A short while ago I did some simulations on an MTM design to see what degree of lobing would result in the vertical plane for a given Xover frequency and order.
The one thing that concerned me though was that with odd order BW Xovers there may be a risk of the 3dB higher level at the Xover frequency causing significantly more power dissipation in the tweeter than a LR4 Xover.
To check out this issue, I put together a Spice netlist to simulate the dissipation in a tweeter via various types of Xover.
As a signal source, I chose a pink noise generator, but I’m not sure if normal music program material has the same spectral density as pink noise. Can anybody advise me on this or if they know of a better one?
Peter
The one thing that concerned me though was that with odd order BW Xovers there may be a risk of the 3dB higher level at the Xover frequency causing significantly more power dissipation in the tweeter than a LR4 Xover.
To check out this issue, I put together a Spice netlist to simulate the dissipation in a tweeter via various types of Xover.
As a signal source, I chose a pink noise generator, but I’m not sure if normal music program material has the same spectral density as pink noise. Can anybody advise me on this or if they know of a better one?
Peter
Pink noise is octave-based as is our hearing and most music/musical instruments/their harmonics.
It is much closer to what and how we hear than white noise although obviously it will not perfectly represent all music.
If you listen to pink noise to us it sounds as if the noise is even across all frequencies, on the other hand white noise sounds like pure hiss (all treble) to us.
Mind you if a tweeter survives a white noise power test it will be indestructible in the real world.
It is much closer to what and how we hear than white noise although obviously it will not perfectly represent all music.
If you listen to pink noise to us it sounds as if the noise is even across all frequencies, on the other hand white noise sounds like pure hiss (all treble) to us.
Mind you if a tweeter survives a white noise power test it will be indestructible in the real world.
Hi,
White noise is a terrible option, it has as much power above 10KHz
as below. Pink noise is better but still wrong, music is not constant
power per octave, your best bet is pink noise with a 1KHz baxadall
tone control set to +6dB bass and -6dB treble to approximate roughly.
Not that the OP's original question makes any real sense anyway,
zaph|audio - bargain aluminum mtm
Is a LR4 acoustic that will totally screw up
the calculations as its second order electrical.
Put that in your pipe and smoke it.
rgds, sreten.
Last edited:
The way I understand it pink noise has equal power for each octave.
So there is as much energy between 100-200Hz as there is from 10k-20k.
White noise has equal power across all frequencies.
So there is as much energy between 100-200Hz as there is from 10k-10.1k.
As far as I know gamelan (traditional balinese music) has the highest treble density.
So there is as much energy between 100-200Hz as there is from 10k-20k.
White noise has equal power across all frequencies.
So there is as much energy between 100-200Hz as there is from 10k-10.1k.
As far as I know gamelan (traditional balinese music) has the highest treble density.
As usual sreten takes even the smallest miss-statement as a condemnation of ones expertise. Yes above 10 kHz there is little energy in any music, which is why I mostly ignore it. But I have measured recordings with a near flat spectrum up to 10 kHz which means that pink noise, with all its HF fall-off is not really the best choice.
Years ago we had a few compression driver failures from VC burnout. I was baffled at first until I realized that when an amp clips (which in a club is most of the time) the spectral content moves ONLY upward in frequency to ever higher and higher frequencies. When we limited the content above 10 kHz the problem went away. While this was pro, it still highlights the issue in that conservative design for thermal is a mistake. I always design for a flat spectrum (to 10 kHz!!!!,sreten!) and then I know that nothing bad will happen. If you want to be "right on the edge" then this is unwise, but I think it equally unwise in DIY to be so conservative.
Years ago we had a few compression driver failures from VC burnout. I was baffled at first until I realized that when an amp clips (which in a club is most of the time) the spectral content moves ONLY upward in frequency to ever higher and higher frequencies. When we limited the content above 10 kHz the problem went away. While this was pro, it still highlights the issue in that conservative design for thermal is a mistake. I always design for a flat spectrum (to 10 kHz!!!!,sreten!) and then I know that nothing bad will happen. If you want to be "right on the edge" then this is unwise, but I think it equally unwise in DIY to be so conservative.
