AVS › AVS Forum › Audio › Speakers › Effect of Compression on Peaks
New Posts  All Forums:Forum Nav:

Effect of Compression on Peaks

post #1 of 48
2/8/12 at 7:35am
Thread Starter 
I'm trying to reconcile how we think in terms of peak output for program peaks, versus known levels of power compression in speakers with continuous signals (eg pink noise for 5 minutes).

Planning for peaks in system
When choosing speakers and amps for a room, people often neglect to consider the maximum level the system will need to handle. This is a combination of what volume level they're aiming for (average dialog levels with 20dB peaks for home theater), the size of the room, listening distance, speaker placement, room treatment, and amplification.

Since almost all systems can handle the average dialog levels, it's handling those peaks that we should be concerned with. Many speakers don't have enough sensitivity, thus requiring enormous amps to handle the peaks, or their power handling doesn't allow them to take the peak signal without compression and distortion. It's the latter that I'm concerned about here.

At what level do speakers compress?
I've heard it a couple times on AVS that speaker power compression occurs at 1/8th to 1/10th power. I wasn't sure whether that was continuous or peak, and with what level signal.

Recently, I've re-read the Audioholics article Loudspeakers & Power Ratings: What's the Deal Part II.

In it, Gene DellaSala and Paul Appolino reference three JBL Pro speakers where the power compression is given. It's in terms of fractions of continuous power rating. For instance, the JBL woofer 2226. It's rated 600w AES cont. pink noise. They give three amounts of power compression:
60w 0.7dB (1/10)
300w 2.5dB (1/2)
600w 4.0 dB (rated power)

This means with this 97dB sensitive speaker, instead of getting 113.3dB at 12 feet with 600 watts, you're getting 109dB (I realize this is more than enough SPL, but it's just an example, and it's applicable to 90dB speakers too).

This is after the AES pink noise signal 50-500Hz for 5 minutes.

I made a Google spreadsheet showing the power compression of three speakers referenced in the article, as well as another JBL one I'd researched earlier that only shows compression at 1/10 and 1/2 power.

This would seem to confirm, at least for this speaker, compression starting at 1/10th rated power continuous. There's also distortion of varying amounts at different frequencies: at 40Hz, 3rd harmonic distortion is at -20dB (1%), at only 60w of power. They don't give us the distortion at full power.

What does this mean for peaks?
On the one hand, one could apply this 1/10th rule also to peak power. For most speakers, this would severely limit the level it can reach, resulting in diminished peaks. That peak that was supposed to be 15dB over the average level only reaches 12dB. And the odd-order distortion makes us wince.

On the other hand, a 1 second peak of an orchestral stab or a film's gunshot is not the same thing as a 5 minute continuous signal. We know a speaker can much better handle a brief spike, where the voice coil wasn't heated up much previously, and has time to cool down in the subsequent seconds.

What I'm Asking
Is anyone familiar with what sort of compression and distortion result at what typical fractions of peak power handling, with peak signals for brief periods, like the THX burst signal they use when testing / specifying speakers?

Is it less compression for a 1 second signal with a 20dB crest factor? Or more compression, because we're talking about 4x the watts as the continuous signal?
Reply
Reply

AVS Top Picks

post #2 of 48
2/8/12 at 9:58am
Quote:
Originally Posted by Eyleron View Post

I'm trying to reconcile how we think in terms of peak output for program peaks, versus known levels of power compression in speakers with continuous signals (eg pink noise for 5 minutes).

Planning for peaks in system
When choosing speakers and amps for a room, people often neglect to consider the maximum level the system will need to handle. This is a combination of what volume level they're aiming for (average dialog levels with 20dB peaks for home theater), the size of the room, listening distance, speaker placement, room treatment, and amplification.

Since almost all systems can handle the average dialog levels, it's handling those peaks that we should be concerned with. Many speakers don't have enough sensitivity, thus requiring enormous amps to handle the peaks, or their power handling doesn't allow them to take the peak signal without compression and distortion. It's the latter that I'm concerned about here.

At what level do speakers compress?
I've heard it a couple times on AVS that speaker power compression occurs at 1/8th to 1/10th power. I wasn't sure whether that was continuous or peak, and with what level signal.

Recently, I've re-read the Audioholics article Loudspeakers & Power Ratings: What's the Deal Part II.

