Re: Creating impulse responses
- From: Bob Cain <arcane@xxxxxxxxxxxxxxxxx>
- Date: Sat, 06 May 2006 01:33:11 -0700
Don Pearce wrote:
On 5 May 2006 05:59:39 -0700, "David Spearritt"
<djspearritt@xxxxxxxxx> wrote:
None of the impulsive sources are good for a decent response
measurement, there's simply not enough energy to excite the room and
get enough S/N ratio. You are best off with a swept sine.
A swept sine is absolutely the last thing you want. It sets up and
maintains standing waves, which leave you absolutely nowhere.
Sorry, but this is wrong. The cross correlation performed between the
response and the reference deals with standing waves just fine. Were you
to remove them you would not be accurately measuring the IR of what causes
them.
The optimum is probably an MLS.
It's good too but it is very sensitive to distortion whereas the swept sin
effectively eliminates it.
The problem with both of these, of course, is that they involve a
reproduction system to generate them, so there is an in-built
uncertainty of what exactly you have. A second problem is that they
are not omnidirectional at all frequencies - something that is
reasonably easy to obtain with a spark.
Absolutely. But the driver directionality may be ok depending on the
application and the characteristics of the driver can be measured
separately and mathematically factored out of the room measurement.
Of course with the spark you
don't fire it once, capture it and analyse. You fire it repeatedly and
average the captured waveforms until you have enough to smooth out the
noise.
I've experimented with this using a spark gap generator I made from a
modified (lots more capacitance) strobe light and found that there is just
too little low frequency energy for signal to rise much above noise however
many iterations you average (within reason, of course.) There are also
difficulties with the averaging unless the trigger is synchronized with
respect to the sample clock edges. There are computational methods to do
that using fractional period delays based on an average group delay
calculation but it is computationally very heavy and a pain in the ass to
wait for.
Another impulse option is exploding wire. You suddenly dump lots of high
voltage energy into a thin wire and its sudden vaporization packs the SPL
of a shotgun. I've built one from a kit but chickened out of turning it on
because it stores enough 5kV energy in a large capacitor bank to be
absolutely lethal in the event of an accident or of momentary stupidity.
Until I considered my skin I thought it would be ideal because of the high
level of the signal over ambient noise and the fact that the wire vaporizes
in time less than an audio sample period. That the wire is vaporized and
its mass ejected very rapidly yields much more LF content than a spark
which is just a heating of the air.
Bob
--
"Things should be described as simply as possible, but no simpler."
A. Einstein
.
- References:
- Creating impulse responses
- From: Carey Carlan
- Re: Creating impulse responses
- From: Don Pearce
- Re: Creating impulse responses
- From: Carey Carlan
- Re: Creating impulse responses
- From: Don Pearce
- Re: Creating impulse responses
- From: David Spearritt
- Re: Creating impulse responses
- From: Don Pearce
- Creating impulse responses
- Prev by Date: Re: Why do these mikes exist?
- Next by Date: Re: Why do these mikes exist?
- Previous by thread: Re: Creating impulse responses
- Next by thread: Re: Creating impulse responses
- Index(es):
Relevant Pages
|
Loading
