I read:
The instantaneous part means to me "peak", so I would not divide the voltage by the square root of 2 - but maybe that just my reading of the term. Anyway, it is explained above - thanks.
Hi Barleywater
It’s worth pointing out that in audio two frames of reference are used and occasionally without realizing one isn’t the other.
The “peak value” as used in say with a signal the peak to average ratio, refers to the peak envelope value compared to the average value which is the long term heating value if one is dealing with heating power or a peak capture SPL meter.
If you use cool edit or other software, you can find these peak and average values for a given music track or short event.
For loudspeaker use, the “pink noise” or equal energy in all octaves noise signal has a Peak to average ratio of 6dB, this is used in many measurement systems is actually a little less dynamic range than a highly compressed FM station. Other systems use the old style 10dB peak to average ration which was the custom in RF and in this case more like a lot more like the dynamics of pop music.
In the latter case, a SLOW SPL meter has value X while the peak hold function is +10dB higher than X.
Some uncompressed music / hifi recordings can have 30dB or more peak to average but these sound “quiet” when played because the peak level is set by the medium while what we hear as loudness is much more like the average level (and why meter movement VU meters ignored short peaks).
OR, it could refer to the peak vs average value in a single sine wave or half sine wave.
But if one is dealing with sound specifications, one is stuck with the envelope definition while the true purest will examine the mic signal on a storage oscilloscope or the software equivalent.
For loudspeakers, the only real “peak spl” is the one you get when measure in a standard way, hopefully how it will be used and you convert the mic voltage.
Fwiw, 388 Watts with an 82dB sensitivity would produce 107.8dB in a perfect world, not 133dB.
10 X (Log (input power)) + 1W1M sensitivity = max output @1M in a perfect world.
133dB is actually harder than you might imagine to produce.
Take that driver, produce say 80dB spl at 1 meter (which is 68dB if you’re listening 4 meters from the speaker) , measure the harmonic distortion etc, raise the level 3dB and repeat.
Usually, somewhere around 1/10 to 1/8 the driver rated power (a figure which is derived in a near death test), you will see the response curve SHAPE start to change relative to the 1W response shape and this is when you are reaching the beginning of power compression or mechanical non-linearity etc.
Best
Tom
It’s worth pointing out that in audio two frames of reference are used and occasionally without realizing one isn’t the other.
The “peak value” as used in say with a signal the peak to average ratio, refers to the peak envelope value compared to the average value which is the long term heating value if one is dealing with heating power or a peak capture SPL meter.
If you use cool edit or other software, you can find these peak and average values for a given music track or short event.
For loudspeaker use, the “pink noise” or equal energy in all octaves noise signal has a Peak to average ratio of 6dB, this is used in many measurement systems is actually a little less dynamic range than a highly compressed FM station. Other systems use the old style 10dB peak to average ration which was the custom in RF and in this case more like a lot more like the dynamics of pop music.
In the latter case, a SLOW SPL meter has value X while the peak hold function is +10dB higher than X.
Some uncompressed music / hifi recordings can have 30dB or more peak to average but these sound “quiet” when played because the peak level is set by the medium while what we hear as loudness is much more like the average level (and why meter movement VU meters ignored short peaks).
OR, it could refer to the peak vs average value in a single sine wave or half sine wave.
But if one is dealing with sound specifications, one is stuck with the envelope definition while the true purest will examine the mic signal on a storage oscilloscope or the software equivalent.
For loudspeakers, the only real “peak spl” is the one you get when measure in a standard way, hopefully how it will be used and you convert the mic voltage.
Fwiw, 388 Watts with an 82dB sensitivity would produce 107.8dB in a perfect world, not 133dB.
10 X (Log (input power)) + 1W1M sensitivity = max output @1M in a perfect world.
133dB is actually harder than you might imagine to produce.
Take that driver, produce say 80dB spl at 1 meter (which is 68dB if you’re listening 4 meters from the speaker) , measure the harmonic distortion etc, raise the level 3dB and repeat.
Usually, somewhere around 1/10 to 1/8 the driver rated power (a figure which is derived in a near death test), you will see the response curve SHAPE start to change relative to the 1W response shape and this is when you are reaching the beginning of power compression or mechanical non-linearity etc.
Best
Tom
Tom,
Thanks for shake out of math. 10 log for power, 20 log for voltage.
Continuing with burst testing, the brick wall 1kHz low pass test signal is filtered with Linkwitz-Riley 1kHz 24dB/Oct high pass, and also with 1kHz 1024 point FFT. Equal normalization is done to both and a response recording is made. Referencing for displayed results is -6dB at just below 1kHz for LR4 response:
The LR4 high pass crossover filter results in generation of harmonic distortion and intermodulation distortion components, and in reality extend <1kHz. All of these show up as sum and difference side bands to desired pass band components.
FIR crossover results in lower level, well defined harmonic distortion generation.
