filter response and 6 first order phase shifters (per channel) were required in order to finally achieve phase coherence and proper summing (which causes strong cancellation when tweeter polarity is reversed) between the outputs of the horns in the crossover region (around 1600Hz). It was not an easy task indeed.
So the bottomline is, one needs to know his electronics and acoustics well, to get good results from higher order crossovers. As Eva says achieving phase coherence and proper summing may need extra circuitry.
consort_ee_um, my first order filter was different, but it was better too..period (I didn’t measure its frequency response though). Anyway these discussions encourage me to learn more, so that I can design custom higher order filters that perform well..
By the way, what you guys think about some very popular websites which lure the readers into discarding their simple crossovers and embracing higher order actives, without mentioning anything about the above issues? Why do they hide these things? Maybe I am a victim, and there are others too, like Dan..
reji
Me too. That’s why I am trying to believe that complex crossover circuits should have their advantages afterall..
> But if things get complicated, there is more room for errors.
> higher order filters are harder to get right because they produce narrower dips or bumps in the frquency (and polar) response when they are not aligned properly.
> A lot of tweaking in filter response and 6 first order phase shifters (per channel) were required in order to finally achieve phase coherence and proper summing (which causes strong cancellation when tweeter polarity is reversed) between the outputs of the horns in the crossover region (around 1600Hz). It was not an easy task indeed.
So the bottomline is, one needs to know his electronics and acoustics well, to get good results from higher order crossovers. As Eva says achieving phase coherence and proper summing may need extra circuitry.
consort_ee_um, my first order filter was different, but it was better too..period (I didn’t measure its frequency response though). Anyway these discussions encourage me to learn more, so that I can design custom higher order filters that perform well..
By the way, what you guys think about some very popular websites which lure the readers into discarding their simple crossovers and embracing higher order actives, without mentioning anything about the above issues? Why do they hide these things? Maybe I am a victim, and there are others too, like Dan..
reji
Hi,
The other speaker was just a lab setup. So I didn´t give to much thought to it.
So I will try the tweeter displacement in my next project.
It should be beneficial and comes with only the cost of some wood.
But after all, it was just a "what can we do with a sine burst" thing and it shows a clearly measurable difference.
Regards
Jürgen
this is one impression I got from the whole thing. The time aligned box was a well designed, very good sounding speaker. But I beleave it wouldn´t be as good without the tweeter displacement.
The other speaker was just a lab setup. So I didn´t give to much thought to it.
So I will try the tweeter displacement in my next project.
It should be beneficial and comes with only the cost of some wood.
Some people discuss the sound of amplifiers, although the specs are very close.
But after all, it was just a "what can we do with a sine burst" thing and it shows a clearly measurable difference.
I don´t see me as a victim. But I don´t like webpages, articles and books that only explain some simplified textbook circuits. Always there should be a link to reality.
Regards
Jürgen
juergenk said:
You have to consider that a sine burst contains infinite frequencies and not just the fundamental frequency. Also, what these plots show is is just group delay.
And last but not least, note that the first purpose of any crossover filter is to avoid feeding more electrical energy to the driver outside its operating frequency range than it can handle in a linear and undistorted way. First order filters often fail to achieve that goal unless crossover points are set very conservatively, thus preventing an efficient use of the operating frequency range of the drivers. On the other hand, second order filters are usually enough for direct radiating hi-fi drivers, while horns tend to require sharper filters in order to exploit their operating frequency range as much as possible while keeping unwanted frequencies away.
Then, after dealing with energy issues, is comes summing and time alignment...
> Active crossover circuits that do not include phase correction circuitry are only marginally useable.
m8o, thanks for mentioning this. Though I had read from linkwitzlab.com, this important piece of information had escaped my eyes.
>"PLEASE TIME ALIGN US!!"
Yes..that's another important point..
so I have some interesting work at hand..only trouble is my DIY music system and myself are now at diametrically opposite points on the earth..have to wait couple of months more..
reji
m8o, thanks for mentioning this. Though I had read from linkwitzlab.com, this important piece of information had escaped my eyes.
>"PLEASE TIME ALIGN US!!"
Yes..that's another important point..
so I have some interesting work at hand..only trouble is my DIY music system and myself are now at diametrically opposite points on the earth..have to wait couple of months more..
reji
hello philiprej
there are a few important things for me :
- First, as it has been said before, you must take care of drivers time alignment, the higher order your crossover is, the more important this is for sure. This can be done by physically offsetting drivers or by using all pass delay filters. for tweeters, a small waveguide is a very good bet. If you don't , music will loose coherency, what you could call "life".
- Also, i guess from your previous statements that your bass and mid drivers have cones made of paper or polypropylene, because if not, it would be very unlikely that you like the sound of a first order crossover on them. If it's the case, using 24db slopes is clearly overkill. The advantages are not so big, except on stressing the drivers less, but if you chosen your crossover frequencies carefully this should clearly not be a problem, especially in a three way.
