To Mezzo:
Please do not think I am now rejecting the series xover. I will be testing it again (soon) against the notch-filtered parallel xover in A-B mono as you suggest, and then in stereo. I am still convinced that the series crossover has some strong advantages. On the other hand, I think the parallel xover with notch filter does deal better with the 2 KHz peak.
To everybody:
Last night, I invited two friends over, and we had an extended listening session into the early hours of the morning. (My neighbours must love me.) One of these guys didn't like the clone at all previously (with either the original parallel or the series xover), but is now won over - I think.
We all agreed that vocals in particular are now much smoother, and that treble detail has improved also (removal of the masking effect of the 2 KHz peak from the mid-woofer?) In fact, it was felt that the treble was now a little exposed and a drop in tweeter level was desirable (corresponding to TG's recommendation). I had been using a 5 ohm resistor, but replaced it with 6 ohms, which we all agreed gave a more balanced sound.
We had done some A-B mono testing first, and one of my visitors felt there was a small amount of midrange constriction with the notch filter in place. While I disagree, I think the comment is understandable, as a more forward balance can be initially impressive. He did agree that the removal of the glare was obvious, though - and later in stereo testing also agreed that background vocalists/instruments were clearer.
The speaker is still quite revealing of differences in recording quality, and we found some recordings that remain a little hard to listen to, though not nearly as bad as before. While the 2.5 may not be the most analytical speaker around, it is still very revealing, even with the notch filter! Nevertheless, most of the marginal recordings that had been fatiguing previously were now able to be enjoyed for the music. And the good ones sound magnificent.
We switched back a couple of times in stereo between filtered and unfiltered modes (till I got tired of soldering and lugging the speakers around!), and there was a general consensus that image depth improved with the filter in place.
The clone is still not flawless (it still has that lower mid dryness and occasional excessive sibilance), but to my ears (and those of my friends) its major fault was the midrange glare and everything else is forgiveable, given its overall magic. Why Proac did not include a notch filter is beyond me! It wouldn't add much to the original's cost.
Like yourself, Al.M, I'll be carrying out further listening tests over the coming weeks, and will also be doing further comparisons with Mezzo's series filter.
By the way, I'm sure I've read somewhere that notch and resonant peak filters should be placed as close as possible to the driver. This is what I did, anyway, as my crossovers are in the bottom of the cabinet (on a removable panel). I attached the notch filter to a side wall, adjacent to the 8535. Hopefully, the magnetic field isn't affecting it too much. It sounds right, anyway.
Please do not think I am now rejecting the series xover. I will be testing it again (soon) against the notch-filtered parallel xover in A-B mono as you suggest, and then in stereo. I am still convinced that the series crossover has some strong advantages. On the other hand, I think the parallel xover with notch filter does deal better with the 2 KHz peak.
To everybody:
Last night, I invited two friends over, and we had an extended listening session into the early hours of the morning. (My neighbours must love me.) One of these guys didn't like the clone at all previously (with either the original parallel or the series xover), but is now won over - I think.
We all agreed that vocals in particular are now much smoother, and that treble detail has improved also (removal of the masking effect of the 2 KHz peak from the mid-woofer?) In fact, it was felt that the treble was now a little exposed and a drop in tweeter level was desirable (corresponding to TG's recommendation). I had been using a 5 ohm resistor, but replaced it with 6 ohms, which we all agreed gave a more balanced sound.
We had done some A-B mono testing first, and one of my visitors felt there was a small amount of midrange constriction with the notch filter in place. While I disagree, I think the comment is understandable, as a more forward balance can be initially impressive. He did agree that the removal of the glare was obvious, though - and later in stereo testing also agreed that background vocalists/instruments were clearer.
The speaker is still quite revealing of differences in recording quality, and we found some recordings that remain a little hard to listen to, though not nearly as bad as before. While the 2.5 may not be the most analytical speaker around, it is still very revealing, even with the notch filter! Nevertheless, most of the marginal recordings that had been fatiguing previously were now able to be enjoyed for the music. And the good ones sound magnificent.
We switched back a couple of times in stereo between filtered and unfiltered modes (till I got tired of soldering and lugging the speakers around!), and there was a general consensus that image depth improved with the filter in place.
The clone is still not flawless (it still has that lower mid dryness and occasional excessive sibilance), but to my ears (and those of my friends) its major fault was the midrange glare and everything else is forgiveable, given its overall magic. Why Proac did not include a notch filter is beyond me! It wouldn't add much to the original's cost.