In it, Gene DellaSala and Paul Appolino reference three JBL Pro speakers where the power compression is given. It's in terms of fractions of continuous power rating. For instance, the JBL woofer 2226. It's rated 600w AES cont. pink noise. They give three amounts of power compression:
60w 0.7dB (1/10)
300w 2.5dB (1/2)
600w 4.0 dB (rated power)

This means with this 97dB sensitive speaker, instead of getting 113.3dB at 12 feet with 600 watts, you're getting 109dB (I realize this is more than enough SPL, but it's just an example, and it's applicable to 90dB speakers too).

This is after the AES pink noise signal 50-500Hz for 5 minutes.

I made a Google spreadsheet showing the power compression of three speakers referenced in the article, as well as another JBL one I'd researched earlier that only shows compression at 1/10 and 1/2 power.

This would seem to confirm, at least for this speaker, compression starting at 1/10th rated power continuous. There's also distortion of varying amounts at different frequencies: at 40Hz, 3rd harmonic distortion is at -20dB (1%), at only 60w of power. They don't give us the distortion at full power.

What does this mean for peaks?
On the one hand, one could apply this 1/10th rule also to peak power. For most speakers, this would severely limit the level it can reach, resulting in diminished peaks. That peak that was supposed to be 15dB over the average level only reaches 12dB. And the odd-order distortion makes us wince.

On the other hand, a 1 second peak of an orchestral stab or a film's gunshot is not the same thing as a 5 minute continuous signal. We know a speaker can much better handle a brief spike, where the voice coil wasn't heated up much previously, and has time to cool down in the subsequent seconds.

What I'm Asking
Is anyone familiar with what sort of compression and distortion result at what typical fractions of peak power handling, with peak signals for brief periods, like the THX burst signal they use when testing / specifying speakers?

Is it less compression for a 1 second signal with a 20dB crest factor? Or more compression, because we're talking about 4x the watts as the continuous signal?

This is why I use speakers that can play much louder than reference and chose speakers that sound good doing so!
Reply
Reply
post #3 of 48
2/8/12 at 1:14pm
Power compression has little effect on peaks. Power compression results from long term heating of the voice coil, the result of which is that the resistance of the voice coil increases. Increased resistance means less power flow and therefore reduced sensitivity, so output is non-linear with respect to increased power applied.

What this really means to the user is that if you double output power you won't necessarily get 3dB additional output, and that if you push the driver hard enough long enough more power won't give any more output. It will create more heat, and that can lead to driver failure.

This points out the fallacy of manufacturer 'maximum SPL' figures, because those are calculated as linear with respect to 1m/1w sensitivity, and assume that power compression does not exist. In reality you can expect the real maximum SPL to be 3 to 6dB lower than claimed.
Reply
Reply
post #4 of 48
2/8/12 at 1:18pm


This does not suffer from power compression at reference listening at my LP which loses 12 dB's at the seats from the speakers. Basically I need 117 dB's of peaks to reach reference. I will assume these suffer less compression from most because they can scare the crap out of you more than any speaker I have tried. It could be the extra detail at loud levels as well.
Reply
Reply
post #5 of 48
2/8/12 at 1:54pm
Thread Starter 
Quote:
Originally Posted by Bill Fitzmaurice View Post

Power compression has little effect on peaks. Power compression results from long term heating of the voice coil, the result of which is that the resistance of the voice coil increases. Increased resistance means less power flow and therefore reduced sensitivity, so output is non-linear with respect to increased power applied.

Thanks! So, the only distortion / compression we'd expect to see on transient peaks would be from what the author says about the voice coil moving beyond the magnetic field?
Quote:
Originally Posted by Audioholics Article View Post

Speaker “clipping” like amplifier clipping will occur when the excursion required to reproduce the signal results in the VC being propelled OUT of the magnetic field, reducing the motor force (B*L).

So this is more of a magnetic/mechanic compression than thermal compression, and it sounds like it could occur on peaks, too?

Quote:
Originally Posted by Bill Fitzmaurice View Post

This points out the fallacy of manufacturer 'maximum SPL' figures, because those are calculated as linear with respect to 1m/1w sensitivity, and assume that power compression does not exist. In reality you can expect the real maximum SPL to be 3 to 6dB lower than claimed.

So would you say that those numbers can be trusted with regard to typical transient peaks?
Reply
Reply
post #6 of 48
2/8/12 at 2:35pm
Quote:
Originally Posted by Eyleron View Post

Thanks! So, the only distortion / compression we'd expect to see on transient peaks would be from what the author says about the voice coil moving beyond the magnetic field?

That will be the most serious effect. The lows are squashed when the coil can't move far enough to reproduce them, the highs aren't, so high THD results.


Quote:


So this is more of a magnetic/mechanic compression than thermal compression, and it sounds like it could occur on peaks, too?

