What you are seeing is the reason why in horns with a defined cutoff the general advice is to cross over an octave or two above that frequency. The resulting slope and group delay will then be mainly that of the filter put in place and not the one created by the horn giving out all of a sudden.
In the filters with excess group delay what goes in doesn't come out the same in time. If you want to make it a non issue and can accept some latency then use a linear phase complementary crossover away from cutoff.
This is why I posted images of the step response belonging to the crossover filters I showed before. There is no excess group delay in the linear phase crossover and the step is virtually perfect, what is left from perfection is caused by the 30 Hz rolloff of the simulated woofer.
In the filters with excess group delay what goes in doesn't come out the same in time. If you want to make it a non issue and can accept some latency then use a linear phase complementary crossover away from cutoff.
This is when I knew, the same thing that I know today...only now with better words to describe what I've known all along from experience...the technically correct words, mind you. I thank you for that. I've been arguing group delay and its importance for 2 years now....and I am still right! lmao!
I've always been discerning over group delay and I will continue that trend in my designing. I recommend you all the same. If you didn't know, now ya know =)
The sound of a piano, drums, guitar, the whole spectrum really....yes it matters.
if you can't hear it, at least keep it correct for those of us who can
Last edited:
There is another way. Just make the waveguide HP the crossover, but set the waveguide such that the woofers delay is compensated for and the crossover can linear phase and passive, or active. Makes things much more practical than looking for a 150 Hz horn. As I said this 2 x idea seems ridiculous to me. (Warning!!! can't be done for pro applications!!)
This is the golden key!
I've got the general concept down. I've played and manipulated it in musical instruments for so long now, its second nature.
Timbre - : the quality of tone distinctive of a particular singing voice or musical instrument or the resonance by which the ear recognizes and identifies a voiced speech sound
Due to the different organization of frequencies arriving to the table of perception at similar or intended levels....Timbre has changed.
Somewhere in there, is, "I don't mind this change" and "this is actually more accurate"
Last edited:
Now you are talking about "perception". You need to stay in one domain at a time. Perception seems to be the one you are most confident in, most people think that of themselves. "I hear it, therefore it is!" Dangerous territory that.
Geddes you are also good with words, I like that. All points taken. Thank you for helping put GD on the map as a "Time Domain Effect" more technically referred to as "start up transient" another trait of GD outside of....actually, what's its other trait?
Last edited:
Group delay is the part of the phase that is responsible for those things I highlighted. I guess even that is hard to say without acknowledging that phase is too, its just....which parts...and group delay is no less a topic than phase. How we address these things, is a matter design philosophy....
I'm not saying that I've mastered the topic, I'm just saying I have a good perspective on it.
We have the thresholds studies. That's good enough for a lot of people. I personally would just aim for the best, and not think twice about it.
Last edited:
the left with no XO and the right with 48db LR at 500hz.
Some of the amplitude character is unique as far out as 9-13ms in these examples.
Attachments
Last edited:
Not because we can't make some accurate guesses but because there are so many different aspects. All I've learned truly is that there is much more that I don't know. In discovering this aspect of group delay in a more intimate way, I've discovered other traits that I had no idea to anticipate.
This technique of recording signal and comparing it to source signal as a form of deeper analysis? Seems legit... You might be surprised at what you find
This technique of recording signal and comparing it to source signal as a form of deeper analysis? Seems legit... You might be surprised at what you find
Please describe your recording and test chain.
You mean like this?
As measured with a demo version of APL_TDA at the listening spot some time back in 2016. As said, nothing new.
Probably have some REW files too, showing more of the same. As I kept nagging John about getting the APL type of display into REW (which he did).
Mark will have plenty of measurements that would show similar results.
We've already told you how to get there, fluid did so only a couple of posts ago. But you kept on discovering all these new things about GD.
Congrats, two years of study that could be turned into an online course !
maybe titled "Group Delay Revelations 101"
With a suggested course description:
Theoretical analysis of group delay studying examples of means and methods that a speaker builder would not choose to utilize a driver, a horn, or a xover.
The course will examine such examples with an entirely theoretical intuitive focus, without the constraint of staying consistent with known measurement science. (Such constraints put a terrible damper on validating intuitive revelations.)
There will be times however, when revelations are shown to be consistent with known measurement science; in which case the course will expound on why the revelations are new and unique to the audio world.
The course will at all times abstain from studying examples of means and methods that a speaker builder would choose to utilize a driver, a horn, or a xover.
(because that would of course defeat the purpose of this course...)
Meant all in fun Camplo 🙂
best, Mark
Group delay vs Time domain distortion. I've had some important people tell me there was no such aspect of it, rather that, Group delay, is just another view of phase. The recordings of waveform reproduction says other wise. The anomalies witnessed in the amplitude are not phase related in the sense of synchronicity of crest/trough to another signal. If I had to make a point, I could go back through this thread and find that comment made several times over...it is an incorrect statement.
Last edited:
Non-flat group delay means phase distortion and that changes time response. It's still the same data, looked at from different perspective. In a case of a horn, from its amplitude frequency response alone, you can calculate all the rest.
If you don't include the aspect of changes in amplitude vs time you are not providing a whole definition