So my advice is, bypass 2 poles of each sections of your crossover. if you made it linkwitz they are identical so you should have a 12 db crossover , and see if it sounds better for you
- also if i understood it right you bypassed your mid driver in the first order config you like ! this is a huge information, maybe way more important than crossover slope !
in fact , maybe it's a very important factor in your appreciation, may it be for the reason that your mid driver sucks
more seriously, midrange coming out of a bigger driver is clearly not the same thing, the distortion and breakup topics have been covered but do not explain anything about why you should like it better Without your mid driver.
the fact is that the dispersion pattern of your midrange will be much more narrow if coming out of a bigger driver (wich i assume is the case) depending on the space you are listening in, having a narrow dispersion pattern in the midrange could be a clear advantage, bringing information back into the music
anyway, we seriously lack information about your system, your room , the drivers you use. the modification you made is huge and there are many things that could have led to your preference, so please enlighten us !
there are a few important things for me :
- First, as it has been said before, you must take care of drivers time alignment, the higher order your crossover is, the more important this is for sure. This can be done by physically offsetting drivers or by using all pass delay filters. for tweeters, a small waveguide is a very good bet. If you don't , music will loose coherency, what you could call "life".
- Also, i guess from your previous statements that your bass and mid drivers have cones made of paper or polypropylene, because if not, it would be very unlikely that you like the sound of a first order crossover on them. If it's the case, using 24db slopes is clearly overkill. The advantages are not so big, except on stressing the drivers less, but if you chosen your crossover frequencies carefully this should clearly not be a problem, especially in a three way.
So my advice is, bypass 2 poles of each sections of your crossover. if you made it linkwitz they are identical so you should have a 12 db crossover , and see if it sounds better for you
- also if i understood it right you bypassed your mid driver in the first order config you like ! this is a huge information, maybe way more important than crossover slope !
in fact , maybe it's a very important factor in your appreciation, may it be for the reason that your mid driver sucks
more seriously, midrange coming out of a bigger driver is clearly not the same thing, the distortion and breakup topics have been covered but do not explain anything about why you should like it better Without your mid driver.
the fact is that the dispersion pattern of your midrange will be much more narrow if coming out of a bigger driver (wich i assume is the case) depending on the space you are listening in, having a narrow dispersion pattern in the midrange could be a clear advantage, bringing information back into the music
anyway, we seriously lack information about your system, your room , the drivers you use. the modification you made is huge and there are many things that could have led to your preference, so please enlighten us !
First off this thread really belongs in the Loudspeakers section, imho.
A few thoughts:
The TL082 is not a particularly good sounding opamp. I'd expect to hear its effects, period. Don't judge ur results unless you run the control situation through a TL082 as well. I'd pick a much much better opamp myself.
Secondly, as someone noted in passing, the acoustic response is what counts, not only the electrical response of a filter. Get the system correct as a whole.
First order filters + speakers will not pass square waves unless some very special conditions are met. They usually are not. Check out John Dunlavy's work, and earlier B&Os JAES paper on the subject.
So called "first order xovers" work as first order xovers only when the bandwidth of the driver being x'd over is something like two octaves or greater than the xover point.
Driving most tweeters with a first order filter is a sure fire receipe for a signficant increase in distortion of all types - especially if that filter is at or near the F3 point of the tweeter. The distortion is non-linear and increases with level.
I don't like third order filters... just my "feeling".
It is a very good idea to make the impulse response of a multiway system look as much like a single pulse as you can - that means aligning the drivers timewise. Imho.
In my experience, many listeners tend to mistake added distortions for "loudness" and or "presence." Be careful - really, really clean speakers almost always sound less "loud" even at high volumes!!
_-_-bear
A few thoughts:
The TL082 is not a particularly good sounding opamp. I'd expect to hear its effects, period. Don't judge ur results unless you run the control situation through a TL082 as well. I'd pick a much much better opamp myself.
Secondly, as someone noted in passing, the acoustic response is what counts, not only the electrical response of a filter. Get the system correct as a whole.
First order filters + speakers will not pass square waves unless some very special conditions are met. They usually are not. Check out John Dunlavy's work, and earlier B&Os JAES paper on the subject.
So called "first order xovers" work as first order xovers only when the bandwidth of the driver being x'd over is something like two octaves or greater than the xover point.
Driving most tweeters with a first order filter is a sure fire receipe for a signficant increase in distortion of all types - especially if that filter is at or near the F3 point of the tweeter. The distortion is non-linear and increases with level.
I don't like third order filters... just my "feeling".
It is a very good idea to make the impulse response of a multiway system look as much like a single pulse as you can - that means aligning the drivers timewise. Imho.
In my experience, many listeners tend to mistake added distortions for "loudness" and or "presence." Be careful - really, really clean speakers almost always sound less "loud" even at high volumes!!
_-_-bear
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.