I am not totally taken by the popular conviction that healthy individuals of same age groups listen that different. Imagine the evolutionary selective disadvantages in group hunting over stone age...Maybe we just like different?
I think there is such criteria, but I suspect they lay confined into the artistic. I would ask master symphonic conductors to contribute about tone, subtlety, and preference. I wish I could. It would be illuminating.
I think there is such criteria, but I suspect they lay confined into the artistic. I would ask master symphonic conductors to contribute about tone, subtlety, and preference. I wish I could. It would be illuminating.
Finally got console to work!
It was soundcard routing issue. I have downloaded Voxengo audio delay (free) and linear phase parametric equilazer Neon (trial). Before i didn't have detailed phase control , just phase invert - this makes huge difference in depth and space representation. Curently trying out some 3way config, it works great. Neon doesn't have any steeper crossover than 24db/oct, although i wouldn't use them anyway.
Any recommedations for some other linear phase param. eq?
It was soundcard routing issue. I have downloaded Voxengo audio delay (free) and linear phase parametric equilazer Neon (trial). Before i didn't have detailed phase control , just phase invert - this makes huge difference in depth and space representation. Curently trying out some 3way config, it works great. Neon doesn't have any steeper crossover than 24db/oct, although i wouldn't use them anyway.
Any recommedations for some other linear phase param. eq?
Mx said:
Sweet, you now have a whole new world of plugins to explore!
Look at Thunau's Frequency Allocator as a starting point. Very easy to use, sounds great and the 'lite' version is perfect for low latency use.
If you wanting EQ plugins then Waves, Voxengo, Algorithmix all do excellent examples. Some are very pricey though and are really intended for studio mastering.
The other options is to use a convolution engine such as convolver or BruteFIR. These will filter audio according to specialist impulse responses created within programs such as Acourate and Audio Lense.
I would recommend Frequency allocator as a starting point then move onto the convolution method.
ShinOBIWAN said:
Thanks so much for (once again ) excellent reply, and I very much look forward to your updated how to.
ShinOBIWAN said:
Thanks ShinObiwan, I will look at all these plugins.
Just tell me, are you planning to use PC crossover as final solution, or will crossover be inside preamp?
ShinOBIWAN said:
Say you decide to make a chocolate shake for yourself.. do you want to individually taste the milk, ice cream, and chocolate syrup? Or would you instead prefer to simply have an excellent chocolate shake? To get to that excellent shake you need the right amount of blending for those ingredients.
The basic principal applies to the loudspeaker. The drivers don't have the same mass. They don't have the same surround. They don't have the same radiation character. So you do your best to get the right "blending" between the drivers to achieve an *overall* more accurate sound. Doing otherwise leaves you with an excellent sounding (and measuring) tweeter, midrange, and woofer - i.e. a "technically" better speaker.
Say you decide you cook a nice meal for yourself .. do you want to individually taste the soup, the meat, the potato, the salad and the desert? Or would you prefer to simply have an "excellent mash" prep'd in a mixer and then in the shaker? I fear that even the right amount of blending for the ingredients will not help.
How to mix the musicians and their instruments in an orchestra to get a nice music event. Shall they randomly take their seat or change the instruments during the concert?
ScottG said:
uli.brueggemann said:
Say you decide you cook a nice meal for yourself .. do you want to individually taste the soup, the meat, the potato, the salad and the desert? Or would you prefer to simply have an "excellent mash" prep'd in a mixer and then in the shaker? I fear that even the right amount of blending for the ingredients will not help.
How to mix the musicians and their instruments in an orchestra to get a nice music event. Shall they randomly take their seat or change the instruments during the concert?
Uhg... I don't mean to be rude, but I think creating analogies like the ones mentioned above just plain confuses people. I see this happen a lot with respect to audio. In reality, loudspeaker crossovers are unlike milkshakes and nice meals.
m0tion said:
I don't think that either you or uli misunderstood the analogy.. Nor do I think its likely that Shin did either.
It was meant to be humorous and on point (however trite it might be - and admittedly it was exceedingly so). 😀
As far as reality is concerned.. well, that depends on your perspective. 😉
Oh, hah, yeah maybe I was taking your comments a little too seriously. I just really dislike it when really objective things are made to be really subjective. Not that there aren't subjective parts to loudspeaker design, but sometimes it's made out to be a lot more subjective than it really is.
I guess I understood what you tried to say 🙂
My answer was intended just to show that it is dangerous to use analogies (my example should not to confuse anyone). Pretty often analogies do no meet the point and for sure it is not possible to prove anything with an analogy. So my example is also nonsense of course.
In this way I guess you also immediately understood my example. 😉
In fact we cannot discuss different properties of crossovers like this. We will find advantages and disadvantages with each design. And we must live with...
m0tion said:
You've seen Uli and I post some pretty impressive objective measurement showing a class of accuracy rarely seen in commercial loudspeakers. I agree these are important and do correlate to good sound but subjective qualities such as those of filter slopes, especially when viewed on a like for like basis in a direct comparison, highlight that subjectivity is important.