The power compression referred to by JBL is thermal, and is pretty much broadband. Mechanical compression caused by the coil reaching xmax in the lows only affects where the displacement limit has been reached.


Quote:


So would you say that those numbers can be trusted with regard to typical transient peaks?

Transient peaks don't cause thermal power compression because they have too short a duration to cause heating of the coil. So if you keep the average power level well below 1/2 Pe the coil won't get hot enough to cause thermal compression and the peaks won't be chopped so long as they aren't subject to the displacement limit of the driver.
Quote:


I will assume these suffer less compression from most because they can scare the crap out of you more than any speaker I have tried.

It's simply a matter of efficiency. They require less power for a given SPL, less power means a cooler voice coil.
Reply
Reply
post #7 of 48
2/8/12 at 2:40pm
Quote:
Originally Posted by Bill Fitzmaurice View Post

That will be the most serious effect. The lows are squashed when the coil can't move far enough to reproduce them, the highs aren't, so high THD results.


The power compression referred to by JBL is thermal, and is pretty much broadband. Mechanical compression caused by the coil reaching xmax in the lows only affects where the displacement limit has been reached.


Transient peaks don't cause thermal power compression because they have too short a duration to cause heating of the coil. So if you keep the average power level well below 1/2 Pe the coil won't get hot enough to cause thermal compression and the peaks won't be chopped so long as they aren't subject to the displacement limit of the driver.
It's simply a matter of efficiency. They require less power for a given SPL, less power means a cooler voice coil.

Yes, I think I use about 20 watts for the loudest peaks in movies.
Reply
Reply
post #8 of 48
2/8/12 at 5:23pm
Quote:
Originally Posted by Eyleron View Post


On the other hand, a 1 second peak of an orchestral stab or a film's gunshot is not the same thing as a 5 minute continuous signal.

If you look at the actual waveform of a musical piece-you will see that 1 second is a musical eternity. That is short of a few instruments that produce constant tones for long periods.

A gun shot (the peak anyway) is only a very small time period of 1 second. I would guess several microseconds. But have not measured it.

The peaks are VERY short.

It is the continuous-or average levels that do the heating that causes power compression.

When testing with pink noise-that can be a bit brutal-as compared with normal music program. The normal pink noise is either 6 or 10dB crest factor. Normal (whatever that is) music is on the order of 20dB crest factor. So it does not cause as much heating as pink noise.

Of course some musical styles are more like pink noise in their crest factor. So the real answer is that if varies with the musical style.

That is one piece of the puzzel that various people have been struggling with for many years-to simulate music signals with a regular test signal.

However-the nice thing about standards-is that there are so many to choose from!
Reply
Reply
post #9 of 48
2/8/12 at 5:44pm
There are peaks, then there are peaks... I suspect Bill was describing an orchestral crescendo (seconds) or kettle drum smack (decay of 1/2 - 1 s) that can last a while... As for microseconds, well, maybe, but not the period of a single 20 kHz cycle is 50 us, and 1 kHz is a period of 1 ms for one cycle. I would guess 10's of ms at least. Either way, the very loudest peaks are usually very short relative to everything else.
Reply
Reply
post #10 of 48
2/8/12 at 7:36pm
Quote:
Originally Posted by DonH50 View Post

Either way, the very loudest peaks are usually very short relative to everything else.

+1. It's high power over extended periods of that causes power compression. The usual victims are DJ speakers, especially subs, pushed too hard too long because the DJ doesn't have enough of them for the levels he's running.
Reply
Reply
post #11 of 48
2/9/12 at 7:17am
Thread Starter 
Alright, what I'm getting from this is:
Power compression (thermal) won't occur on peaks, because:
  • Transients are too short to heat up the VC significantly, and the drop to lower power allows time to cool.
  • If there were so much high peaks that you COULD heat up the VC, then we're actually talking about a higher continuous power, and so you need to be wary that your average levels are not over the speaker's "Program Power" rating, or 2x Continuous, or 1/2 Peak Power ratings.

If you do send more power in peaks than the speaker is rated for, then you're looking at either distortion and compression due to leaving the magnetic gap, stretching of the spider, or other mechanical limits such as bottoming out, depending on how the manufacturer chose to come up with the rating (onset of X % distortion, speaker destruction, etc.).

I was just confused by articles and posts speaking of power compression, and their sometimes mixing those concepts up with peak output.

