This is precisely Floyd's point as well. From my perspective two channel is far from being surpassed by multichannel as the major source of musical recordings in the home. As such, I have not paid very much attention to surround, except in film and even there I don't consider a major aspect. That surround has the "potential" to change things is probably true. That it has made any real impact as yet is certainly not true.
As I have mentioned before, I was agast to discover that my receiver was going into 2 channel mode because it couldn't understand the signal stream from a Blue Ray disk. This meant no center and no surrounds. I first noticed no center, but even that was not completely obvious at first. Then I noticed that there weren't any surrounds either but those I had hardly missed. My point here is simply that 2 channel goes a long way before more channels are missed. Thats why I don't see 2 channel going away any time soon, if ever. It is ideal for mobil listening on headphones - which seems to be the defacto standard for recordings these days.
Hi Earl,
A comment and a question about some of Sean Olive's latest work with in-room response measured steady-state and 'room correction' algorithms.
The comment, he seems to accept different directivity speakers. He says they should all have reasonably flat direct sound at the listening position, either anechoic or quasi-anechoic. He also says the highs will be rolled off in a steady-state measurement and says, with a perfect room and speaker, you'd see a reasonably straight-line roll-off with no big wiggles. The slope of the roll-off will depend on the directivity.
The question, Harman is doing spatial averaging over several mic positions to measure the in-room response but Sean won't answer questions about how they average. What do you think is the best way to do that? Here are snippets of a couple of posts I made about my own experiments after not getting any answers about that.
--------
Averaging dB is no good. A null should be ignored but it pulls down a simple average too much.
Power averaging, 10*log((10^(db1/10) + ...)/n) ignores the nulls pretty well but it seems to give too much weight to a single peak.
Pressure averaging 20*log((10^(db1/20) + ...)/n) seems better. It ignores the nulls and doesn't give as much weight to a single peak. That's what you'd get if you ran several mics into a mixer and averaged their voltage.
You could take the log thing even further in ignoring the extremes and use 30 instead of 10 or 20. Hmmmm....
--------
Following up on my previous rant about averaging mic readings, here's a little copy paste from my spreadsheet. With 4 of the measurements being 80dB, one with a 90dB peak, and one with a 40dB null, the fuzzy logic between our ears tells us the average used in room correction calcs should be somewhere near 80dB. The 30log average looks pretty good.
A comment and a question about some of Sean Olive's latest work with in-room response measured steady-state and 'room correction' algorithms.
The comment, he seems to accept different directivity speakers. He says they should all have reasonably flat direct sound at the listening position, either anechoic or quasi-anechoic. He also says the highs will be rolled off in a steady-state measurement and says, with a perfect room and speaker, you'd see a reasonably straight-line roll-off with no big wiggles. The slope of the roll-off will depend on the directivity.
The question, Harman is doing spatial averaging over several mic positions to measure the in-room response but Sean won't answer questions about how they average. What do you think is the best way to do that? Here are snippets of a couple of posts I made about my own experiments after not getting any answers about that.
--------
Averaging dB is no good. A null should be ignored but it pulls down a simple average too much.
Power averaging, 10*log((10^(db1/10) + ...)/n) ignores the nulls pretty well but it seems to give too much weight to a single peak.
Pressure averaging 20*log((10^(db1/20) + ...)/n) seems better. It ignores the nulls and doesn't give as much weight to a single peak. That's what you'd get if you ran several mics into a mixer and averaged their voltage.
You could take the log thing even further in ignoring the extremes and use 30 instead of 10 or 20. Hmmmm....
--------
Following up on my previous rant about averaging mic readings, here's a little copy paste from my spreadsheet. With 4 of the measurements being 80dB, one with a 90dB peak, and one with a 40dB null, the fuzzy logic between our ears tells us the average used in room correction calcs should be somewhere near 80dB. The 30log average looks pretty good.
Code:
90
80
80
80
80
40
dB 10log 20log 30log
75.0 83.7 81.6 80.4
Last edited:
Dennis
Good post. You are finding that things can get pretty flaky when there are large differences in levels, that's quite true. But what we found was that all of these problems go away as the range of levels decreses - as they should. We tested all of the different techniques and found that they all gave about the same answers when the FR was fairy smooth. But when things went awry, the numbers did too. To me it's a matter of "how bad is bad?" - always hard to quantify. In the end we stopped worrying about it because when things get better it doesn't matter and since making things better was the goal the problems diminish as time goes on.
