Too Tall wrote on Wed, 02 June 2004 09:49 |
John Sheerin wrote on Tue, 01 June 2004 15:10 |
Hi Curtis, Long weekend here - I'll just reply here instead of in the old thread.
I would recommend that you or someone else who actually has some LAB subs come up with these guidelines. What I would do is get Hornresp, use Brad Litz's model posted on his webpage: http://www.geocities.com/hulkss/V2input.html and calculate out what input voltages are acceptable. Then set your limiter appropriately and give it a try. To model multiple horns in Hornresp, you can double the throat and mouth sizes (or multiply as appropriate), as well as doubling, etc. the front and rear chamber volumes. Then use the built in tool to calculate composite parameters for multiple speakers. This will give you a good starting point for settings.
|
I have only glanced at several horn response programs and that was long enough for me to realize I had little understanding of the process and wouldn't know good from useless results. I would do it if I thought I had enough experience to pull it off, but I don't.
I realize putting down "hard" number for this looks like a fairly certain way of letting yourself in for a boatload of crap from anyone who ever lunches a driver. But in the end it isn't any different from the other figures stated in the specs for RMS and peak capabilities. They are just "guidelines" only this time they relate to the number boxes stacked together. In the end I understand if Eminence and SPL do not wish to undertake this. Hopefully one of the LAB sub owners that have the requisite experience will take on the task. BTW I just started playing with a new feature in Praxis that charts "Maximum level capabilities" for a speaker over a selected range for THD, single frequency 2nd and 3rd harmonics, peak clip and compression from a stimulus Bill Walso has named a "wavelet". It sounds like a "clunk" and a "pop" mixed into one sound. After I get a handle on this I will try talking Al into letting me give the LAB subs a ride. Too Tall
|
Hi Curtis,
Many seem to get all too worried about understanding exactly how a speaker is operating when all they really need is a conceptual understanding of how a system unloads or drops in output relative to points like tuning frequency and the low end loading of a horn. Even this is more "useful knowledge" not neccessary. The background knowledge just gives you a better ability to analyze the results and possibly start closer to the point of interest.
With some of the rather radical EQ I have applied to our B-DEAPs in home theater applications, I have had to determine what useful output levels are and where we're waisting our time and amplifier. Once in a small room, predictions and expected output gets shifted quite a bit, so it is easier to just rely on real measurements of the entire system. My preferred approach is to look for power compression and change in the shape of the response as output is increased. In looking at LAB subs the distortion limits you mention should also work well. The feature sounds a lot like an automated take on what Sigfried Linkwitz and Don Keele made common with a sine burst to determine maximum output before audible distortion.
I would suggest someone setting up with a measurement of the response from say 20-120Hz. Measure outdoors at say 10'-30' or 4-10m away from the configuration being tested. You will need to be far enough to allow your mic to comfortably measure the peak levels. Have a DSP in front of the system so you can test with various high pass settings while monitoring the test voltage across one sub. Start with a measurement at maybe 80-90dB, depending on how far you are from the subs. This is your reference by which to compare the subsequent measurements. Now start increasing the drive level in precise increments of 5 to 10dB until you are in the range of 40-50V across the subs. From here I would suggest increasing in smaller increments and to also watch your amplifiers. Increases in level of 2dB keep you at less than 2x the power per sweep, reducing the likelihood of damage before you see it coming. As you increase the level you should see the response change. In particular you will see the low end start to not increase the same amount by which you increased the level. You will likely also see some change in the response at low cut-off and mid-band.
The response changes from either or both of two factors. First is over excursion. Any conventional driver will have a fall off in motor strength as well as a stiffening of the suspension at excursion limits. Thermal heating of the voice coil will increase impedance(Re) and reduce current flowing through the motor, resulting in less motor strength which is observed as lower sensitivity and higher system Q. Note that as this loss of motor strength onsets some systems will actually increase output at over a small frequency range while loosing output at others before output no longer increases. In other words, you will in effect
see at what frequencies and at what drive levels the parameters are changing. This is a very good indication of the useful limits of the device you are testing.
Now, if you decide to do this, be warned that carelessness and curiosity can kill drivers. You should also
LISTEN for cries of distress before you increase the level further. So don't come back saying "Mark said I could... but I blew them up!"
So now what do you do with the information you see? Try to understand what is going on. If you first see the low end of the response shelf down and not increase any further, you can use that information to set your high pass filter. Set your filter, and drop down a few increments to establish a new reference curve, and increase the level again. If you set your filter properly, you will be able to sweep the passband to a higher level before you see significant changes to the response. You may even find that you can increase headroom by pulling 2-4dB from some frequency which overloads before the rest of the band.
Unfortunately the
ONLY way to determine maximum long term levels is to kill some drivers while closely monitoring what your input signal is and the duration of it. The problem is that any number you come up with will be entirely contigent on the bandwidth, spectral content and duration of the input signal, as well as dependent on the filter settings. Change the settings or the content and the point of failure changes. All you will ever really be able to get is a confirmation that the subwoofer lived to woof again under a given test condition and duration. The usefulness of this number will entirely depend on how the test compares to your typical usage.
In my oppinion, the above testing gives you a good indication of the well-behaved output limitations which should correlate rather well to maintaining consistent sound. There are of course all sorts of tricks you can play if you are willing to accept distortions from amplifiers and push the system to the limit constantly. Such conditions are a crap shoot which can only be characterized on a case-by-case basis. The whole point of a sub like the LAB is to keep yourself in the former operating range.
This got pretty lengthy, but I also want mention that you should be able to set up a similar test using any program with dual FFT capability. Basically are looking to compare the input signal to the amplifier or to the speaker (depending on if you want to include the amplifier behavior) where you can calibrate the two inputs to an appropriate display level. Changes in the response can only be from the response of the speaker changing. With this setup you could effectively watch loudspeaker compression in real-time.
My 10 cents,