Absolutely, I agree completely. But one thing I realized is that if the speaker is large enough that baffle transition happens below the Schroeder frequency, then it should probably be treated differently than if it happens higher in frequency. Since the cause is changing directivity, and since directivity in the modal region is ambiguous, I'm not sure that electronic equalization makes much sense. So I would not suggest baffle step compensation on a physically large speaker used indoors. Rather than throwing more power at it, I think it would be better to use a multisub approach. You can smooth the modal range and the ripples from baffle diffraction with the same method, killing two birds with one stone.
You're right, of course. The feature I was focused on was self-interference, and I simplified for brevity. My point is that I've found adding sources is a very useful way to counter floor bounce, similar to the multisub approach but a little higher in frequency. Vertical arrays do not suffer floor bounce for this very reason.
Same thing with 2.5-way speakers, when the "helper" woofer is allowed to blend with the midwoofer through the region where floor bounce would likely occur otherwise. Or use what I sometimes refer to as "flanking subs", essentially the same as a 2.5-way, just subs placed fairly close to the mains but in a different vertical position, allowed to be blended with the mains through the floor bounce region. Naturally, these have to be well behaved in that region for quality to be good, but that's not usually a problem.
Similarly, a large format midrange that is allowed to reach fairly low, overlapping with a woofer that reaches fairly high - and I'm talking overlap in the 100Hz to 300Hz range or so - this approach works well to mitigate floor bounce, allowing the midrange to be placed at ear level and still not suffer a floor bounce notch.
I'm curious why I never see reference to using an "anechoic microphone" . . . ie recessing the microphone in a sound absorbing box or tube (of "rigid" fiberglass or rockwool, for example, 2 ft x 2 ft x 4 ft, open at the smaller ends) that would absorb first reflections before they reach the microphone itself. I have seen this done for testing the high frequency response of microphones (where it reduces the whole apparatus to a convenient tabletop size and significantly reduces problems from environmental noise), and can't see any conceptual reason why it shouldn't work for speaker testing.
I don't know of any with a tight enough pattern that are reasonably flat on-axis and reasonably cheap. Any suggestions ? ? ?
Deward, what parameters should we look for that let you(and potentially us) know your speakers are removing the effects the recording process?
thanks,
Dan
thanks,
Dan
It's certainly possible and you've hit on the methodology.
I know a couple of professional designers that just lift the speaker up on a forklift with a mic on an extention pole to get better measurements. Just one more reason not to sell that forklift you haven't used much recently.
Best Regards,
TerryO
LOL . . . and assuming you're someplace where it's (occasionally) quiet outside . . .
Another reason that you always keep your night-vision googles handy. Actually, the testing outside is usually taking place at about 3 pm. That gives you enough time to get set up after you close down the local bars.😀
Best Regards,
TerryO
Alas, none that are reasonably priced, no. I use a cardioid and like it, but a tighter pattern would be better.
Those are all valid concepts in handling floor bounce.
In my designs i usually account for the floor as an
integral part of the speaker.
e.g.
Dipol 08 open baffle line array, Strahlerzeile mit offener Schallwand
A floor standing speaker is usually not thought to
be elevated, so i cannot even see the need for
measurements cancelling floor bounce.
For development of drivers (transducers) itself, this
is different. Since i also make bending wave transducers,
i want to know their impulse response without room
boundaries if possible.
But in case of a completed speaker ...
A speaker having proper impulse response on a fork
lift, but is suffering from floor bounce - floor standing
being its manner of operation - is good for nothing.
With the panel form bending wave transducers similar
solutions like with arrays or "bounce filling 2.5 ways" can
be found, since the propagation of bending waves
is dispersive: Above the coincidence frequency of the
panel, the bending wave can reach the floor faster
via the panel, than components radiated directly from
the exciter via airpath.
That means floor bounce can be dealt with by adjusting
the positions (height above floor) of the exciter(s) and
the coincidence frequency of the panel to mitigate floor
bounce.
Again the floor is seen as an integral part of the
speaker. I can hardly imagine a speaker performing
independently from floor elevation (and listening distance).
e.g.
http://www.quadesl.com/graphics/quadGraphics/quad63_frequencyresponse.jpg
So even measurement of complete floorstanding speakers
without floor does not really make sense to me, yielding results
which are purely theoretical.
Last edited:
I think diffraction modeling tools are better than they're getting credit for, but they aren't great.
Here's a careful comparison of several free modeling tools, against anechoic measures from Seas. I suspect that the hash above 2 kHz is reflection of a tweeter off the cone in the box, but can't validate.
Windowing:
The biggest problem with short time windows is less the lowest valid frequency, but more the resolution. Lowest valid frequency = frequency resolution of measure. i.e. each FFT bin is an "average" with this bandwidth. Then there is the issue that some windows provide more smoothing, others let more junk from nearby frequencies bleed into and affect the amplitude at a different frequency.
Dave
<> ABBA
Here's a careful comparison of several free modeling tools, against anechoic measures from Seas. I suspect that the hash above 2 kHz is reflection of a tweeter off the cone in the box, but can't validate.
Windowing:
The biggest problem with short time windows is less the lowest valid frequency, but more the resolution. Lowest valid frequency = frequency resolution of measure. i.e. each FFT bin is an "average" with this bandwidth. Then there is the issue that some windows provide more smoothing, others let more junk from nearby frequencies bleed into and affect the amplitude at a different frequency.
Dave
<> ABBA
Attachments
Hello,
Line array microphone!
DIY with condenser capsules + summing amplifier.
Anyone?
- Elias
Line array microphone!DIY with condenser capsules + summing amplifier.
Anyone?
- Elias
Hello,
Overlay impulse responses of single microphone
taken at dedicated points in space ?
This way you can simulate any desired mic array.
Hello,
I think best to use time-freq resolution that matches the ear prosessing bandwidth. I've been using Bark. Within a critical bandwidth regardless of the ripple the perceived loudness remains the same if the total energy remains the same. But ripples look bad for the eye! But sometimes the ear don't perceive. The why to present the ripples in the response? To make a wrong conclusion? No thanks.
- Elias
I think best to use time-freq resolution that matches the ear prosessing bandwidth. I've been using Bark. Within a critical bandwidth regardless of the ripple the perceived loudness remains the same if the total energy remains the same. But ripples look bad for the eye! But sometimes the ear don't perceive. The why to present the ripples in the response? To make a wrong conclusion? No thanks.
- Elias
Even the best super-cardioids are no more that 10dB down at 90 degrees, and only 5-6dB down at 60 degrees. It should be possible to do *much* better than that with a few panels of fiberglass/rockwool . . .
I don't get it. As Markus said, gating is virtually perfect above some frequency and there are other ways below this transition. There are in fact ways to do measurements in reflective spaces that completly ignore reflections (See Weinreich in ASA about 50 years ago). That it is "not possible" is completely false. That it takes some skill to get right is correct.
Directional microphones are not really the answer since they don't work well enough. Signal processing is the answer - like gating, for example.
Directional microphones are not really the answer since they don't work well enough. Signal processing is the answer - like gating, for example.
I agree with you. I think anechoic measurements (and/or pseudo-anechoic measurements) are useful but I think room effects should be understood and addressed too.
I like having anechoic measurements to know what the speaker does without room influence. But I also like to know what the speaker will do when installed as intended, which means I will project my ideas about what effects the environment will have. This requires some assumptions on my part, but they aren't entirely speculative. As you've rightly said, the designer can know with some reasonable certainty that the speaker intended for home hifi use will be in proximity to a floor, walls and ceiling. So I think designing for specific placement makes a lot of sense.
- Status
- Not open for further replies.