At LFs there is likely to be a greater problem than at HF. Some frequency averaging is necessary to get away from very sharp peaks and dips that can cause trouble. With a little freq averaging before doing the spatial average I don't find any major issues.
Hey Brandon,
It sorta did but it's only dealing with measurements in an anechoic chamber -- what angles to measure to simulate early reflections, etc. And it doesn't explain how they average the measurements over several angles, just says they do. If you're measuring in a room, you can just gate the impulse to get that stuff instead of having to simulate it.
It sorta did but it's only dealing with measurements in an anechoic chamber -- what angles to measure to simulate early reflections, etc. And it doesn't explain how they average the measurements over several angles, just says they do. If you're measuring in a room, you can just gate the impulse to get that stuff instead of having to simulate it.
The "normal" solution to the directivity problem is to use a waveguide. But I've noticed that large pistons beam as well, and some of them have nicely controlled directivity.
Why not use a small woofer as a tweeter, instead of a dome?
I was looking at the response graph of the $10 Peerless computer speakers, and noticed that their off-axis response is relatively well behaved.
Thoughts?
Why not use a small woofer as a tweeter, instead of a dome?
I was looking at the response graph of the $10 Peerless computer speakers, and noticed that their off-axis response is relatively well behaved.
Thoughts?
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
The directivity is constantly changing. Its a very well behaved piston, a great looking speaker actually, but as with ALL pistons, the beam changes with frequency. I don't think that this a good idea in a premium speaker. Its useful in some situations, but not as a main. A waveguide on that speaker might work well - thats workable.
I've been thinking along similar lines with waveguides for small full range drivers.
I think if one took care you could get a directivity roughly similar to Dr. Geddes's designs (and a similar low crossover for a 2 way).
Obviously it'd suffer various waveform problems compared to a compression driver even with a carefully designed phase plug, as well as much more limited dynamics. But considering small full range drivers of reasonable quality are ~$20 from Aura, Peerless and others, it might be a decent value.
I think if one took care you could get a directivity roughly similar to Dr. Geddes's designs (and a similar low crossover for a 2 way).
Obviously it'd suffer various waveform problems compared to a compression driver even with a carefully designed phase plug, as well as much more limited dynamics. But considering small full range drivers of reasonable quality are ~$20 from Aura, Peerless and others, it might be a decent value.
I appreciate the information, but is it proper to promote products in this kind of thread?
Well, it is about waveguides, right
You may say its about OS waveguides in general
Throath looks strange and different from OS
But tell us how much it cost
Would be relevant to judge what kind of thing this is
But when you say "nearest supplier" indicates a stricktly commercial product made fore profit
If so it belong in vendors
You may say its about OS waveguides in general
Throath looks strange and different from OS
But tell us how much it cost
Would be relevant to judge what kind of thing this is
But when you say "nearest supplier" indicates a stricktly commercial product made fore profit
If so it belong in vendors
Last edited:
Throat is longer so we can cut to desired diameter. Drawing and explanation below:
http://www.diyaudio.com/forums/multi-way/103872-geddes-waveguides-358.html#post1991238
http://www.diyaudio.com/forums/multi-way/103872-geddes-waveguides-358.html#post1991238
You could get the directivity, but not over the same bandwidth. Thats what the phase plug buys you - another octave or two. Its an inexpensive driver approach, but I'm not sure that it saves money in the end if you need another driver and waveguide to cover the bandwidth.
Most rooms are made well damped, not live like I suggest. And most rooms don;t have enough LF absorption. Basically, the standard technique for a bad room is to put a lot of absorption arround. This tends to damp the HFs but not the LFs. This makes the room dead, and the bass boomy. In that situation a wider directivity speaker might just add some "life" back into the room. But the imaging is going to be off.
Hello jzagaja,
Looks very good. Do you have some directivity measurements you could post for your waveguide? When the waveguide is cut to the correct diameter, do you mount a flange to attach the driver?
Hello jzagaja,
The top of your cabinet does not look the best choice to me:
- there seems to be a sharp edge => diffraction
- distance between wavequide and this edge remains contant for the top part => effect concentrated at 1 specific frequency (kinda like mounting a round speaker in the center of a round baffle)
In what material will the cabinet be constructed?
The top of your cabinet does not look the best choice to me:
- there seems to be a sharp edge => diffraction
- distance between wavequide and this edge remains contant for the top part => effect concentrated at 1 specific frequency (kinda like mounting a round speaker in the center of a round baffle)
In what material will the cabinet be constructed?