These plots don't show energy but level without regard to time so the harpsichord plot only prooves that it has more HF content than grand piano, but it says nothing about a tweeters ability to handle the material at hand.
To make any judgement we need to know the absolute SPL at typical listening distance. I know that I have never put my head into an harpsichord but listen at 2-10 meter. What SPL are we talking about?
I would like to point out also that there are speakers that have other crossover points than 1kHz..
Also one needs to understand the dynamic nature of typical acoustic instruments and music.
/Peter
To make any judgement we need to know the absolute SPL at typical listening distance. I know that I have never put my head into an harpsichord but listen at 2-10 meter. What SPL are we talking about?
I would like to point out also that there are speakers that have other crossover points than 1kHz..
Also one needs to understand the dynamic nature of typical acoustic instruments and music.
/Peter
Peter, the plot supplied was nearly identical to yours. Your's proved that tweeters didn't take a lot of energy, why doesn't Ron's prove to the contrary? Additionally, why does time need to play a role in the term energy? It sure seems like you are trying to be argumentative for the sake of it.
I'm sorry if you took offense to that. I just don't understand your argument. You told Dr. Geddes that your graph and example shows that for many types of program material thermal compression was a non-issue. Your graph doesn't represent energy expended over time, nor does Ron's, but Ron's you said didn't show it as an issue.
Obviously the conversion of electrical energy into mechanical energy happens over a period of time, and that the length of that time impacts thermal compression. What I didn't understand was why your graph showed it as a non-issue, but Ron's graph doesn't show it as an issue. Are you saying that a piano is more representative of common program material? Are these acoustic recordings even representative of normal program material? What about rock music, studio jazz recordings, movie soundtracks, etc. You make a good point though that even if reference levels indicate peaks of 105db's (as in THX movie reference levels), that this may not mean the tweeter needs to sustain this level, since the program material may not contain an equal amount of energy up that high, and most THX speakers not made by JBL have pretty high crossover points, well above 1khz. I don't know if that's true, I haven't seen a waterfall plot of a movie soundtrack before to even know how much energy exists above 1khz.
Modeling a tweeter like a tiny woofer, using measured parameters, and calculating the missing ones, I find that it's the lower treble point, around the crossover, which most drives the mechanical limits issue, and when looking at the power graphs, the thermal limits as well. With a normal tweeter, crossover points between 2khz and 3khz seem to limit maximum output to around 100-110db's, regardless of thermal issues. Efficiency plays a roll in the amount of power it takes, my example had a roughly 89db efficiency, and thus power levels to reach that higher number were in excess of 100 watts, some times many times that number if the crossover point was low enough. You simply can not run a normal dome tweeter down to 1khz at all, it reached xmax with just a handful of watts, but isn't all that common either. I didn't model one that low, but I did model it down to 1.8khz, which seems a common crossover point in "audiophile" designs, but less so in home theater designs.
Obviously the conversion of electrical energy into mechanical energy happens over a period of time, and that the length of that time impacts thermal compression. What I didn't understand was why your graph showed it as a non-issue, but Ron's graph doesn't show it as an issue. Are you saying that a piano is more representative of common program material? Are these acoustic recordings even representative of normal program material? What about rock music, studio jazz recordings, movie soundtracks, etc. You make a good point though that even if reference levels indicate peaks of 105db's (as in THX movie reference levels), that this may not mean the tweeter needs to sustain this level, since the program material may not contain an equal amount of energy up that high, and most THX speakers not made by JBL have pretty high crossover points, well above 1khz. I don't know if that's true, I haven't seen a waterfall plot of a movie soundtrack before to even know how much energy exists above 1khz.
Modeling a tweeter like a tiny woofer, using measured parameters, and calculating the missing ones, I find that it's the lower treble point, around the crossover, which most drives the mechanical limits issue, and when looking at the power graphs, the thermal limits as well. With a normal tweeter, crossover points between 2khz and 3khz seem to limit maximum output to around 100-110db's, regardless of thermal issues. Efficiency plays a roll in the amount of power it takes, my example had a roughly 89db efficiency, and thus power levels to reach that higher number were in excess of 100 watts, some times many times that number if the crossover point was low enough. You simply can not run a normal dome tweeter down to 1khz at all, it reached xmax with just a handful of watts, but isn't all that common either. I didn't model one that low, but I did model it down to 1.8khz, which seems a common crossover point in "audiophile" designs, but less so in home theater designs.
pjpoes said:
The plot themself does not proove much, you need to understand the dynamics of the instrument and the typical SPL. Just because there is much HF info like in the harpsichord plot does not mean it means "much energy" into the tweeter.