Like yourself, Al.M, I'll be carrying out further listening tests over the coming weeks, and will also be doing further comparisons with Mezzo's series filter.
By the way, I'm sure I've read somewhere that notch and resonant peak filters should be placed as close as possible to the driver. This is what I did, anyway, as my crossovers are in the bottom of the cabinet (on a removable panel). I attached the notch filter to a side wall, adjacent to the 8535. Hopefully, the magnetic field isn't affecting it too much. It sounds right, anyway.
I do not have large thing to add, all is known as 😉
I could make comparisons with two on line clone, one in series and one in //, and for me and several person which came listened one found the filter series superior, not because it was me which with optimized the filter series because I is not known as to any person in which loudspeakers found the filter series 😉
Then which to believe? What means our ears or the theory applied in the filters?
For information, PROAC update its response 2.5 for soon with a corrector of impedance according to infos 😉
The impedance at all linear and they in one were not surely
taken account according to the remarks read on certain forum.
My filter series presents a curve of impedance linear than the original clone, therefore the whole of the sound scene is detached better around the instruments, the medium on the proac original was slightly masked.
So really found you the filter series lower than the filter//,
I would remove it my site but in any case it will remain in my clone
of 2.5 😉
Mezzo 😉
I could make comparisons with two on line clone, one in series and one in //, and for me and several person which came listened one found the filter series superior, not because it was me which with optimized the filter series because I is not known as to any person in which loudspeakers found the filter series 😉
Then which to believe? What means our ears or the theory applied in the filters?
For information, PROAC update its response 2.5 for soon with a corrector of impedance according to infos 😉
The impedance at all linear and they in one were not surely
taken account according to the remarks read on certain forum.
My filter series presents a curve of impedance linear than the original clone, therefore the whole of the sound scene is detached better around the instruments, the medium on the proac original was slightly masked.
So really found you the filter series lower than the filter//,
I would remove it my site but in any case it will remain in my clone
of 2.5 😉
Mezzo 😉
Mezzo:
I am not saying at this stage that the notch-modified parallel crossover is better or worse than your series crossover. The only tests I have carried out over the last two days have been between the original parallel xover and the notch-filtered parallel xover.
My previous comments on the series crossover remain the same: that is, I agree that it is preferable overall to the original parallel crossover:
The series crossover provides more space or air around instruments and vocalists, giving each what I could best describe as an individual dynamic, and produces better layering in the depth plane. There is a clearer picture of what is going on behind the scenes. I also feel that the drivers integrate better than they do in the original, and that there is a greater sense of continuity in the midrange. I feel that the problem at 2 KHz is much better dealt with by the series xover, and that the listening experience is more enjoyable than with the original. I still have no hesitation in recommending the series xover.
The new modification to the parallel crossover (the addition of the notch filter) is, however, excellent, and improves this version of the speaker by a large amount. At the moment, I have not compared the series and notch-filtered parallel options, but I will be doing so soon. It may end up being a hard choice for me, or a clear-cut one - but at the moment I simply don't know.
The series xover remains IMO a viable option for clone builders and I urge others to try it along with all published options. I suspect many will be won over by its qualities.
I am not saying at this stage that the notch-modified parallel crossover is better or worse than your series crossover. The only tests I have carried out over the last two days have been between the original parallel xover and the notch-filtered parallel xover.
My previous comments on the series crossover remain the same: that is, I agree that it is preferable overall to the original parallel crossover:
The series crossover provides more space or air around instruments and vocalists, giving each what I could best describe as an individual dynamic, and produces better layering in the depth plane. There is a clearer picture of what is going on behind the scenes. I also feel that the drivers integrate better than they do in the original, and that there is a greater sense of continuity in the midrange. I feel that the problem at 2 KHz is much better dealt with by the series xover, and that the listening experience is more enjoyable than with the original. I still have no hesitation in recommending the series xover.
The new modification to the parallel crossover (the addition of the notch filter) is, however, excellent, and improves this version of the speaker by a large amount. At the moment, I have not compared the series and notch-filtered parallel options, but I will be doing so soon. It may end up being a hard choice for me, or a clear-cut one - but at the moment I simply don't know.
The series xover remains IMO a viable option for clone builders and I urge others to try it along with all published options. I suspect many will be won over by its qualities.