Thanks for clearing this up!
Reply
Reply
post #12 of 48
2/9/12 at 7:32am
I always just thought of compression is when I turn up the volume it does not get louder. Basically when I measure turning the dial 3 dB's had the same measurement. This meant the peaks were now cut off or squashed compared to something that can still hit those peaks and the difference was dynamics! I always try to keep things simple as I do understand what Ivan and Bill were saying. It seems when it comes to dynamics they both know as heard in their products.
Reply
Reply
post #13 of 48
2/9/12 at 7:52am
Quote:
Originally Posted by MKtheater View Post

This is why I use speakers that can play much louder than reference and chose speakers that sound good doing so!

Agree - low efficiency designs are a sign of lack of speaker design skill.

Now that does not mean that every high efficiency speaker is designed well. There are good and bad sounding high efficiency speakers.

The bottom line is that you will always get better (more realistic) reproduction with a good sounding high efficiency speaker than a low efficiency one (can't reproduce dynamics)
Reply
Reply
post #14 of 48
2/9/12 at 8:10am
Thread Starter 
Quote:
Originally Posted by MKtheater View Post

I always just thought of compression is when I turn up the volume it does not get louder. Basically when I measure turning the dial 3 dB's had the same measurement. This meant the peaks were now cut off or squashed compared to something that can still hit those peaks and the difference was dynamics! I always try to keep things simple as I do understand what Ivan and Bill were saying. It seems when it comes to dynamics they both know as heard in their products.

But to measure compression and "it does not get louder" when we're talking about only peaks...that's difficult to measure. I read that most sound level meters won't register super short duration peaks. Does REW?

So to know whether the peak got louder or not, one is relying on one's impression of the passage, separated at least by several seconds from one test to the next. I think this is why people don't realize that due to amplifier clipping on insensitive speakers, or a lack of power handling in the speaker, or a lack of amplifier, they're not getting those peaks. It's more subtle and almost impossible for us to measure.

We measure with test tones, because that's not a peak signal, that's a continuous signal.

And when talking about average levels of film sound tracks at least, I think most 90+dB sensitive speakers can handle 85dB average level without compressing. Music is another issue, of course.
Reply
Reply
post #15 of 48
2/9/12 at 8:22am
Thread Starter 
Quote:
Originally Posted by Dionyz View Post

Agree - low efficiency designs are a sign of lack of speaker design skill.
...
The bottom line is that you will always get better (more realistic) reproduction with a good sounding high efficiency speaker than a low efficiency one (can't reproduce dynamics)

To be fair, I think you have to have a caveat about the desired listening level. If someone wants to listen to music as they heard it in an opera house seated far away, where the average level is the equivalent of -20dB down, then that's their prerogative.

Likewise, people may want to listen -15dBFS, well below reference level. They only need 70dB program levels and 90dB peaks. Or, they listen much closer to their speakers than do you and I. They will still get the dynamic range from average to peaks.

So I wouldn't say that those 86dB speakers are bad designs. I mean, even a 95dB speaker is a "bad design" with regards to listening 50 feet away. And then there's power handling. If the speaker sounds great with its low efficiency, but support peak power handling of 3200 watts, well that's great. It's not efficient, in the sense that a Ferrari doesn't win fuel economy awards, but it's not inherently a bad design. It's a design that may be bad for ME.

It's all about the intended purpose and goals.

EDIT: Okay, I have a question that maybe sways in Dionyz's favor: what about the noise floor? For movies, what's the lowest level of recorded soundtrack? At reference level, is there content below 30dB? How about 50dB? If there's content at 50dB and I'm listening at -20dBFS, then the 50dB content is pushed down to 30dB, so I'd better be sure my noise floor is well below 30dB. If it's not, then I'm missing soft sounds in the soundtrack, and it's not "realistic."
Reply
Reply
post #16 of 48
2/9/12 at 8:29am
Lowest level on a sound track (movie or otherwise is 22dB) Way below the noise floor in a quiet residential room (typically 30-35dB).
Reply
Reply
post #17 of 48
2/9/12 at 9:02am
Thread Starter 
Quote:
Originally Posted by Dennis Erskine View Post

Lowest level on a sound track (movie or otherwise is 22dB) Way below the noise floor in a quiet residential room (typically 30-35dB).

Thanks. Oof. Maybe I'd better find a way to mount the projector farther away, get rid of the 60Hz hum from my sump pump battery backup, and get my wife to move the cats' water fountain farther away than the adjoining room!
Reply
Reply
post #18 of 48
2/9/12 at 9:26am
Quote:
Originally Posted by Eyleron View Post

I'm trying to reconcile how we think in terms of peak output for program peaks, versus known levels of power compression in speakers with continuous signals (eg pink noise for 5 minutes).

You're asking many of the right questions, many enthusiasts don't want to even skim the surface.