The difference when crossover is implemented with LR4 v 1024 point FIR for my Pluto type speaker is night and day. Transient performance is truly cleaner. Sustained high levels are truly cleaner.
Regards,
Andrew
Thanks for shake out of math. 10 log for power, 20 log for voltage.
Continuing with burst testing, the brick wall 1kHz low pass test signal is filtered with Linkwitz-Riley 1kHz 24dB/Oct high pass, and also with 1kHz 1024 point FFT. Equal normalization is done to both and a response recording is made. Referencing for displayed results is -6dB at just below 1kHz for LR4 response:
The LR4 high pass crossover filter results in generation of harmonic distortion and intermodulation distortion components, and in reality extend <1kHz. All of these show up as sum and difference side bands to desired pass band components.
FIR crossover results in lower level, well defined harmonic distortion generation.
The difference when crossover is implemented with LR4 v 1024 point FIR for my Pluto type speaker is night and day. Transient performance is truly cleaner. Sustained high levels are truly cleaner.
Regards,
Andrew
Barleywater,
Thanks for clearing up the details.
I think you would find (good) compression drivers to be quite clean at the same 133 dB SPL as you estimate your 82dB/1watt cone driver to be at 12".
With about 2 watts of input I measured 131.7 dBA at 13" from a B&C DE82TN 1.4” exit 3” diaphragm driver crossed at 1250 Hz (LR24) using a pink noise signal. At 6.5" from the driver's screen the level was 136.8 dBA.
At 2 watts, the driver is quite clean, at 128 watts, not so clean, but the level would be 167.7 dBA.
Probably quite a bit louder than slapping the palm of your hand against your ear, and smoke would emit from the cone driver long before that level
.Art
Last edited:
Art i agree. The one benefit of horn drivers, the low distortion due to their high efficiency can be wasted, even lower THD. (ive not been impressed yet, but im still young). The reason i am öt keen is due to the heavy duty diaphragm and voice coil former etc which is necessarmx for their intended use.
I'm interested in what Barleywater is doing with the steep filters in terms of power handling and distortion. So far, I have not been crazy about the brickwall crossovers I've heard, but I have an open mind and ears.
If better power handling can be demonstrated for drivers like the little fullrange, that would go a long way in design. Getting clean, in room peaks at 105-110dB out of a "tweeter" like that would be great. Not breaking it would be even better.
What would be a good test level at 1M to simulate in-room peaks just above 100dB SPL? Remembering that two drivers would be used to achieve those levels in-room.
If better power handling can be demonstrated for drivers like the little fullrange, that would go a long way in design. Getting clean, in room peaks at 105-110dB out of a "tweeter" like that would be great. Not breaking it would be even better.
What would be a good test level at 1M to simulate in-room peaks just above 100dB SPL? Remembering that two drivers would be used to achieve those levels in-room.
A tip for anyone wanting to do some experiments with active speakers without damaging them:
If the speaker has a port, simply disconnect the spade connectors at the drive units and thread your temporary cables in through the port(s). You can put a protection cap in series with the tweeter in the cable.
Just did this to a pair of B&W speakers in under an hour. Sounding good, but then they sounded very good in passive form too.
Kinda tempted to arrange a couple of relays so I can instantly swap between active and passive...
If the speaker has a port, simply disconnect the spade connectors at the drive units and thread your temporary cables in through the port(s). You can put a protection cap in series with the tweeter in the cable.
Just did this to a pair of B&W speakers in under an hour. Sounding good, but then they sounded very good in passive form too.
Kinda tempted to arrange a couple of relays so I can instantly swap between active and passive...
In case anyone is interested, here are a couple of basic measurements of speakers before and after conversion to active (passing cables in through the ports!). My regret is that I didn't do anything to measure the insertion loss of the passive crossover. I'll do it when I convert them back.
In both cases the measurement is of a single speaker at about 1m in the listening room, traces averaged with 1/3 octave smoothing, using Room EQ Wizard and a WM61a mic with cal file off the WWW. The active crossover is your basic Butterworth-style linear phase done by a PC, without any phase or delay tweaking of any kind. 'Voicing' was highly involved and took almost 5 minutes. (I merely set the amp volume controls to get similar-ish traces between passive/active.)
The B&W DS2, is a two way, and I used an active crossover frequency of 2800 Hz 6th order. Passive is the red trace, and active the blue. Not a lot of difference!
The Tannoy R2 is a three way, and I used active crossover frequencies of 200 Hz and 2800 Hz, both 9th order. Passive is the magenta trace and red the active.
I've had a bit of a speaker-measuring, tweaking and listening binge this last couple of days, including writing my own software to correct individual drivers in various ways. What seems clear is that the basic character of the speaker comes through the passive-to-active conversion, and that a flat frequency and phase response does nothing to help a truly crap speaker!