Last edited:
Dear gedlee, you complain about reactions about your "small miss-statements" .... but you continue flavouring your comments with them, impossible not to fall for that bait 
Now it's your :
In fact pink noise just compensates white noise rising HF content to get a much more useful constant power per octave response.
Constant is very different from falling.
To put some numbers into it [tm] here's some samples of average energy content at different frequencies as produced by the instruments recorded, this is NOT a speaker system frequency response curve.
What they clearly show is that even a synthetic signal such as pink noise has much higher HF energy compared to these actual music samples.
Suggesting even brighter unbalanced white noise to test tweeter survivability approaches nonsense.
If any, this music if played loud would have a higher than normal HF content (distortion) and only then approach pink noise, which thus shows itself as a very sensible test source.
I haver a dream that someday World Audio Standards organizations will adopt pink noise as a valid test source.
Oh !!! they already did?
Over 40 years ago?
Clever chaps
Here's some samples of energy content by music type.
Pulled straight from AVI loudspeakers data, posted in:
Spectral Content of Music (Page 1) - Reference - AVI HiFi Forum
NOTE 2: pink noise has much stronger deep bass and above all high frequencies than these average recordings.
Testing with white noise?
Well, that woukd be a flawed test
Now it's your :
In fact pink noise just compensates white noise rising HF content to get a much more useful constant power per octave response.
Constant is very different from falling.
To put some numbers into it [tm] here's some samples of average energy content at different frequencies as produced by the instruments recorded, this is NOT a speaker system frequency response curve.
What they clearly show is that even a synthetic signal such as pink noise has much higher HF energy compared to these actual music samples.
Suggesting even brighter unbalanced white noise to test tweeter survivability approaches nonsense.
If any, this music if played loud would have a higher than normal HF content (distortion) and only then approach pink noise, which thus shows itself as a very sensible test source.
I haver a dream that someday World Audio Standards organizations will adopt pink noise as a valid test source.
Oh !!! they already did?
Over 40 years ago?
Clever chaps
Here's some samples of energy content by music type.
NOTE: they chose a fragment with the organ playing full blast, that's why it shows so much bass.
Pulled straight from AVI loudspeakers data, posted in:
Spectral Content of Music (Page 1) - Reference - AVI HiFi Forum
NOTE 2: pink noise has much stronger deep bass and above all high frequencies than these average recordings.
Testing with white noise?
Well, that woukd be a flawed test
JMFahey's power curves confirm something Lynn Olson said about shallow 6dB/octave tweeter filters. They barely keep up with the rising power in normal music at lower frequencies, so readily distort due to overload. We seem to have 6dB power rise per octave there as you go to lower frequency.
All this argues for as high a crossover as you can get away with and slopes sufficiently steep to get the Fs and cone breakup regions well down.
I have compared BW3 and LR4 on a 3kHz crossover with 8" bass. BW3 naturally pushes the drivers harder than LR4 at crossover. It also has about 10dB less suppression of the troublesome regions. So it is a more distorted sound. On the other hand, BW3 has a lovely overall sound. It has 3dB more power than LR4 at crossover, in fact power delivery is flat. Frequency response is also flat due to the miracle of 90 degree phase.
The downside is it lobes, which is what the D'Appolito MTM design is meant to correct in PA type applications. But because you don't get something for nothing, Lobing is replaced by combing.
Bottom line, IMO, BW3 is essential for MTM, but make sure the drivers have good overlap in the first place. I would expect it to work well with small 5" bass, because the first troublesome bass cone resonance is around 6kHz.
Off-topic perhaps, but I have a notion that MTTM is theoretically better, and that series wiring the drivers has benefits.
All this argues for as high a crossover as you can get away with and slopes sufficiently steep to get the Fs and cone breakup regions well down.
An externally hosted image should be here but it was not working when we last tested it.
I have compared BW3 and LR4 on a 3kHz crossover with 8" bass. BW3 naturally pushes the drivers harder than LR4 at crossover. It also has about 10dB less suppression of the troublesome regions. So it is a more distorted sound. On the other hand, BW3 has a lovely overall sound. It has 3dB more power than LR4 at crossover, in fact power delivery is flat. Frequency response is also flat due to the miracle of 90 degree phase.
The downside is it lobes, which is what the D'Appolito MTM design is meant to correct in PA type applications. But because you don't get something for nothing, Lobing is replaced by combing.
