Thanks Chris! Which CD and horn were you using?
There clearly seems to be some extra delayed energy in the impulse response, but what this means I'm not so sure...
Cheers,
Andreas
Please understand that what you are trying to do will not be simple. You have a multipath problem and you will have to understand how one separates different signals in that case. You will have to write code that identifies signals based on timing, not on spectrum. As experienced as I am at signal processing, this is not a task that I would take on lightly.
Basically, I would do things differently. I believe that it would be easier to identify the HOM from far field data, at least those that count, rather than near field, because the near field has things happening that are far more complex than the far field. And since many near field effects do not propagate, one is not concerned with those anyways.
If you are willing to do the measurements, and they will not be easy, then I am willing to help. But I don't see the task, as you have presented it, as having much potential.
Basically, I would do things differently. I believe that it would be easier to identify the HOM from far field data, at least those that count, rather than near field, because the near field has things happening that are far more complex than the far field. And since many near field effects do not propagate, one is not concerned with those anyways.
If you are willing to do the measurements, and they will not be easy, then I am willing to help. But I don't see the task, as you have presented it, as having much potential.
Hi Earl,
The next step would be to throw some foam in and look for differences. I've just finished a 30ppi block that fits nicely in the horn. If there are internal reflections, they'll be more attenuated by the horn, so I'd expect a click response that looks more like the raw CD at the tail end. While I agree the problem is complex, at the end of it all we're trying to cut down on internal reflections within the horn itself, so even just some visible improvement to the waveform is a good start (IMO) with minimal test gear and expertise.
I'm interested in the measurements you suggest. At the moment I have access to a reasonable-sized grassy back yard, and also a couple of bits of 18mm ply around 3'x4' each that I'm willing to sacrifice. I'm a reasonably competent woodworker, so sinking the horns into the baffle would be easy enough.
I have REW, Audacity, and the test gear (including an impedance jig), compression drivers and horns mentioned in my previous post. Would that lot be adequate?
If not, what else would be needed?
Not sure if it'll be of interest, but I also have some cheap pro audio coaxials that ought to be quite problematic (sharp edges in a couple of places in the horn path), and I'd be happy to experiment with those, too.
In fact, I might go and "click" those to see what a bad horn (coaxials) vs half-decent horn (RCF HF94) looks like.
Chris
The next step would be to throw some foam in and look for differences. I've just finished a 30ppi block that fits nicely in the horn. If there are internal reflections, they'll be more attenuated by the horn, so I'd expect a click response that looks more like the raw CD at the tail end. While I agree the problem is complex, at the end of it all we're trying to cut down on internal reflections within the horn itself, so even just some visible improvement to the waveform is a good start (IMO) with minimal test gear and expertise.
I'm interested in the measurements you suggest. At the moment I have access to a reasonable-sized grassy back yard, and also a couple of bits of 18mm ply around 3'x4' each that I'm willing to sacrifice. I'm a reasonably competent woodworker, so sinking the horns into the baffle would be easy enough.
I have REW, Audacity, and the test gear (including an impedance jig), compression drivers and horns mentioned in my previous post. Would that lot be adequate?
If not, what else would be needed?
Not sure if it'll be of interest, but I also have some cheap pro audio coaxials that ought to be quite problematic (sharp edges in a couple of places in the horn path), and I'd be happy to experiment with those, too.
In fact, I might go and "click" those to see what a bad horn (coaxials) vs half-decent horn (RCF HF94) looks like.
Chris
Hi,
i've just started playing with the foam plugs, and beginning to look at impulse response differences. (test bed: BMS coax on XT1464)
Also been trying to understand Dave Gunness's whitepapers and patents ...and what he calls temporal eq. Excellent resource material IMO.
Here's one whitepaper for example, that appears to be very much about HOMs and their measurement ........yes???
http://fulcrum-acoustic.com/assets/...dspeaker-transient-response-with-dsp-2005.pdf
Temporal EQ (TQ) | Fulcrum Acoustic
i've just started playing with the foam plugs, and beginning to look at impulse response differences. (test bed: BMS coax on XT1464)
Also been trying to understand Dave Gunness's whitepapers and patents ...and what he calls temporal eq. Excellent resource material IMO.
Here's one whitepaper for example, that appears to be very much about HOMs and their measurement ........yes???
http://fulcrum-acoustic.com/assets/...dspeaker-transient-response-with-dsp-2005.pdf
Temporal EQ (TQ) | Fulcrum Acoustic
Chris
Some years back I was able to show a difference in impulse responses in the far field with and without the foam. It showed a lowering of a delayed peak in the impulse and I think that this might have been an HOM. But I am suspect of some of the claims being made by others as to having measured them and "fixed" them. For example, in the Gunness paper, the very first part of the paper is incorrect so I can't take the rest of it seriously.
As to what you would need to do a "good job", I would say that you have all the equipment that it would take to do things right. Putting a waveguide in an infinite plane gets one what is required. You don't have an infinite plane, but you have a pretty large baffle. Some sound absorption at all the edges to prevent diffraction and you have a "virtual" infinite plane. One then expands the sound field into radiation modes and looks at the actual velocity contours in the plane of the mouth. We know what these should be if there were no HOM and any difference would be due to HOMs.
These measurements are not too hard to make, but the analysis is tricking because the data is what is called "ill-conditioned". In doing the reverse problem - i.e. what are the aperture modes given the field - one must deal with a nearly singular situation (the forward problem is not ill-conditioned, namely what is the sound field given the aperture velocities.) There are ways to deal with this, but it is not just your every day algebra problem. But if you get me the data, I'll see if I can analyze it.
Some years back I was able to show a difference in impulse responses in the far field with and without the foam. It showed a lowering of a delayed peak in the impulse and I think that this might have been an HOM. But I am suspect of some of the claims being made by others as to having measured them and "fixed" them. For example, in the Gunness paper, the very first part of the paper is incorrect so I can't take the rest of it seriously.
As to what you would need to do a "good job", I would say that you have all the equipment that it would take to do things right. Putting a waveguide in an infinite plane gets one what is required. You don't have an infinite plane, but you have a pretty large baffle. Some sound absorption at all the edges to prevent diffraction and you have a "virtual" infinite plane. One then expands the sound field into radiation modes and looks at the actual velocity contours in the plane of the mouth. We know what these should be if there were no HOM and any difference would be due to HOMs.
These measurements are not too hard to make, but the analysis is tricking because the data is what is called "ill-conditioned". In doing the reverse problem - i.e. what are the aperture modes given the field - one must deal with a nearly singular situation (the forward problem is not ill-conditioned, namely what is the sound field given the aperture velocities.) There are ways to deal with this, but it is not just your every day algebra problem. But if you get me the data, I'll see if I can analyze it.
Quick browse though the first paper, the charts look like wavelet charts. Yes, I think this reveal issues that seem to associate with reflection from mouth termination. I looked at some past measurements doing wavelet transformed, that is what I interpret from what I saw.
Sent from my iPhone using Tapatalk
Hi Earl,
I'm happy to do the measurements. Let me know exactly what needs doing and I'll find a nice patch of weather and get cracking.
I'll do measurements of the 8" coaxial unit, too, if the process doesn't take a full day. There's some sharp edges in that unit that might provide something more obvious to look for in the data.
Chris
I'm happy to do the measurements. Let me know exactly what needs doing and I'll find a nice patch of weather and get cracking.
I'll do measurements of the 8" coaxial unit, too, if the process doesn't take a full day. There's some sharp edges in that unit that might provide something more obvious to look for in the data.
Chris
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.