Nadir,
that reassures me 😉
I awaits the same things that you 😉
for information, I made the comparative test with the modification of Jacques and my filter series 😉
Aldo
clone 2.5 original is very very well but it is softer with a certain
reserve. it slightly misses life I think😉
that reassures me 😉
I awaits the same things that you 😉
for information, I made the comparative test with the modification of Jacques and my filter series 😉
Aldo
clone 2.5 original is very very well but it is softer with a certain
reserve. it slightly misses life I think😉
Sorry, I meant to include this link to the audioclone series crossover for others to try:
http://audioclone.free.fr/PROAC 2.5 SERIE.gif
http://audioclone.free.fr/PROAC 2.5 SERIE.gif
In view of Mezzo's concerns over my tentative comments on the Proac 2.5 series crossover versus the notch-filtered parallel crossover, I decided to have yet another late night and do things properly. Last time I was relying on memory alone, as the series crossover had been removed to allow the parallel crossover notch-filter modification to be evaluated and there was quite a gap between then and the last time I had heard the series crossover. I probably didn't make this clear enough.
Memory can play tricks, especially with recollections of sound, so I have now spent several hours listening to the parallel crossover (with notch filter) against the audioclone series crossover. Listening tests were carried out with A-B mono switching and then with both sides changed for stereo evaluation. It proved a frustrating experience.
If this had been a straight comparison between the original (Jacq-modified) parallel xover and the audioclone series xover, I would have had no hesitation in choosing the series as I have in the past. In fact, the series xover was, until the notch-filter tests, a permanent part of my clone. As I suspected, with the notch filter in place in the parallel xover, the issues are less clear. I won't repeat myself by describing in detail my impressions of the two, as I have already rambled on at length starting on page 1 of this thread, and the standard clone without notch filter has been well covered elsewhere.
For me, the fact that the notch filter totally eliminates the 2 KHz peak and its detrimental effects is a major advance for the parallel xover. As I thought, mono comparisons with the series xover showed that it is not quite as smooth here with some recordings, though it is considerably better in this respect (IMO) than the parallel without the notch filter.
On the other hand, the comparisons confirmed that there are other qualities in the midrange with the series xover that I really like, and my previous positive comments on the other aspects of its attractive, airy and dynamic sound remain unchanged . . .
I have concluded for the moment that, provided the parallel xover has the notch filter, the choice between the two comes down to their quite different approaches to interpreting music. I don't think that either is "better" (I know Mezzo will disagree). With some recordings, I found myself preferring the modified parallel crossover and with others, the series. I found it very difficult to decide. Other listeners will find it easier, I am sure. They are both IMO very good options for the clone. I would like to be able to throw a switch from one to the other, depending on the CD or LP I am listening to at the time (a bit impractical, and switches usually degrade sound quality).
At this early stage, I still think clone-builders should try both alternatives and then make a decision. They are different, but either way it's a great speaker. Just make sure the parallel crossover has both Jacq's mods and the Troels Gravesen notch filter (and the tweeter polarity inverted at either the input or output). For the series crossover, follow the schematic exactly. As always, let your own ears decide.
Memory can play tricks, especially with recollections of sound, so I have now spent several hours listening to the parallel crossover (with notch filter) against the audioclone series crossover. Listening tests were carried out with A-B mono switching and then with both sides changed for stereo evaluation. It proved a frustrating experience.
If this had been a straight comparison between the original (Jacq-modified) parallel xover and the audioclone series xover, I would have had no hesitation in choosing the series as I have in the past. In fact, the series xover was, until the notch-filter tests, a permanent part of my clone. As I suspected, with the notch filter in place in the parallel xover, the issues are less clear. I won't repeat myself by describing in detail my impressions of the two, as I have already rambled on at length starting on page 1 of this thread, and the standard clone without notch filter has been well covered elsewhere.
For me, the fact that the notch filter totally eliminates the 2 KHz peak and its detrimental effects is a major advance for the parallel xover. As I thought, mono comparisons with the series xover showed that it is not quite as smooth here with some recordings, though it is considerably better in this respect (IMO) than the parallel without the notch filter.
On the other hand, the comparisons confirmed that there are other qualities in the midrange with the series xover that I really like, and my previous positive comments on the other aspects of its attractive, airy and dynamic sound remain unchanged . . .
I have concluded for the moment that, provided the parallel xover has the notch filter, the choice between the two comes down to their quite different approaches to interpreting music. I don't think that either is "better" (I know Mezzo will disagree). With some recordings, I found myself preferring the modified parallel crossover and with others, the series. I found it very difficult to decide. Other listeners will find it easier, I am sure. They are both IMO very good options for the clone. I would like to be able to throw a switch from one to the other, depending on the CD or LP I am listening to at the time (a bit impractical, and switches usually degrade sound quality).