Bottom line, IMO, BW3 is essential for MTM, but make sure the drivers have good overlap in the first place. I would expect it to work well with small 5" bass, because the first troublesome bass cone resonance is around 6kHz.
Off-topic perhaps, but I have a notion that MTTM is theoretically better, and that series wiring the drivers has benefits.
This is not a simple subject, because little research has been done, and what little research there was is quite old. (If anyone is aware of newer investigations please post here, and PM me, I 'd love to see!)
You can reference "Power Transmission Through Crossover Networks"
AES E-Library Power Transmission Through Crossover Networks
I don't have the print version here, but I distinctly recall a few things:
1) I believe the author ends up using pink noise, filtered -6 dB below 40 Hz and above 2 kHz, as a rough approximation. Of course, different pieces of music can be very different, but they all have far less energy in the high frequencies. White noise is way too harsh, though if your tweeter can survive this it should be bulletproof.
2) The transient (peak) characteristics can be very different from the RMS power. But still, less energy at high frequencies.
3) Steeper crossovers do NOT necessarily reduce the power into the driver! You say: HUH?!?! WHAT?!?! Here's why: the crossover slopes are logarithmic, but actual power is not. A steeper crossover slope reduces energy below the crossover frequency…but allows MORE power just above the crossover frequency. Look carefully at plots of crossover response and you can see this. However, it is deceptive because it is logarithmic. The more-filled-in "shoulder" or "knee" contains as much energy as what looks like the so-much-smaller "skirt" of the rolloff, if you plot it out linearly.
4) Now, steeper high pass slopes DO reduce EXCURSION, which I suspect causes more tweeter failures than thermal. Especially, if the high pass is not too far above resonance, and there is not impedance compensation, the resonance peak at impedance can "negate" the crossover, and you get more power and excursion at a low frequency which is hard for the tweeter to handle. Being sure the crossover rolls off properly *may* be more important than the exact frequency or slope.
5) A Linkwitz-Riley crossover will indeed reduce the power into the tweeter…because the crossover frequency is "actually higher." That is, it is -6 dB instead of -3 dB, so of course there is less power. But that is an apples-to-oranges comparison, because the L-R -3 dB is higher.
The basic solution here is
- Use robust tweeters
- Do NOT clip the amplifier!*
- Use L-R or not for acoustic reasons, not due to power transmission.
*Clipping introduces somewhat more power as the waveform squares out, but another AES paper showed this was not actually the primary cause of burnout. The primary problem was that with a 100W amp, 100W of clipped bass can become maybe 200W. But, say, 3W of much softer treble buried in the same music could become 200W of treble and poof! goes the tweeter.
You can reference "Power Transmission Through Crossover Networks"
AES E-Library Power Transmission Through Crossover Networks
I don't have the print version here, but I distinctly recall a few things:
1) I believe the author ends up using pink noise, filtered -6 dB below 40 Hz and above 2 kHz, as a rough approximation. Of course, different pieces of music can be very different, but they all have far less energy in the high frequencies. White noise is way too harsh, though if your tweeter can survive this it should be bulletproof.
2) The transient (peak) characteristics can be very different from the RMS power. But still, less energy at high frequencies.
3) Steeper crossovers do NOT necessarily reduce the power into the driver! You say: HUH?!?! WHAT?!?! Here's why: the crossover slopes are logarithmic, but actual power is not. A steeper crossover slope reduces energy below the crossover frequency…but allows MORE power just above the crossover frequency. Look carefully at plots of crossover response and you can see this. However, it is deceptive because it is logarithmic. The more-filled-in "shoulder" or "knee" contains as much energy as what looks like the so-much-smaller "skirt" of the rolloff, if you plot it out linearly.
4) Now, steeper high pass slopes DO reduce EXCURSION, which I suspect causes more tweeter failures than thermal. Especially, if the high pass is not too far above resonance, and there is not impedance compensation, the resonance peak at impedance can "negate" the crossover, and you get more power and excursion at a low frequency which is hard for the tweeter to handle. Being sure the crossover rolls off properly *may* be more important than the exact frequency or slope.