---

Quote:
Originally Posted by Bill Fitzmaurice View Post

Power compression has little effect on peaks. Power compression results from long term heating of the voice coil, the result of which is that the resistance of the voice coil increases.

This is an aspect that rarely gets discussed, however power compression isn't only due to the thermal aspects of the heating of the voice coil. Yes, the majority of compression as we know it results from the ability of the voice coil to dissipate heat, but another component to compression is magnetic flux compression. Although the typical thermal compression most discussed is more long term, the effects of magnetic compression are instantaneous. These magnetic compression distortions, can dull transients, and render reproduction as lifeless, dull and unrealistic.

Certainly, thermal compression is a big issue. The vast majority of mfrs. have no idea of their products compression characteristics, other than their likely not so good. This is why typical residential speakers are inadequate for the huge dynamic swings associated with realistic HT. Many enthusiasts merely discount the effects of compression, because they are mostly subtractive in nature. But subtractive they are, compression robs the playback of life/realism. These effects are insidious because it's not an offensive distortion. It just thins out the presentation,...no life or snap. Oftentimes, without a direct comparison, the un-initiated have no idea of what they're missing.

The magnetic issue lies in the fact that when the drummer hits the kick drum pedal, and the beater strikes the head, the signal path results in the voltage is impressed across the loudspeaker's terminals. This results in the current flow in the VC, and subsequently the driver attempts to track the signal accurately. The problem is that when current flow begins in the VC, there exists three separate, and entirely different sources of magnetic flux in the gap. The permanent magnet's flux, the signal voltage VC flux, and lastly the flux that's generated by the varying eddy currents in the pole pieces. This is the problem, the flux modulation compression.

Whereby the thermal compression effects are a function of time, the magnetic flux compression is instantaneous. Magnetic saturation of any of the motor elements needs to be entirely avoided, as it's effects instantly impact waveform shape, the peak, and peak capability.

So just as reducing drive levels to avoid thermal buildup and subsequent losses, avoiding magnetic compression should be a goal to be mindful of even if it's only for the split second of the top of any given waveform. Depending on the frequency, we can reduce drive levels via spreading the signal over multiple drive units, ie multiple subwoofers, mid-bass units etc. Fortunately this flux modulation compression is most prevalent with high current levels that are typically associated with low frequencies,...thus enabling implementing multiple drivers covering the given passband without acoustic interaction issues that would accompany multiple drivers at higher frequencies.

I've experimented with the effects of compression, and be it thermal or magnetic, and spreading the drive signal across multiple drive units, thereby reducing the effects of potential compression, can be quite dramatic.

---

Quote:
Originally Posted by Bill Fitzmaurice View Post

+1. It's high power over extended periods of that causes power compression. The usual victims are DJ speakers, especially subs, pushed too hard too long because the DJ doesn't have enough of them for the levels he's running.

Good point Bill. Man I hate some un-informed DJ, hitting his speakers hard with clipped material, that's already badly compressed/high RMS material, and the mains and subs (if he's got 'em) are just sounding like garbage. That's how I got into DJ'ing (a brief foray) beginning back in the 80's. I just couldn't believe how bad some of the sound systems were, and how the DJ's operated them.

---

Quote:
Originally Posted by Eyleron View Post

Alright, what I'm getting from this is:
Power compression (thermal) won't occur on peaks, because:
  • Transients are too short to heat up the VC significantly, and the drop to lower power allows time to cool.
  • If there were so much high peaks that you COULD heat up the VC, then we're actually talking about a higher continuous power, and so you need to be wary that your average levels are not over the speaker's "Program Power" rating, or 2x Continuous, or 1/2 Peak Power ratings.

If you do send more power in peaks than the speaker is rated for, then you're looking at either distortion and compression due to leaving the magnetic gap, stretching of the spider, or other mechanical limits such as bottoming out, depending on how the manufacturer chose to come up with the rating (onset of X % distortion, speaker destruction, etc.).

I was just confused by articles and posts speaking of power compression, and their sometimes mixing those concepts up with peak output.

Thanks for clearing this up!

Compression, dynamic capability, transient capability, these effects are all tightly inter-related. In the context of high performance HT, these elements are essential for a realistic experience. Our ear/brain interface is incredibly forgiving in both the frequency domain, and in the time domain. I mean it's amazing how we can extract a decent experience out of some pretty bad acoustical environments. However, the realism that accompanies natural dynamics, accurate signal tracking, and the relative absence of the insidious effects of compression, can make or break the illusion. Realism in audio playback should easily have the ability to startle, to frighten, to convey an impending sense of doom. Yes most system do a decent job, but when you experience the realism that comes with a system that's designed to eliminate dynamic constraints, it peels away one last layer between you and the recorded event.