Anyway, the Tannoys are sounding rather wonderful, now.
In both cases the measurement is of a single speaker at about 1m in the listening room, traces averaged with 1/3 octave smoothing, using Room EQ Wizard and a WM61a mic with cal file off the WWW. The active crossover is your basic Butterworth-style linear phase done by a PC, without any phase or delay tweaking of any kind. 'Voicing' was highly involved and took almost 5 minutes. (I merely set the amp volume controls to get similar-ish traces between passive/active.)
The B&W DS2, is a two way, and I used an active crossover frequency of 2800 Hz 6th order. Passive is the red trace, and active the blue. Not a lot of difference!
The Tannoy R2 is a three way, and I used active crossover frequencies of 200 Hz and 2800 Hz, both 9th order. Passive is the magenta trace and red the active.
I've had a bit of a speaker-measuring, tweaking and listening binge this last couple of days, including writing my own software to correct individual drivers in various ways. What seems clear is that the basic character of the speaker comes through the passive-to-active conversion, and that a flat frequency and phase response does nothing to help a truly crap speaker!
Anyway, the Tannoys are sounding rather wonderful, now.
Attachments
I assume the bad passive is the 3way? One thing that put me off building a 3 way is the inversion of 1 driver, eg midrange, to align phase at xo points with a 2nd order passive network. Its bad enough in 2ways with a 2nd order network, aligning phase at Fc but sacrificing the original phase in the higher harmonics. Active for 3 ways and above makes perfect sense, but I feel its less significant in 2ways or a fullrange system. My stuck-in-the-dark-ages opinion is simply: The better a driver is, the lesser the need for EQ. Treat the cause to cure the symptom.
1/3 octave smoothing is not revealing, and graph scale is tiny. What of reverse null response of crossovers? And what forms of waveform temporal distortions are revealed with burst testing? Distortion?
Phase response of your DSP crossovers may be flat, but what does REW show for acoustic response? Phase mayhem starts with coupling capacitors, including active capacitance of speaker diaphragms along with capacitors found in coupling between amplifier stages, as well voice coil behavior.
What? Low distortion due to their high efficiency? Whole principle of compression driver and horn is harnessing non linear properties of sound in high pressure air. The lower the distortion, the less efficient.
Regards,
Andrew
You know far more about this stuff than I do - I've never even contemplated building a passive crossover.
However, after experimenting with active I'm beginning to get the feeling that yes, if the drivers aren't basically right, you can't do a lot with EQ. I would guess that an anomaly in a driver's frequency response also corresponds to phase, and other (possibly uncorrectable) anomalies caused by 'natural' physical factors. I wonder if that means that your brain naturally accepts and 'hears through' it. Messing about with EQ may end up just making it sound weird, even though the results look good on a graph.
That's very interesting - I had an idea it was the case as I rummaged about in it. So in the original configuration, the middle driver is also doing a fair bit of bass as well..? Why use this configuration rather than a true three way? Is it the difficulties in designing the crossover that mondo mentioned earlier?
Do you have any experience of this speaker yourself? I'd be interested in your listening impressions.
I bow to your expertise. My crossover software is capable of correcting phase and amplitude if only I can make the right measurements. However, I have my doubts that I'm going to hear a huge difference, compared to the phase mayhem that's occurring in the room, or as I move around relative to the drivers. I was greatly admiring your graphs in an earlier post, but do you still get a square wave if you move the mic a few millimetres?
I was reading an old HiFi World review of the similar R3 last night and you can see the LF crossover in the reponse and imedance curves. It allows the use of narrower cabinets and gives easy (excessive) baffle step compensation. The reviewer preferred to use the foam bung in the port, even though the port was broadly tuned and largely flattened the upper bass impedance peak. One thing about the R3 was that the bottom of the cabinet was not part of the true volume, being for sand loading.
Coherence once formed is maintained in direct sound, microphone may be moved significant distance, and gated response capturing direct sound reveals this.
Yes, getting good corrections requires good measurements.
Mayhem in the room is status quo for evolutionary development of hearing. Source coherence makes it easier for mind to tie reflections to it rather than having to apply more attention to assess if new source may have entered room. The more intact a reproduced sound's harmonic phase structure, the more plausible the reproduction.
Regards,
Andrew
Yes, getting good corrections requires good measurements.
Mayhem in the room is status quo for evolutionary development of hearing. Source coherence makes it easier for mind to tie reflections to it rather than having to apply more attention to assess if new source may have entered room. The more intact a reproduced sound's harmonic phase structure, the more plausible the reproduction.
Regards,
Andrew
Yes Falvio, I agree. For me passive have certain practical advantages. It took me many years and much listening to admit that they can be just as good (sometimes better) than active crossovers. Yes, I was an active crossover snob!
But nothing on your new website
But nothing on your new website
and I thought, new website, I had not heard anything about that
and then I notice latest news ...2008
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.