Peak power same thing.
Edit: Dadod, saw your post after I posted mine. Would have made it superfluous.
Last edited:
Yes I did get good results from thye tests, as well as realising how fallible our perception is and how easily the results can be biased by many things other than hearing.
I also learn a lot doing and being part of extensive listening tests for clarity for the next generation of USA vehicle communication systems, these were mainly modified rhyme tests done under controlled conditions, all participants had hearing tests before hand...
So sorry I am a believer in DBT as part of the design cycle and testing.
The striving was when I had carried out mods or tried cables that created differences that according to many where "night and day" differences, that I found under blind conditions just were not there... I realised that so many things work to alter our perceptions and quite often you are fooling yourself when you think there is a change, when there isn't....
A perfect 100W square wave at 20Hz has 81W at 20Hz, 9W at 60Hz, 3.2W at 100Hz etc. By the time you get to 1kHz there isn't much power left. Let's do a rough estimate assuming the worst case of a 6dB/octave crossover. At 1kHz the component will be 32mW, and half will go each way. We can approximate the other LF components by adding 32mW for each of the 50 of them. So we have about 1.6W of LF energy going to the tweeter.
The easiest way to estimate the HF energy is to add up the bigger LF components and see what is left. We already have the first few above. To them we can add 1.65W at 140Hz, 1W at 180Hz, 670mW at 220Hz, 480mW at 260Hz, 360mW at 300Hz. I'm getting bored so I will stop there and add up: 97.36W. So at most there is 2.64W of HF energy (above 300Hz - let alone above 1kHz). So we have less than 4.2W going to the tweeter.
However, all the Fourier components are lined up (well, they were before the crossover phase shifted them) so if displacement is the limiting factor (rather than heat) then you might need a somewhat more powerful tweeter than this. Nevertheless, I think this shows that considering a low frequency square wave is a rather bad way to determine tweeter power requirements.
The easiest way to estimate the HF energy is to add up the bigger LF components and see what is left. We already have the first few above. To them we can add 1.65W at 140Hz, 1W at 180Hz, 670mW at 220Hz, 480mW at 260Hz, 360mW at 300Hz. I'm getting bored so I will stop there and add up: 97.36W. So at most there is 2.64W of HF energy (above 300Hz - let alone above 1kHz). So we have less than 4.2W going to the tweeter.
However, all the Fourier components are lined up (well, they were before the crossover phase shifted them) so if displacement is the limiting factor (rather than heat) then you might need a somewhat more powerful tweeter than this. Nevertheless, I think this shows that considering a low frequency square wave is a rather bad way to determine tweeter power requirements.
The proposed "drum hit" does not look like a square wave, it could be possible for it to have a very high amplitude initial attack and a lower fundamental (whose peak is of necessity delayed in time)...
If one makes the assumption that for a given multi-way multi-amp system the drivers all have the same sensitivity, AND that one wants to assure that there is no clipping of peak signals, then I suppose the design would call for amplifiers of equal power. (or clipping at the same SPL levels, I suppose is more accurate).
Otoh, if one used an 85dB sensitive bass section and a 108dB sensitive HF horn, then equal amplifier powers clearly are not required. So, the "aim/rule" is not about power, but acoustic output WRT input power?
_-_-
If one makes the assumption that for a given multi-way multi-amp system the drivers all have the same sensitivity, AND that one wants to assure that there is no clipping of peak signals, then I suppose the design would call for amplifiers of equal power. (or clipping at the same SPL levels, I suppose is more accurate).
Otoh, if one used an 85dB sensitive bass section and a 108dB sensitive HF horn, then equal amplifier powers clearly are not required. So, the "aim/rule" is not about power, but acoustic output WRT input power?
_-_-
Swamps & the "imaginary driver"
Going back to my imaginary HF driver for a moment.
There was a discussion as to the audibility of an amplifier with distortion under 0.1% earlier. (or was it 1.0%??)
The driver I suggested had a spec alleged to be 128db/1m (max power) and 0.1% measured THD.
The other thing that was put forward (back a few pages now) was that amplifier distortion was going to be swamped by speaker distortion.
Now, I asked what the likely result of dropping level in any (reasonably well designed) driver is going to have on THD? Will it remain constant? Eg. at 128dB it's 0.1% and at 100dB it's still 0.1%?? Or will the THD reduce in % with a reduction of level, and what would a "typical characteristic curve" for that THD vs Power input look like?
I don't actually know the answer to the last question. Do not think I have seen it presented that way, and I don't have any data to put points on a graph to approximate it either. Anyone have this??
Well, here's the way it looks to me...
At +128dB, 0.1%THD the distortion sits at a level of +68dB
At +100dB, 0.1% THD the distortion sits at a level of +48dB (assuming no change in THD)
At +100dB, 0.01% THD the distortion sits at +20dB (or -80dB down) (assuming a drop in THD with power in...)