I personally think Scott's analogy was apt and is just what I experience when cranking up the slopes. You can literally hear the drivers doing there own thing. Sure it measures better but thats why a good does of subjectivity is important also.
I like to show some reason why a XO of a lower order my sound better than a higher order. See appended picture.
The picture displays some lowpass XOs at corner frequency 200 Hz. The blue curve is a Butterworth order 1. Of course the amplitude response is quite flat. The lower graph shows the time domain behaviour, here as minimum phase step response. We can see that the blue curve starts quickly (it introduces less delay) and it has a soft transition to 1.
The red curve is a Butterwort order 10. It may be desired to select if a steep slope is necessary to suppress e.g. resonance frequencies of a cone. The tradeoff is its ringing behaviour which you can clearly see in the step response display. It also creates a bigger time delay.
The green curve is a Bessel filter of order 10. So the slope is the same like the Butterworth of 10th order. You certainly will notice that it looks like a compromise between the other curves. The transition for the amplitude response is more smooth. An interesting point is the step response. It has no overshoot despite the high filter order. But it also shows some delay.
Here we get to a point where we can see the advantages of the digital world. Now we are able to design crossovers according to our needs, it is not mandatory to follow the old tradition of analog passiv crossovers. We can compensate also for the delays. We can use linear phase crossovers. No problem to design crossovers with an subtractive approach.
At the end we find a way to use the best part of a given driver and to integrate it with other drivers in a proper way. This is my understanding of "cooking" as a method of synthesis 🙂
The picture displays some lowpass XOs at corner frequency 200 Hz. The blue curve is a Butterworth order 1. Of course the amplitude response is quite flat. The lower graph shows the time domain behaviour, here as minimum phase step response. We can see that the blue curve starts quickly (it introduces less delay) and it has a soft transition to 1.
The red curve is a Butterwort order 10. It may be desired to select if a steep slope is necessary to suppress e.g. resonance frequencies of a cone. The tradeoff is its ringing behaviour which you can clearly see in the step response display. It also creates a bigger time delay.
The green curve is a Bessel filter of order 10. So the slope is the same like the Butterworth of 10th order. You certainly will notice that it looks like a compromise between the other curves. The transition for the amplitude response is more smooth. An interesting point is the step response. It has no overshoot despite the high filter order. But it also shows some delay.
Here we get to a point where we can see the advantages of the digital world. Now we are able to design crossovers according to our needs, it is not mandatory to follow the old tradition of analog passiv crossovers. We can compensate also for the delays. We can use linear phase crossovers. No problem to design crossovers with an subtractive approach.
At the end we find a way to use the best part of a given driver and to integrate it with other drivers in a proper way. This is my understanding of "cooking" as a method of synthesis 🙂
Attachments
Dear Uli,
Thanks for the pictures! Can I ask you if we can compensate this ringing with linear phase filter and your substraction XO method used in Acourate. Does it still matter whether we choose Butterworth or Bessels?
Thanks for the pictures! Can I ask you if we can compensate this ringing with linear phase filter and your substraction XO method used in Acourate. Does it still matter whether we choose Butterworth or Bessels?
The ringing of filters which you can see in a lowpass step response is a property of the filter type. So filters of order 1 do not ring and also filter of Bessel type. All other filters introduce more or less ringing. This is independant if it is a minimum phase or linear phase filter.
A higher order means a steeper slope. The Neville-Thiele filter is a filter which has quite a high slope but less ringing than e.g. a Butterworth with the same slope steepness.
So I prefer to use today either Bessel or NT filters.
A higher order means a steeper slope. The Neville-Thiele filter is a filter which has quite a high slope but less ringing than e.g. a Butterworth with the same slope steepness.
So I prefer to use today either Bessel or NT filters.
m0tion said:
Thanks for the interest m0tion,
Since my last post there been lots of tweaking, listening and enjoying.
This project is pretty much at a conclusion barring a few last photo's, measurements of the final performance, a summary of implementation and maybe some text on my thoughts about the outcome.
I've got a few amp related issues that I need to sort before I can really call the system finish but I will post back with those final thoughts, if only to bring the thread to a conclusion and let folks know how it turned out.
BTW I've had a small amount of interest from people wishing to hear the speakers and a couple have been already so if anyone is interested I'd be thrilled to give a relaxed and thoroughly informal demo, the drink and food is on me. My email is at the bottom of this post if your interested.
ShinOBIWAN said:
That's a neat project, I'd love to hear it.
If you don't mind, could you bring it over to my house? 😉
If you can't, then I'll just have to read other people's impressions after they've heard your new speakers.
Best Regards,
TerryO
I will put together some info on the finished design but in the meantime here's another shot.
Imaging isn't so good in this position but I'll maybe try others to improve the situation 😀
Imaging isn't so good in this position but I'll maybe try others to improve the situation 😀
An externally hosted image should be here but it was not working when we last tested it.
- Status
- Not open for further replies.