Good questions, dynamic realism is very important. Our Hunter/Gatherer brains clearly delineate the difference.

Good lookin' out
Reply
Reply
post #19 of 48
2/9/12 at 9:28am
How I measured was I would watch a scene at say -10MV and record the peaks of that scene. I would turn up the MV to -5MV and sure enough the peaks went to 5 dB's more. I then turned it up to 0 MV and it went up 2 dB's rather than another 5dB's. +5 MV resulted in the same level as 0MV so it is not getting louder just compressing. distorting, or going to blow up. This is just an example. I measured an SVS sub before which spec'd 124 dB's in a corner(no frequency specified). So I measured my SVS in the corner at 1 meter during WOTW and I would hit 120 dB's during the scene. I turned the volume up 3 dB's and it would never get above 120 dB's no matter how loud I turned it up so there was no point playing it a MV 0 because -5 MV was just as loud. I basically played it at the loudest it could take before it compress so the peak was 120 dB's and not higher. This sub had limiters and such so going higher did not blow it up. Usually without limiters you start hearing noises and such. Just a simple test, we used to have threads like this all the time with people testing their subs with different scenes to see where compression and their limits were so we could get an idea on what others were experiencing with the same scenes in their own rooms. Of course people turned it into a contest and then the anal members would say it was just a contest and was meaningless. It was fun to do and every room was different and it was nice to see what others were watching at. We measured at normal watching levels and then did the compression tests. It was more fun than just taking compression sweeps in REW although it was just graphing the same thing except using sine waves rather than actual movie content.
Reply
Reply
post #20 of 48
2/9/12 at 9:28am
Quote:
Originally Posted by Dennis Erskine View Post

Lowest level on a sound track (movie or otherwise is 22dB) Way below the noise floor in a quiet residential room (typically 30-35dB).

Great point Dennis.

The astute HT enthusiast, increases dynamic range via lowering their noise floor. This is an aspect I struggle with in my room.
Reply
Reply
post #21 of 48
2/9/12 at 9:50am
Compression anecdote;

Having DJ'd for a period, and FOH engineered everything from a small private event, to outdoor mega-shows, I've always experimented with various things.

While DJ'ing (for a while I'd liaise, procure the gig, set-up and tune, then let the pro spin the vinyl), I discovered how dramatic subwoofer compression can be with dance music.

I learned most of the shows I could do with a pair of subs. But I discovered bringing 4x what I really needed, gave the LF this immediacy, this quality that lacked compression. Essentially the same playback levels, just not running the subs so hard,... spreading the drive signal across several boxes,...just really changes everything for the better.

The LF became much more responsive even at the same playback level. This effect wasn't subtle, not at all,...it was a huge difference. It really had no more extension, no more level (I checked, I routinely monitored levels), it just had this apparent "speed", or lack of sluggishness that I never even knew was capable. You spread that drive voltage across more motors, everything seems to speed up. Yep, woofer speed

Several elements at play here; The lack of overshoot, the VC being in the linear Bl aspect of the gap, no frequency dependent thermal compression effects. And as I know now, no magnetic compression. I was loving what I'd discovered. Also, I only used the best LF drivers, and maintained great working condition. No rub, no buzz,..and no rattles etc.

From that time forward, upon discovering that, I've always way over spec'd LF for any event I'm involved with. (as long as the client pays) One very memorable event that I'll never forget; a very nice small, outdoor live event I did. I played music for a couple hours, the act performed for a couple hours, and I played more music the remaining time. I brought (8) double 15" JBL's with all new drivers. I only needed 1 top, and 1 sub per side, but I brought 2 tops merely for wider frontal coverage (front fill actually), which I aimed tightly inward for front row seating. Anyway, stunning LF, just simply stunning. No room modal issues being outdoors, just solid extension down to 25-30hz. The mains had TAD HF beryllium compression drives in them, the finest,....very sweet, like $2k or so a piece just for the compression drive. I loved those things,....nothing I heard around that time was anything close. All sounded well, but I noticed some imaging/interaction problems,..kinda a blurring of the MF/HF. So to find the issue, I disconnected the inward fill mains, and only had the one regular two way per side. BAM! Perfect. The entire system became perfectly focused,..no destructive interference from adjacent boxes not meshing, no interference (this is huge) from acoustics being outdoors, just wonderful. Control room type studio monitor clarity. No sidewall issues, no modal issues,.. just a wonderful window into the recording.