What IF the reduction in distortion at +95dB (a fairly normal average listening level) was greater still and was 0.005%? Then we'd be looking at numbers approaching that of pretty well designed low distortion amplifiers, and close to redbook CD numbers, lower than numbers from LP and probably magnetic tape...
So, the reason I bring this up is twofold.
First, such a driver does exist - actually there are a number of them out there, afaik.
(I own a pair, fwiw) They do sound substantially different and cleaner than *any* dome or cone when compared - blind or not)
Secondly, it seems obvious to me that an amplifier with 1.0% or 0.1% THD is going dominate the distortion when heard through such a driver! Not the other way around.
Also, the HF is what dominates the perception of distortion, since LF is Low Passed, and is all sinusoidal type signals. SO, if you are going to percieve "grit" or "hardness" it's all from the HF. One may perceive a difference in bass and midbass, etc... based on the ratios of the harmonic spectra in those drivers, certainly. But the LF character is the lesser half of this issue, imho.
Bottom line, I don't personally buy into this conjecture that amplifiers are not going to be heard because the distortion is "swamped" by speaker's distortion. (I mean it can and does happen that way - but that's not state-of-the-art, now is it?)
_-_-bear
PS. don't hang me if I got the specs for the driver wrong, it's been 25 years since I looked at it... heh. but the idea remains something to consider
Going back to my imaginary HF driver for a moment.
There was a discussion as to the audibility of an amplifier with distortion under 0.1% earlier. (or was it 1.0%??)
The driver I suggested had a spec alleged to be 128db/1m (max power) and 0.1% measured THD.
The other thing that was put forward (back a few pages now) was that amplifier distortion was going to be swamped by speaker distortion.
Now, I asked what the likely result of dropping level in any (reasonably well designed) driver is going to have on THD? Will it remain constant? Eg. at 128dB it's 0.1% and at 100dB it's still 0.1%?? Or will the THD reduce in % with a reduction of level, and what would a "typical characteristic curve" for that THD vs Power input look like?
I don't actually know the answer to the last question. Do not think I have seen it presented that way, and I don't have any data to put points on a graph to approximate it either. Anyone have this??
Well, here's the way it looks to me...
At +128dB, 0.1%THD the distortion sits at a level of +68dB
At +100dB, 0.1% THD the distortion sits at a level of +48dB (assuming no change in THD)
At +100dB, 0.01% THD the distortion sits at +20dB (or -80dB down) (assuming a drop in THD with power in...)
What IF the reduction in distortion at +95dB (a fairly normal average listening level) was greater still and was 0.005%? Then we'd be looking at numbers approaching that of pretty well designed low distortion amplifiers, and close to redbook CD numbers, lower than numbers from LP and probably magnetic tape...
So, the reason I bring this up is twofold.
First, such a driver does exist - actually there are a number of them out there, afaik.
(I own a pair, fwiw) They do sound substantially different and cleaner than *any* dome or cone when compared - blind or not)
Secondly, it seems obvious to me that an amplifier with 1.0% or 0.1% THD is going dominate the distortion when heard through such a driver! Not the other way around.
Also, the HF is what dominates the perception of distortion, since LF is Low Passed, and is all sinusoidal type signals. SO, if you are going to percieve "grit" or "hardness" it's all from the HF. One may perceive a difference in bass and midbass, etc... based on the ratios of the harmonic spectra in those drivers, certainly. But the LF character is the lesser half of this issue, imho.
Bottom line, I don't personally buy into this conjecture that amplifiers are not going to be heard because the distortion is "swamped" by speaker's distortion. (I mean it can and does happen that way - but that's not state-of-the-art, now is it?)
_-_-bear
PS. don't hang me if I got the specs for the driver wrong, it's been 25 years since I looked at it... heh. but the idea remains something to consider
Last edited:
Odd you make no mention of ON Semi/Motorola's fourth generation of originally Toshiba based power devices, MJL 4281/4302. Still more linear, Ft extended from 30 to 35 MHz, power up from 200 to 230W, operating voltage raised to 350V, still better SOAR.
On your comments and mention of the Otala/Lohstroh "low TIM" power amp, I still maintain that was one of the best sounding audio amps I have ever heard. It's a pity so many people dismiss it out of hand mostly due to its lowish power output, nominally 25/50W into 8/4 Ohms, but still 100 V/us slew rate and just 20 dB of global NFB. I was indeed a fool to sell off my unit in a fit of madness.
Marce et al.
I have my system set up such that I can run (admittedly sighted) perfectly instant and very quick AB comparisons on the fly.....this is how to learn and discern VERY fine differences, and reliably.
More to the point, once I have instigated some kind of change to my lounge room main system, I then typically 'let the system run' playing ALL kinds of music, and then listen intently, or casually from a bunch of different positions apart from the sofa sweetspot 'ideal' listening spot....ie the sunroom/workroom, the laundry, the backyard clothes line, the front yard letterbox, the bathroom etc, and if I am giving it some herbs I might go for a long stroll up and down the street and do a roadie 'sound check'....and that might be with every window/door sealed, or the front door opened not more than a few inches.