At this early stage, I still think clone-builders should try both alternatives and then make a decision. They are different, but either way it's a great speaker. Just make sure the parallel crossover has both Jacq's mods and the Troels Gravesen notch filter (and the tweeter polarity inverted at either the input or output). For the series crossover, follow the schematic exactly. As always, let your own ears decide.
Nadir: I also have a nicely built DIY Proac 2.5 with original parrallel Jacqs' x-over, and accordingly I am following this thread with some interest!
May I make a suggestion with your evaluation technique of the two xo's, that you try NOT doing quick A-B tests to see which you prefer. I have found that when two audio components sound basically very similar, A-B tests can be very confusing...with everything ending up sounding the same!
You should sit with one of the xo's for about a week listening to all your favourite tracks, and then switch back to the other xo...the difference should be immediately noticeable! What happens here is that your brain (and ears) have had a chance to acclimatise to one type of sound over a long period, and when it changes, the difference is more noticeable. With A-B tests, the sonic signature of one xo, over another, is not able to imprint on your mind.
I find ultimately though, that a combination of A-B tests and long-term listening is best. But the latter technique allows you to better guage which sound you might prefer.
For what it's worth...
Regards,
Steve M.
May I make a suggestion with your evaluation technique of the two xo's, that you try NOT doing quick A-B tests to see which you prefer. I have found that when two audio components sound basically very similar, A-B tests can be very confusing...with everything ending up sounding the same!
You should sit with one of the xo's for about a week listening to all your favourite tracks, and then switch back to the other xo...the difference should be immediately noticeable! What happens here is that your brain (and ears) have had a chance to acclimatise to one type of sound over a long period, and when it changes, the difference is more noticeable. With A-B tests, the sonic signature of one xo, over another, is not able to imprint on your mind.
I find ultimately though, that a combination of A-B tests and long-term listening is best. But the latter technique allows you to better guage which sound you might prefer.
For what it's worth...
Regards,
Steve M.
Nadir
Just one more thing, you are probably doing this anyway but when listening to the parralel xo be sure to bi-wire as I have found this improves separation and imaging slightly, particularly for some SS amps. Although for me valve amplification is superior on the clones over SS.
With the series xo it appears you are not able to take advantage of bi-wiring, assuming this is needed or you beleive in it. But it appears the series xo is making up for this anyway.
We should try and regroup a week later as a few others on the Madisound forum are trialing things as well.
Al.M
Just one more thing, you are probably doing this anyway but when listening to the parralel xo be sure to bi-wire as I have found this improves separation and imaging slightly, particularly for some SS amps. Although for me valve amplification is superior on the clones over SS.
With the series xo it appears you are not able to take advantage of bi-wiring, assuming this is needed or you beleive in it. But it appears the series xo is making up for this anyway.
We should try and regroup a week later as a few others on the Madisound forum are trialing things as well.
Al.M
Steve M.:
I do agree with you, but when it comes to the effect in the critical 2 KHz area, I've found the A-B mono tests are very revealing and unambiguous. In this case, they did not end up sounding the same in any of the tests I have mentioned.
Short-term listening in stereo is probably the most problematic. I have often found that where the effects are subtle, the change-over time of 20 minutes or so is sufficient to blur the memory. In this case, the removal of the midrange glare with the notch filter in the parallel xover is pretty obvious though, as is the very different sound of the audioclone series xover.
You are right also in your comments on long-term listening tests, and this is what I am doing at the moment with the notch filter mod. It really is the only way to confirm one's impressions. I have done this at length with the series xover vs the parallel xover (before the notch filter mod), so am pretty sure what I think about those two options.
It's a pity in some ways that this discussion has been complicated by having three options to consider. If the notch filter hadn't come along when it did, things would be simpler (parallel vs series). Nevertheless, I think the growing number of choices will prove a plus for clone-builders in the long run.
Al.M:
Yes, I always listen to the parallel-xovered clone bi-wired (tried it single-wired back in the early days and agree that bi-wiring is essential).
I use both SS and valve amps, and agree that the clone is generally better with valve amplification, though SS does usually give it better bass. If SS is used, it needs to be uncharacteristically warm in the midrange. My favourite is my aging but lovely Sugden C51/P51 combination, which does have some valve-like characteristics. I've borrowed a few friends' SS amps which sounded uncomfortably cold and, in one case, quite harsh.
Agreed on the regroupment. I hope that many cloners will trial both the notch filter mod and the series xover. This should be interesting.
I do agree with you, but when it comes to the effect in the critical 2 KHz area, I've found the A-B mono tests are very revealing and unambiguous. In this case, they did not end up sounding the same in any of the tests I have mentioned.