5) A Linkwitz-Riley crossover will indeed reduce the power into the tweeter…because the crossover frequency is "actually higher." That is, it is -6 dB instead of -3 dB, so of course there is less power. But that is an apples-to-oranges comparison, because the L-R -3 dB is higher.
The basic solution here is
- Use robust tweeters
- Do NOT clip the amplifier!*
- Use L-R or not for acoustic reasons, not due to power transmission.
*Clipping introduces somewhat more power as the waveform squares out, but another AES paper showed this was not actually the primary cause of burnout. The primary problem was that with a 100W amp, 100W of clipped bass can become maybe 200W. But, say, 3W of much softer treble buried in the same music could become 200W of treble and poof! goes the tweeter.
Last edited:
That all makes sense.
I wanted to add a couple of instructive MTM designs. Lynn Olson actually tuned the Ariel with Pink Noise:
The Ariel, Part II
The SB Acoustics Elok design is more of a start than anything. You are encouraged to adjust it:
SB Acoustics :: Elok
With inefficient 5" bass units, the woofers are probably more at risk than the tweeter IMO.
I wanted to add a couple of instructive MTM designs. Lynn Olson actually tuned the Ariel with Pink Noise:
The Ariel, Part II
The SB Acoustics Elok design is more of a start than anything. You are encouraged to adjust it:
SB Acoustics :: Elok
With inefficient 5" bass units, the woofers are probably more at risk than the tweeter IMO.
Hi,
there is nothing small about the miss-statement, its wrong.
The above is nonsense. Limiting content above 10KHz makes
no difference to the consequence of clipping of an amplifier,
your just talking lazy self justifying "expertise" nonsense.
That you think this is all about apperent expertise is your
problem, it isn't. Its about making sensible sense for the OP.
rgds, sreten.
That statement is simply wrong. If I clip a 10 kHz signal then I get 20 kHz, 30 kHz etc. Filter out these higher harmonics going to the tweeter and they won't affect it.
What sort of filter did you find worked to control clipping high harmonics, Earl? I can only think of a small 0.1mH coil and/or a tweeter Zobel.
Modern ferrofluid tweeters seem tough as old boots to me. Can't they handle instantaneous 220W? Some of the SEAS tweeters can take a surprising 90W long-term. H1212-06 27TBFC/G
Well, this is using pink noise 12dB rolled off at 5kHz according to IEC 268-5. So I'm guessing the tweeter only gets about 10W maximum power really, which is the power rating of the resistors I use too.
Attachments
Last edited:
Hi,
Wrong ?
Your taking advanced utterly pedantic garbage in every respect.
Yeah lets get rid of everthing above 10KHz ... your the expert.
In general pragmatic terms what I said is correct, and its quite
pathetic your using such a p*ss poor argument to seem to be right.
You were and are wrong from the first off about white noise, and
it doesn't get more any complicated, other than your wriggling.
rgds, sreten.
Last edited:
Maybe read the cursive, bolded text. While in home environment maybe not so critical for the masses. Otherwise there's a lot of red noise in the thread
Peter
I think is has become necessary to emphasize that my primary objective as stated in the second sentence of my OP, is to determine the difference in the power dissipation in a tweeter when using a BW3 compared to a LR4 Xover. In order to establish this, I asked for info on a suitable signal source that replicates the spectral density of normal music program material.
I am not asking for a signal source to determine the maximum dissipation in a tweeter. Providing links to existing designs and/or Xover schematics misses the point completely.
Peter
PS. My hearing aid clips well before my amps!!
I am not asking for a signal source to determine the maximum dissipation in a tweeter. Providing links to existing designs and/or Xover schematics misses the point completely.
Peter
PS. My hearing aid clips well before my amps!!
If you want to measure the dissipation I suppose white noise would show up any effect more plainly simply because there is more energy to dissipate.
You could always band-limit it to 2 octaves either side of the xover point.
Of course in real life the effect would be much smaller.
I think in essence I'm saying use white noise to see if there is a difference in the first place and if there is repeat with pink to see its magnitude in more realistic circumstances.
You could always band-limit it to 2 octaves either side of the xover point.
Of course in real life the effect would be much smaller.
I think in essence I'm saying use white noise to see if there is a difference in the first place and if there is repeat with pink to see its magnitude in more realistic circumstances.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.