tinitus said:
This is something I'm wondering too. I have read very high average operating temperatures on some ribbons, such as this:
http://www.madisound.com/catalog/PDF/Eminence/neopro5i.pdf
How does such a high typical temperature effect matters? Also the ribbon will presumably heat and cool very quickly with such a small thermal mass? And then the metal will presumably change size marginally due to heating (though corrigations or suspension will likely make this unimportant).
Despite this I suspect that at such high frequencies power compression effects would not be audible as such, I can't imagine it heating and cooling significantly at only frequencies above 3khz or so? Just a general rise in temperature would be experienced?
I also wonder about planars. They are usually "heatsinked" with thier aluminium components but that will probably only serve to reduce long term temperature rise?
panomaniac said:
Let me be clear - the closed box vs. ported box is a secondary factor to having a lot of subs. Its not the type that matters so much as it is the number and thier location. The room is a much bigger factor than the box type.
In my deisgns, the woofer goes all the way up to 1 kHz so cone motion is a no-no. From that regard I want to lower the cone excursion so I use a closed box. I have no stong prference for ported, closed, bandpass, whatever if they are LF limited - only the number and location.
This is also on my list to gradually build dynamics capability. The last time I looked at information, it mentioned 120db (at the listening point) but did not mention the frequency content. So this is still a mystery.graaf said:
For design of frequency response, would go for 10Hz~50KHz as the ultimate goal.
How realistic things sound also depends on recording technique which is just as important, if not more, as the speaker systems in terms of technical maturity.
soongsc said:
Time for a reality check. Consider what it would take to generate a 120 dB level at the listening position. Let's start by saying the listening position is 3 M from the source. So we start by adding 9 dB, or we need 129 dB at 1M. Even with a system that has a nominal sensitivity of 100dB/2.83V at 1 M this would mean you need an amplifier with a voltage swing of 90 volts RMS (1000w/8 ohms, 2000W/4ohms). I'm guessing that such a system would have an impedance of less than 8 ohms so you can see the demands on amplifier power are pretty extreme.
Now tell me you want to drive the system with a SET. 🙂
I actually have done some calculation, and there are some amps around here that claim 1000W or more into 8ohms transient capability (I do have some doubts though). I am quite sure no music is going to play 120db rms in reality, but we do need the transient voltage capability and the driver max SPL capability. If we use active XO, impedance is not a limiting factor.john k... said:
The good side is that there is now time schedule.
😀
Are we talking 120db PEAK? If so, then maybe I can understand.
Crickey, most rock & roll technical riders call for "only" about 112dB at the mix position. But of course that's RMS, not peak. That's loud, let me tell you.
Classical music is typically recorded at -20 or -22 dB down from peak as an average level. So dynamics from the average to the peak are only 22dB. That's still a lot.
Of course from pianissimo to peak will be more.
Pop, jazz and rock have a much smaller average to peak ratio. Typical is 18dB for dynamic stuff, less than 10dB for metal.
So, if your efficient system is using 1/2 watt for average levels, how much power do you need for the +22dB peaks of classical music? I'll let you do the math.
Crickey, most rock & roll technical riders call for "only" about 112dB at the mix position. But of course that's RMS, not peak. That's loud, let me tell you.
Classical music is typically recorded at -20 or -22 dB down from peak as an average level. So dynamics from the average to the peak are only 22dB. That's still a lot.
Of course from pianissimo to peak will be more.
Pop, jazz and rock have a much smaller average to peak ratio. Typical is 18dB for dynamic stuff, less than 10dB for metal.
So, if your efficient system is using 1/2 watt for average levels, how much power do you need for the +22dB peaks of classical music? I'll let you do the math.
We recently used an amp rated at 4400 watts to test our loudspeakers for power compression. But it went into compression itself. When we tested it all we could get out of it in the steady state was about 100 watts.
We asked the manufacturer and got some song and dance about how the amp limits, etc. after 100 millisec. or so - now get this - "because of mains current draw". I loved that one. It wasn't the amp that was the limiting factor it was the local power company!!!
We asked the manufacturer and got some song and dance about how the amp limits, etc. after 100 millisec. or so - now get this - "because of mains current draw". I loved that one. It wasn't the amp that was the limiting factor it was the local power company!!!
gedlee said:
Probably not as unreasonable as you might think. I do have an amp circa 1980 rated at 325/650/1000 into 8/4/2 ohms and we tried to test it at the shop where I bought it. No problem at 8 ohms. No problem at 4 ohms, but at 2 ohms lights in the store got dim just before the breakers went and we were still under 400 watts. A quick calculation showed that we were sucking close to 15 amps from the power supply (assuming 100% efficiency) and the circuit the amp was plugged into couldn't handle it.