Becasue if you have a fixed power level it means more energy the longer the duration. P=W/t.
The FFT plots does not show energy, they show relative level.
If you analyze a signal in the FFT plot it is not climbing higher and higher the longer the duration. For all we know the high frequency content could be some fast transients during a couple of seconds in the middle of the piece and the rest is ongoing low frequency stuff for minutes.
The plot shows that this and that frequency has occured during the sample at the indicated max level but not how long.. and it's the time that is interesting when feeding power into the VC (well both are interesting but I hope you get my point).
That's not what I said or meant. I objected against Earl's way of turning the plot into proof for making a case. He have measured a budget dome tweeter and decided that dome tweeters are inferior period!
That does not make sense IMO.
What does it matter if the harpsichord (or whatever instrument or composition) have mostly HF content if it is at a low level? One must take the dynamics of the instrument/music into consideration
and also actual SPL.
Absolutely not, but I want to show that for many situations thermal compression is not a problem since HF content in much music is low. We simply do not tolerate high levels of HF.. it is hard on the ears and music and instruments have evolved according to that.
http://www.stereophile.com/reference/1106hot/index1.html
Of course levels and spectra differs between styles but again we normally do not appreciate high levels of HF with exceptions of clean transients. And of course large rooms needs more from the speakers and many listeners listen at levels that will give them an early onset of hearing problems.
Also many people overestimate the leves they actually listen at.
Exactly!
Waterall graph is used to look at the decay of a system as a function of frequency. Not the same as a spectral plot or a spectrogram (which maybe was what you were thinking of?).
I don't know your definition of normal but many tweeters are good up to 110-120dB with good performance used from 2-3k.
A 90dB sensitive speaker needs a 1000W amp to reach 120dB SPL. I seldom need more than 110dB peak though so 90dB speaker with 100W amp is good for me and a high quality dome tweeter is not a limit.
/Peter
CSD waterfalls show decay, but I was referring to the spectral waterfall plots. I forget the program commonly used over on AVS Forum for this, it's often used to show the low frequency content of movies (You know how movie guys are about bass). Anyway, it shows the change in amplitude as changes in color, like a radar graph, with time shown as x axis, and range across the y-axis. It seems like this would be a better tool to look at the high frequency content over a period of time.
We have been over this discussion time and time again.
My measurements used an "average" dome tweeter, not a poor one, and an "average" compression driver, not a high power one. It's ludicrous to imply that any dome can compete with a compression driver for power and thermal handling. If the dome is "good enough for you", then fine. But many people are finding that the dynamics available from a good compression driver are indeed quite audible and attractive. If you like domes, then stick with them. I've never heard one that I liked.
My measurements used an "average" dome tweeter, not a poor one, and an "average" compression driver, not a high power one. It's ludicrous to imply that any dome can compete with a compression driver for power and thermal handling. If the dome is "good enough for you", then fine. But many people are finding that the dynamics available from a good compression driver are indeed quite audible and attractive. If you like domes, then stick with them. I've never heard one that I liked.
Pan said:
Actually the FFT plot doesn't show the peak level, but rather the average relative levels.
We can square the magnitude of the frequency response and integrate over the tweeter's bandwidth to get the total power going into it (if we set the signal's 0dB to a known voltage). And this is just average power during the whole piece (long term power handling), a high level transient could cause problems with short term power handling.
This is mostly correct, but just squaring the voltage assumes that you have the RMS values. The spectrum is not the RMS values. To be exact, you need to find the Power Spectral Density (PSD) over the duration of the piece and then integrate over frequency to find the average power disipated into the tweeter over this period of time. If you want to know a more instantaneous level, at a peak for example, then you would just find the PSD over that time slice and again integrate over frequency to find the power in this period.
