A 100% accurate (i.e. transient perfect) speaker is impossible to build.
With digital crossovers you can get quite close as you may see here. If you want to achieve temporal accuracy down to the lower cutoff frequency then this is at the expense of increased total delay - which may or may not be a problem depending on the usage.
http://www.klein-hummel.com/klein-h...nitoring_studio-monitors_main-monitors_O500C#
With analog topologies the best that you can achieve is an overall behaviour of a wide 4th order bandpass. The rise-time will be dependant on the upper cutoff frequency and the decay will depend on the lower cutoff frequency. The shape will also be influenced by the Qs of the high- and lowpass part of the transfer function.
Regards
Charles
With digital crossovers you can get quite close as you may see here. If you want to achieve temporal accuracy down to the lower cutoff frequency then this is at the expense of increased total delay - which may or may not be a problem depending on the usage.
http://www.klein-hummel.com/klein-h...nitoring_studio-monitors_main-monitors_O500C#
With analog topologies the best that you can achieve is an overall behaviour of a wide 4th order bandpass. The rise-time will be dependant on the upper cutoff frequency and the decay will depend on the lower cutoff frequency. The shape will also be influenced by the Qs of the high- and lowpass part of the transfer function.
Regards
Charles
Iain McNeill said:
I know an individual driver or a fullrange driver is linear phase. And I also know there's no true t-p speaker in practice.
I want to have an example of a multi-driver speaker that is only linear phase but not (even quasi) transient perfect (or zero phase).
I think using digital crossovers with time delay that is linear to frequency, we can make a speaker, even with higher order rolloffs, have the linear phase property (i.e., group delay exists but it changes linearly across frequencies). But this is not transient perfect.
If we ignore the real world drivers' limited band pass, a true t-p (or zero phase wrap) speaker can be obtained only by using a first order slopes whether it uses digital or analog crossovers.
Am I correct?
Jay_WJ said:
Jay,
There are a number of approaches to a transient perfect speaker (bandwidth limited of course) besides first order slopes. Lipshitz and Vanderkooy as well as Eric Baekgaard have published papers in the JAES on the subject. John K has done a lot of work on this including the filler driver approach and higher order TP systems using subtractive delay.
Here are a few links to some of John's studies if you haven't seen them:
John Ks ICTA project: http://www.musicanddesign.com/icta_cross.html
A variety of pages John has written on the subject can be found here: http://www.geocities.com/kreskovs/John1.html
Regards,
Dennis
djarchow said:
Thanks for the reply. I've read all of John K's articles about this subject. As far as I understand, all of the T-P designs he discusses use sort of quasi 1st order acoustic rolloffs---close to 1st order slopes near drivers' large overlap region but higher rolloffs far from it (or at least a driver is faithful to 1st order rolloff). So essentially they just consider practical implementations of BW1---not exactly BW1 but close to it in practice. Only exception is the higher order subtractive delay approach via digital xover. But I think this is not T-P but only linear phase. That's what I meant in my above post.
Jay,
The subtractive delay approach, at least how John is using it, is TP, again while accepting the bandpass limited nature of the system as a whole. See his ICTA project.
John's filler driver designs use a 2nd order slopes for the high pass and low pass assuming a gamma of 1 (Q=.5) which is the optimal as relates to lobing. The filler driver slope is a first order bandpass though. I designed one of these a couple years back with the Revelator midwoofers, a Vifa MG filler, and the Millenium tweeter. While they sounded great, the almost 6 foot tall cabinets had about zero WAF and are sitting out in the garage now.
Regards,
Dennis
The subtractive delay approach, at least how John is using it, is TP, again while accepting the bandpass limited nature of the system as a whole. See his ICTA project.
John's filler driver designs use a 2nd order slopes for the high pass and low pass assuming a gamma of 1 (Q=.5) which is the optimal as relates to lobing. The filler driver slope is a first order bandpass though. I designed one of these a couple years back with the Revelator midwoofers, a Vifa MG filler, and the Millenium tweeter. While they sounded great, the almost 6 foot tall cabinets had about zero WAF and are sitting out in the garage now.
Regards,
Dennis
These here use the subtractive-delay approach:
http://www.psiaudio.com
The A623 and A723 also in an analog fashion and the A823 with a digital crossover:
http://www.studerundrevox.de/index.php?page=222
When you want to use the constant-voltage approach like John K's ICTA then you will either have to use the lowest filter order possible (that't what I do with my Manger system with 1st order lowpass and 2nd order highpass) or you will have to use a D'Appolito configuration (what John does).
But theoretically you can do all kinds of possible symmetrical and asymmetrical crossovers using the constant-voltage approach, but you will always have increased lobing and humps.
Regards
Charles
http://www.psiaudio.com
The A623 and A723 also in an analog fashion and the A823 with a digital crossover:
http://www.studerundrevox.de/index.php?page=222
When you want to use the constant-voltage approach like John K's ICTA then you will either have to use the lowest filter order possible (that't what I do with my Manger system with 1st order lowpass and 2nd order highpass) or you will have to use a D'Appolito configuration (what John does).
But theoretically you can do all kinds of possible symmetrical and asymmetrical crossovers using the constant-voltage approach, but you will always have increased lobing and humps.
Regards
Charles
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.