We all know that 'the proof is in the eating', and going for that sound check walk up and down the opposite side footpath, or all around the house really is informative/the final arbiter....if the sound ain't right it's going to sound 'not right' from everywhere....the converse thankfully is also true.
I recently posted pics of the Van Alstine ABX box, and very curiously/tellingly there was zero discussion brought forward.
My take is, if this comparator box is regarded as the 'standard' for ABX testing, all such testing is quite invalid except for substantially 'gross' differences.
Inserting any such comparator box into any system will most definitely change the system wrt the system stand alone condition, and from what I see of the Van Alstine box the changes are significant, and sufficient to cause masking that significantly reduces the sensitivity/accuracy of the purported testing.
Most notable is the lack of ANY measured performance data, more especially the line level gain adjustment stage.
I have never seen detailed ABX system disclosure/performance measurement data provided in conjunction with 'ABX proofs'...please correct me.
Dan.
Last edited:
That s a FFT and per defintion you cant use this to prove your (lack of) point unless you can psuchologically visualize the sum of the amplitudes, wich i doubt.
What is the level of the signal during the first millisecond once you have added all thoses hamonics amplitude values..?..
In the rising hedge of a square you can see that full amplitude is reached from the start and then sustained, it s not difficult to imagine that the max value is sustained thanks to harmonics, the fundamental max contribution to the level occur at pi/2 and 2pi/3, that is, when the fundamental (wich is a sine) reach its maximum values wich are equal to the max value of the sustained part of the signal.
You are talking of average amplitude and power.
The tweeter will see the full amplitude but during short durations such that average power (and hence level) are very low.
As to being a bad requirement i do not agree, certainly that a perfect square wave is not that adequate but a 50Hz fundamental with harmonics up to say 10KHz is relevant, that s not different from an orchestral hit when all instruments mark a given note with a fortissimo.
It isn't, and your conclusion doesn't follow anyway. But don't let that get in the way of a good indignant hand-wave based on no actual data or analysis.
Wow, we've hashed out time and again the power/energy spectra of most music. Peak power across the audio spectrum is relatively flat (err... dependant on the instruments at play), whereas the amount of energy does fall with increasing frequency. I realize I'm repeating other people's posts.
Conclusions to draw from that WRT (home, I know HF loss in PA is a very different beast) active amplifier setups:
1.) We need essentially the same peak output (dB) across the audio spectrum.
2.) By and large, tweeters are more sensitive than mid/bass drivers.
3.) Scale power/gain (keep that SNR up!) needs appropriately to serve that goal. You can also get away with a lot less beefy PSU/SOAR on your higher frequency amps.
4.) It's probably easier/cheaper to build for your sub amps (or go with a pro class-D amp because the $/watt is stupid good) and repeat the design for the higher frequency amps.
Conclusions to draw from that WRT (home, I know HF loss in PA is a very different beast) active amplifier setups:
1.) We need essentially the same peak output (dB) across the audio spectrum.
2.) By and large, tweeters are more sensitive than mid/bass drivers.
3.) Scale power/gain (keep that SNR up!) needs appropriately to serve that goal. You can also get away with a lot less beefy PSU/SOAR on your higher frequency amps.
4.) It's probably easier/cheaper to build for your sub amps (or go with a pro class-D amp because the $/watt is stupid good) and repeat the design for the higher frequency amps.
I looked, as per stated in my previous post.
The onus is on you to provide full disclosure of all ABX test conditions if you are to propose any such tests as to be valid.......scientific procedure.
So far ALL of your ABX assertions have been without provision of any such test condition data.
Dan.
That s well summarized.
The pic below is a 50Hz/50V pp square wave in red and the corresponding signals with a 2 way 6db/octave filter (one inductance and a cap..) and 8R loads, we can see the action of the filter in the time domain.
Given that amplitude of the harmonics decrease with frequency the picture would be no different if i had restricted the square wave to harmonics up to 10KHz.
Attachments
Last edited:
I have never used nor mentioned the Van Alstine box. You must be thinking of someone else.
I think you don't. There's no connection with Civil Engineering there. The point was in the next paragraph. Saying that the relative level of the confidence with anyone's ears (or other skills) would depend on his experience and expertise (fact#1)...
One's ears can mislead. One's assumption too. Some people are better at making assumptions and some people are better at using their ears.
An example of your wrong assumption (and frequent generalization). There is simply no way you can judge if you are more conservative than I am or not regarding our own ears. You have to understand the "4 quadrant" I was talking about (fact#2)...
If you ever (or frequently) made a mistake regarding your ears, it doesn't tell that your ears are bad. It only tells that you are OVER-CONFIDENT with your ears.
Correct. And we are in agreement. So where is the difference? If you think there is difference, I think the difference is in your head (your assumptions and generalization).
- Status
- Not open for further replies.