I don't know the watts vs. impedance spec of the amp you have but I suspect the manufacture was honest. You can’t pull that kind of continuous power out of the wall for without a special circuit without tripping something. 4400 w come down to 23+ amps into 8, 33 into 4 and 47 amps into 2 ohms. Plugged into a 120 outlet you couldn’t even get the required voltage across an 8 ohm load to give an instantaneous 4400 watts. Assuming a full 120 V RMS across 8 ohms you only reach 1800 watts (15 amps).
soongsc said:
Impedance is still a factor. What's the impedance of the woofer system? Mids? tweeter? You have to consider but the max current capability of the amp and the rail voltage. You need rail voltages in excess of +/- 90 volts DC to reach 1000w (instantaneous) into 8 ohms.
John
Your comments are certainly true, but I think that they are all pretty obviuos. My point was more one of people actually believing what the marketing documents say. It really doesn't matter where the limitation is, does it? There is a limit to the power that can be delivered to a load.
Your comments are certainly true, but I think that they are all pretty obviuos. My point was more one of people actually believing what the marketing documents say. It really doesn't matter where the limitation is, does it? There is a limit to the power that can be delivered to a load.
Hi
The issue of power and time are highly over looked, if one takes a peak hold SLM and walks around the house, one finds that many ordinary things produce very high peak SPL’s. As our ears judge “loudness” based on some time constant and spectrum , it is normal that things with big peaks but low average level are not judged to be “loud”.
For example, tossing a metal spoon onto a tile floor produced a peak nearly 140dB, sitting in a car and slamming the door was about 145 dB.
Obviously that target (being able to produce peaks levels of real life) is a bit out of trajectory for most home size guns.
At work, many applications for our stuff is in large reproducing systems so listening at 10 meters requires 20dB over the 1M level so even with 100dB+ sensitivity, one still uses peaks of thousands of Watts. Add to that the fact that there is a separate subwoofer generally so these are two way bi-amped systems at a minimum.
Here is a thread where Langston Holland measures the peak power output vs time for some for the current crop of monster amps. Scroll about half way down.
http://srforums.prosoundweb.com/index.php/m/0/24996/48/0/#msg_24996
Its funny, part of the thing that made me pursue this microphone thing I’m working on was having that peak capacity at home. When you measure normal commercial recordings one finds they typically take up a shockingly small part of the medium’s total dynamic range.
Anyway, I don’t know if the fireworks link works yet ( a new web site which isn’t done yet) but that has 70dB dynamic range, more than any CD I’ve seen and it can use up all your headroom and still not sound loud. Even the Harley recording has about 60 dB dynamic range.
It is then partly an issue of what is there in real life vs what is in the recording so far as what dynamic capacity is needed to reproduce an event. Headroom is your friend when re-producing a signal.
Best,
Tom Danley
The issue of power and time are highly over looked, if one takes a peak hold SLM and walks around the house, one finds that many ordinary things produce very high peak SPL’s. As our ears judge “loudness” based on some time constant and spectrum , it is normal that things with big peaks but low average level are not judged to be “loud”.
For example, tossing a metal spoon onto a tile floor produced a peak nearly 140dB, sitting in a car and slamming the door was about 145 dB.
Obviously that target (being able to produce peaks levels of real life) is a bit out of trajectory for most home size guns.
At work, many applications for our stuff is in large reproducing systems so listening at 10 meters requires 20dB over the 1M level so even with 100dB+ sensitivity, one still uses peaks of thousands of Watts. Add to that the fact that there is a separate subwoofer generally so these are two way bi-amped systems at a minimum.
Here is a thread where Langston Holland measures the peak power output vs time for some for the current crop of monster amps. Scroll about half way down.
http://srforums.prosoundweb.com/index.php/m/0/24996/48/0/#msg_24996
Its funny, part of the thing that made me pursue this microphone thing I’m working on was having that peak capacity at home. When you measure normal commercial recordings one finds they typically take up a shockingly small part of the medium’s total dynamic range.
Anyway, I don’t know if the fireworks link works yet ( a new web site which isn’t done yet) but that has 70dB dynamic range, more than any CD I’ve seen and it can use up all your headroom and still not sound loud. Even the Harley recording has about 60 dB dynamic range.
It is then partly an issue of what is there in real life vs what is in the recording so far as what dynamic capacity is needed to reproduce an event. Headroom is your friend when re-producing a signal.
Best,
Tom Danley
Tom
Remember that as far as the thermal stuff goes - the base topic of this thread - its the power that is dumped, not how loud it is. The temperature problems occur almost instantaneously with the large signals even if we don't hear them as loud. This is one of the inderlying interests that I have. These thermal issues may occur even at low levels if the signals dynamic range is sufficient.
Remember that as far as the thermal stuff goes - the base topic of this thread - its the power that is dumped, not how loud it is. The temperature problems occur almost instantaneously with the large signals even if we don't hear them as loud. This is one of the inderlying interests that I have. These thermal issues may occur even at low levels if the signals dynamic range is sufficient.
Hi Earl
Well if one were considering the electrically related issues and if this were a MFR wishing to “find out” on a production unit, I think I would do something like the following.
For each driver in the system, monitor/ record the Voltage and Current going to the driver and record the SPL.
Use band limited pink noise suitable for the drivers frequency range or a sine at Rmin.
Slowly increase the level, recording the Voltage, Current and SPL until the Driver fails thermally.