The take-aways here include; headroom is our friend,...this comes up so, so often. Also, (OT but whatever) there is so much negative interaction between a multi cabinet approach to MF/HF. Even the engineered "J" line arrays have issues, that's why the work Tom Danley's doing in the mega/single cabinet pro live sound stuff makes so much sense,...plus the dude an audio genius anyway.



Thanks
Reply
Reply
post #22 of 48
2/9/12 at 9:55am
Thread Starter 
Quote:
Originally Posted by MKtheater View Post

How I measured was I would watch a scene at say -10MV and record the peaks of that scene.

Was this with REW in RTA mode, where you can watch the spectrum components and see the THD?

I suspect this is limited by the duration of the peaks and the frequency one is attempting to capture?

I'll have to try this. I've watched THD on sine waves, but I haven't tried recording signals and playing back to analyze yet.
Reply
Reply
post #23 of 48
2/9/12 at 10:07am
Quote:
Originally Posted by Eyleron View Post

To be fair, I think you have to have a caveat about the desired listening level. If someone wants to listen to music as they heard it in an opera house seated far away, where the average level is the equivalent of -20dB down, then that's their prerogative.

Likewise, people may want to listen -15dBFS, well below reference level. They only need 70dB program levels and 90dB peaks. Or, they listen much closer to their speakers than do you and I. They will still get the dynamic range from average to peaks.

So I wouldn't say that those 86dB speakers are bad designs. I mean, even a 95dB speaker is a "bad design" with regards to listening 50 feet away. And then there's power handling. If the speaker sounds great with its low efficiency, but support peak power handling of 3200 watts, well that's great. It's not efficient, in the sense that a Ferrari doesn't win fuel economy awards, but it's not inherently a bad design. It's a design that may be bad for ME.

It's all about the intended purpose and goals.

EDIT: Okay, I have a question that maybe sways in Dionyz's favor: what about the noise floor? For movies, what's the lowest level of recorded soundtrack? At reference level, is there content below 30dB? How about 50dB? If there's content at 50dB and I'm listening at -20dBFS, then the 50dB content is pushed down to 30dB, so I'd better be sure my noise floor is well below 30dB. If it's not, then I'm missing soft sounds in the soundtrack, and it's not "realistic."

If you need 3,200 watts rather than 50 watts to get the same quality and sound level output, then it is poor design. It is a sign that the speaker designer incapable of efficient design.

Why on earth would I want to be using a 3,200 watt amp or 64 times more power, to get the same sound (which is what I care about)?
Not to mention you would need to increase you air-conditioning system capacity to accommodate these huge amps.

Thus low efficiency speakers ARE poor design - show speaker designers incompetency and they waste energy
Reply
Reply
post #24 of 48
2/9/12 at 10:08am
Thread Starter 
Quote:
Originally Posted by FOH View Post

Whereby the thermal compression effects are a function of time, the magnetic flux compression is instantaneous. Magnetic saturation of any of the motor elements needs to be entirely avoided, as it's effects instantly impact waveform shape, the peak, and peak capability.

So, to address my question about compression on peaks:
If we have a JBL speaker that they rate max power:
  • 150w continuous (which we see with a sine wave)
  • 300w program (which I take to mean some typical crest factor of music or movies) (and this is the level most often recommended of amp RMS to use in the pro world)
  • 600w peak (which for compressed music might be frequent kick drum, or for movies might be at worst a couple minutes of pounding music with several seconds of gun shot transients at a time)

...where we see a stated thermal compression for the speaker of, say, .7db@800w, 2.5dB@400w, and 4dB@800w,

would you expect to see magnetic compression when playing difficult music peaks, but you keep the power <= 800w peak? How about for films?

Or do you expect that staying under 800w probably means we're avoiding magnetic compression?

Note again that the JBL reported compression for their 600w peak was with a continuous signal, which means they're over-driving their speaker.
Reply
Reply
post #25 of 48
2/9/12 at 10:21am
Thread Starter 
Quote:
Originally Posted by Dionyz View Post

If you need 3,200 watts rather than 50 watts to get the same quality and sound level output, then it is poor design. It is a sign that the speaker designer incapable of efficient design.

Why on earth would I want to be using a 3,200 watt amp or 64 times more power, to get the same sound (which is what I care about)?
Not to mention you would need to increase you air-conditioning system capacity to accommodate these huge amps.

Thus low efficiency speakers ARE poor design - show speaker designers incompetency and they waste energy

Gotcha. The JTR Triple 8 is a poor design, even though it's a 98dB sensitive speaker, because its peak power handling is probably 2400 watts, and that's what I'd need to get 105dB peaks to the back row of people @ 70 feet at my concert.