Short-term listening in stereo is probably the most problematic. I have often found that where the effects are subtle, the change-over time of 20 minutes or so is sufficient to blur the memory. In this case, the removal of the midrange glare with the notch filter in the parallel xover is pretty obvious though, as is the very different sound of the audioclone series xover.
You are right also in your comments on long-term listening tests, and this is what I am doing at the moment with the notch filter mod. It really is the only way to confirm one's impressions. I have done this at length with the series xover vs the parallel xover (before the notch filter mod), so am pretty sure what I think about those two options.
It's a pity in some ways that this discussion has been complicated by having three options to consider. If the notch filter hadn't come along when it did, things would be simpler (parallel vs series). Nevertheless, I think the growing number of choices will prove a plus for clone-builders in the long run.
Al.M:
Yes, I always listen to the parallel-xovered clone bi-wired (tried it single-wired back in the early days and agree that bi-wiring is essential).
I use both SS and valve amps, and agree that the clone is generally better with valve amplification, though SS does usually give it better bass. If SS is used, it needs to be uncharacteristically warm in the midrange. My favourite is my aging but lovely Sugden C51/P51 combination, which does have some valve-like characteristics. I've borrowed a few friends' SS amps which sounded uncomfortably cold and, in one case, quite harsh.
Agreed on the regroupment. I hope that many cloners will trial both the notch filter mod and the series xover. This should be interesting.
Thanks again to Troels G. it looks like we have some more final answers and directions on the 2.5 clone crossover.
See link at: http://members.chello.se/jpo/2_5_clone_measurements_v4.pdf
See link at: http://members.chello.se/jpo/2_5_clone_measurements_v4.pdf
I join with Al.M in thanking Troels for the many hours of work he has obviously put in.
The measurements made by Troels show that the audioclone series xover has serious faults in both frequency and phase, and needs more work before it can be seen as a viable alternative for the clone.
Clearly, what the series xo does is similar to the effect of a "loudness switch" (boosted highs & lows/depressed midrange). This certainly eliminates the 2 KHz peak and its audible effects - and on poorly balanced recordings the effect is undoubtedly pleasing. But accurate? After further listening in comparison to Troels' own efforts, I have to agree regretfully that it is not.
Since my last contribution to this forum, I have listened long and carefully to Troels' own original notch-filter mod to the parallel xover (and his latest fine tuning of its basic components), and am now sure that these are the best sounds I have heard from the clone to date. The elimination of the 2 KHz bump is obvious and the reduction in listening fatigue very satisfying. I can now listen to music without distraction, and feel there are also consequent improvements in upper midrange detail and imaging. I am still a little unsure about Troels' latest fine tuning, as (IMO) it does alter the voicing of the clone more than the simple notch filter addition did. Some cloning purists may object.
I hope that Mezzo will continue development of the series xover. Despite its uneven response, IMO it does have other positive and promising aspects to its sound, and if the frequency response (and phase) can be made more even without re-introducing the midrange peak, it may still prove a contender.
The measurements made by Troels show that the audioclone series xover has serious faults in both frequency and phase, and needs more work before it can be seen as a viable alternative for the clone.
Clearly, what the series xo does is similar to the effect of a "loudness switch" (boosted highs & lows/depressed midrange). This certainly eliminates the 2 KHz peak and its audible effects - and on poorly balanced recordings the effect is undoubtedly pleasing. But accurate? After further listening in comparison to Troels' own efforts, I have to agree regretfully that it is not.
Since my last contribution to this forum, I have listened long and carefully to Troels' own original notch-filter mod to the parallel xover (and his latest fine tuning of its basic components), and am now sure that these are the best sounds I have heard from the clone to date. The elimination of the 2 KHz bump is obvious and the reduction in listening fatigue very satisfying. I can now listen to music without distraction, and feel there are also consequent improvements in upper midrange detail and imaging. I am still a little unsure about Troels' latest fine tuning, as (IMO) it does alter the voicing of the clone more than the simple notch filter addition did. Some cloning purists may object.
I hope that Mezzo will continue development of the series xover. Despite its uneven response, IMO it does have other positive and promising aspects to its sound, and if the frequency response (and phase) can be made more even without re-introducing the midrange peak, it may still prove a contender.
I just thought it interesting to note that the dual problems of 2kHz bump and 3kHz trough are predicted even in simulation.
http://www.diyaudio.com/forums/showthread.php?threadid=8541
http://www.diyaudio.com/forums/showthread.php?threadid=8541
- Status
- Not open for further replies.