The change in effective Rdc would be tracked and then a final temperature could be estimated and a long term power compression curve could be generated.
The rate of power increase here would be slow enough to be way longer than the primary thermal TC’s to the VC is near asymptotic temp for each level (ignoring the long term stuff like magnet heating and such)..
Then, knowing where the driver fails from I^2*Rdc heating, on a new driver set up the same setup except track the change in Rdc when the speaker goes from being driven at Zero power to say –3dB from failure in a step function.
This rate of change should reveal the thermal TC (s) as the coil heats to its asymptotic temp at –3dB from failure.
With that TC for each driver and some idea (from the failure test) how much power it can dissipate and how much the Rdc has changed at a given power, one ought to be able to make a pretty good model with some data to compare against.
Best,
Tom Danley
Well if one were considering the electrically related issues and if this were a MFR wishing to “find out” on a production unit, I think I would do something like the following.
For each driver in the system, monitor/ record the Voltage and Current going to the driver and record the SPL.
Use band limited pink noise suitable for the drivers frequency range or a sine at Rmin.
Slowly increase the level, recording the Voltage, Current and SPL until the Driver fails thermally.
The change in effective Rdc would be tracked and then a final temperature could be estimated and a long term power compression curve could be generated.
The rate of power increase here would be slow enough to be way longer than the primary thermal TC’s to the VC is near asymptotic temp for each level (ignoring the long term stuff like magnet heating and such)..
Then, knowing where the driver fails from I^2*Rdc heating, on a new driver set up the same setup except track the change in Rdc when the speaker goes from being driven at Zero power to say –3dB from failure in a step function.
This rate of change should reveal the thermal TC (s) as the coil heats to its asymptotic temp at –3dB from failure.
With that TC for each driver and some idea (from the failure test) how much power it can dissipate and how much the Rdc has changed at a given power, one ought to be able to make a pretty good model with some data to compare against.
Best,
Tom Danley
There seems to be a trend for higher orders of harmonic distortion to be proportionally less pleasing to the ear. 2nd-order is fairly benign, while something like 5th-order is likely to be painful at a much lower amplitude than 3rd or 4th.gedlee said:
gedlee said:
Obvious to you and me, and probably many others; perhaps not so obivious to the masses. I read the tone of your post as if you had overlooked the issue as well.
Being lazy I would just read Klippels paper on thermal effects in loudspeaker modeling before speculating or embarking on an experimental program from scratch
http://klippel.de/pubs/Klippel papers/Nonlinear_Modeling_of_Heat_Transfer_03.pdf
http://klippel.de/pubs/Klippel papers/Nonlinear_Modeling_of_Heat_Transfer_03.pdf
jcx said:
What Klippel shows in that paper is not the same thing that I am doing. Thermal modeling, as this paper shows, has been arround a long time - its been in SPEAK for nearly 20 years.
What I am interested in is scaling the audiblity, since if its not audible then its not important - except for preventing speaker meltdown.
The models don't help with this aspect.
I do question Klippels results however since he shows a time constant for the voice coil of about 15 seconds, which is far far longer than what it should be theoretically and what I measured in my test.
Its a good paper and well worth reading (I'll read it again) but doesn't address what I am interested in.
This is the same issue as Klippels nonlinear modeling. It doesn't address audibility. Whats the point of modeling, manipulating or worrying about something that you can't hear.
I have noticed the limitation of recordings as well. Back in the 80's, or 90's, there were expanders and peak unlimiters designed specifically to address compressed recordings and limited recordings. I have had the best experience with peak unlimiters.Tom Danley said:
One thing we probably should also consider is that taken the noise level in real situations which in reality, we really have only 70db above ambient noise.
listening tests with "auralizations" of the modeled distortion are a part of Klippel's approach
paper:
http://klippel.de/pubs/Klippel pape...n-Subjective_Evaluation_of_nonlin_Dist_01.pdf
online test:
http://www.klippel-listeningtest.de/lt/background.html
result summary - looks like ~ 20 dB less distortion would satisfy many people with this particular speaker/music
http://www.klippel-listeningtest.de...el=-1&speaker=6 inch driver&music=Music - Pop
he may not have included voice coil thermal modeling in these test models - and certainly further experimental tests of models and meaurements would very good to see
I'm not presuming to instruct Geddes on this subject - I just thought Klippel has so much info online that is directly relevant that it should be pointed to for the rest of us to get up to speed
paper:
http://klippel.de/pubs/Klippel pape...n-Subjective_Evaluation_of_nonlin_Dist_01.pdf
online test:
http://www.klippel-listeningtest.de/lt/background.html
result summary - looks like ~ 20 dB less distortion would satisfy many people with this particular speaker/music
http://www.klippel-listeningtest.de...el=-1&speaker=6 inch driver&music=Music - Pop
he may not have included voice coil thermal modeling in these test models - and certainly further experimental tests of models and meaurements would very good to see
I'm not presuming to instruct Geddes on this subject - I just thought Klippel has so much info online that is directly relevant that it should be pointed to for the rest of us to get up to speed
- Status
- Not open for further replies.