Again, it's about your goals for speaker. An RV is not good at drag racing. A Ferrari is not good to take on a camping trip.

Even though my current interest is with 92dB and up sensitivity speakers, that's only because my goal is clean home theater sound @ > -10dBFS @12 feet in a moderately treated room, without having to buy a lot of amps.

If an 86dB speaker will do that with a good 100w amp and sound great, then I don' see how that's a bad speaker. Personally, I want more headroom than that. If my speaker could handle 200w instead of 100w, I might have already bought an XPA-3.

If I wanted to be -15dB, then an even less efficient speaker would be fine. Why are those bad speakers?
Reply
Reply
post #26 of 48
2/9/12 at 10:27am
Quote:
Originally Posted by Dionyz View Post

Thus low efficiency speakers ARE poor design - show speaker designers incompetency and they waste energy

It's a design choice.


Low, Loud, Small,.....pick two
Reply
Reply
post #27 of 48
2/9/12 at 10:50am
Quote:
Originally Posted by Eyleron View Post

would you expect to see magnetic compression when playing difficult music peaks, but you keep the power <= 800w peak? How about for films?

Or do you expect that staying under 800w probably means we're avoiding magnetic compression?

Note again that the JBL reported compression for their 600w peak was with a continuous signal, which means they're over-driving their speaker.

I don't know how significant it would be, but yes, there'd be magnetic compression. It's a matter of relative significance,..as it's inherent to the process, the physics of the electro-magnetic principles involved.

Thermal compression is generally regarded as the dominant factor in compression. These thermal issues can be measured, etc. Magnetic compression on the other hand is difficult to measure without level of sophistication/equipment etc. But, clearly one can lessen it's significance by reducing levels/multiplying drive units etc.

The appeal to listen to high-end two channel, high sensitivity gear is correctly placed. The dynamic realism afforded is stunning. Likewise, the appeal toward electro-stats is similarly appropriate. They exhibit none of these effects. Yes, they bring their own issues. Upon hearing a properly set up set of Acoustat electrostatics is what launched me into audio pursuits full tilt. Wow, nothing else can do that. This is back when HiFi salons had nice rooms etc., the dude came back, opened the sliding glass door and into the room. I asked if he would turn it up,....he chuckled, and said "that's as high as those will play with distorting". I immediately said that sucks and left. All said, I still today think back to that experience,...it was that good.

30 years later, the sales dude, Vic Suarez, still selling gear today.
Reply
Reply
post #28 of 48
2/9/12 at 11:01am
Thread Starter 
Quote:
Originally Posted by FOH View Post

I don't know how significant it would be, but yes, there'd be magnetic compression. It's a matter of relative significance,..as it's inherent to the process, the physics of the electro-magnetic principles involved. ... Magnetic compression on the other hand is difficult to measure without level of sophistication/equipment etc.

I can imagine measuring changes to the flux is difficult to measure. But do you think measuring as MKTheater reported would work? Record the playback, then monitor for peak levels?

It'd be good to know that the speaker you thought could handle X watts peaks really only handles half that before compression. And if it's the woofer that's compressing, and not the tweeter, then the system frequency response will be getting out of balance.

Have you seen speaker manufacturers provide measurements of peak distortion/compression? Or is it just continuous mode?
Reply
Reply
post #29 of 48
2/9/12 at 11:05am
Thread Starter 
Here's an interesting thread about peaks on sound and vision mag forum: http://forums.soundandvisionmag.com/...happens)/page5. He looks at the 1812 Overture.
Reply
Reply
post #30 of 48
2/9/12 at 12:38pm
Quote:
Originally Posted by Dionyz View Post


Thus low efficiency speakers ARE poor design - show speaker designers incompetency and they waste energy

Tell that to Don Keele.

Horns ruled in the day of tubes and early SS. Once 200 watt and better SS amps became available and the cost of watts fell Don famously wrote that if you have more time than money build horns, if you have more money than time buy direct radiators and amps large enough to drive them. That scenario held for 30 odd years, but today between the cost of fuel to transport pro-touring rigs and HT owners who play at reference the pendulum has swung in the other direction.
Quote:


Low, Loud, Small,.....pick two

You may go low, loud and small, but at the expense of using uber-xmax drivers and uber-watts to push them. And then there's the matter of distortion, which is proportional to excursion. Horns still do everything best in terms of output, distortion and economy of cost, but at the expense of economy of size.
Reply
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Speakers

AVS Top Picks

AVS › AVS Forum › Audio › Speakers › Effect of Compression on Peaks