Dr. Geddes:
Is the lack of thermal compression the major reason for the sonic superiority of compression drivers over domes?
FWIW, I am using the DE250 compression driver in a Geddes 15-inch waveguide and the quality is a night and day difference from not only standard domes, but also any ribbon or quasi-ribbon (Magnepan) that I have ever heard.
Once you hear a good compression driver in a good waveguide, there is no going back to obsolete domes.
Is the lack of thermal compression the major reason for the sonic superiority of compression drivers over domes?
FWIW, I am using the DE250 compression driver in a Geddes 15-inch waveguide and the quality is a night and day difference from not only standard domes, but also any ribbon or quasi-ribbon (Magnepan) that I have ever heard.
Once you hear a good compression driver in a good waveguide, there is no going back to obsolete domes.
gedlee said:
If I remember correct it actually had poor performance and I assumed it was poor cheaply built one. Please fresh up my memory what kind of tweeter was it? Soft or metal dome? Ferro or not?
Or maybe it's ludicrous that you write that considering Michael prooved it to be the other way around in one test using the Seas tweeter and a compression driver.
Yes, and as you see in Davids post, non compressed low distortion 120dB peaks at 1kW input is possible from a dome tweeter. I only have 78W amps at the moment, I have only clipped them at one or two occasions and that was midrange clipping. That power means 107dB tops from my tweeters and there is no compression to speak of and the distortion is low.
How do we know that? It may as well be reduced amp clipping since the typical compression driver has higher voltage sensitivity. Remember, a standard speaker with 88dB sensitivty only reach 108 dB SPL peaks with 100W.. 111dB with 200W and not many people have more power than that. In such situations the amp is the limit if the dome tweeter is a good one.
Again, good dome tweeters manage up to 120dB SPL peaks with little or no compression.
/Peter
breez said:
With "advanced" FFT software you can choose peak level, or average and of course the windowing technique choosen affects the results.
I've done some analyzes of dual sines with different duration and I don't really understand the results. I'll get back when I've seen the light!
/Peter
Music is both, so you can't really say that it's one OR the other. You need to have an idea of how a transducer will response to either. The 4 cycle pulse is good for transients, which is basically a test of the mechanical systems linearity, but it will not test the long term (> 1 ms.) thermal capability.
A closed miked cymbal crash would have a very loud sharp attack, followed by a longer sustain and decay with different harmonic content.
Those who attempt to synthesize acoustic instruments discover how difficult it is. Because of the relationship between the fundamental and all the other higher frequency components that change and evolve over time.
A 10khz sine wave has a ten-thousandth (.0001 ) of a sec period, but some instruments sustain at other frequencies for seconds.
To my ears things like cymbal crashes and breaking glass are hard for tweeters to replicate and can be very revealing of HF quality.
Those who attempt to synthesize acoustic instruments discover how difficult it is. Because of the relationship between the fundamental and all the other higher frequency components that change and evolve over time.
A 10khz sine wave has a ten-thousandth (.0001 ) of a sec period, but some instruments sustain at other frequencies for seconds.
To my ears things like cymbal crashes and breaking glass are hard for tweeters to replicate and can be very revealing of HF quality.
AndrewT said:
Both..I think.
There's a level where the tweeter can go on "forever".. say 0.1W to 1W or something like that and also if the transients are short enough 120dB peaks are managable as we can see in David's post above in other words 1kW or so.The hard part is the emiddle ground and you need to know the system and program material well to get some kind of grip on that but the best is of course to measure during real playback levels with program material of different types. It's not very easy but one can arrange with set ups that manage that. One example being the sterophile article I linked to earlier in this thread.
Yes, typically.
The very nature of acoustic instruments is such that for most part of the envelope there is a fundamental that often has the highest amplitude but some instruments can have some of the first harmonics at the highest level. The HF content is typically higher at the start up of plucked and percussive instruments.
The peak power to the tweeter may be similar to a midrange (with lower demand to the bass typically) but the average power is way lower to the tweeter than to the sub 2k or so range.
With synths and overdrive on your strat (not to mention EQ, compressors and limiters) you can obviously have any signal you want but few people enjoy music with a flat spectrum.. especially at high levels. It sounds harsh and hard on the ears.
/Peter
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.