What is "reference level"? - Page 5 - Home Theater Forum and Systems - HomeTheaterShack.com

Old 02-22-12, 07:37 AM
waricle

Join Date: Jan 2010
Location: bribie island QLD Austral
Posts: 4
Re: What is "reference level"?

Love the disintegrating house smilie!
You wouldn't be dead for quids, would you?
waricle is offline

Old 03-19-12, 04:16 AM
Shackster
David

Join Date: Dec 2009
Location: Minot ND
Posts: 29
Re: What is "reference level"?

Quote:
gsmollin wrote: View Post
No, actually not 10 kW. According to your example, you need 20 kW. The problem comes down to what we mean by "Watts".

The first mistake in all the calculators is that there is no such thing as "Watts RMS" in these calculations.
I think you're conflating amplifier dynamic headroom with spl formulae?
It doesn't matter how you get there: it will take a number of watts using a number of distance to yield a number of SPL.

Since 105dB is referring to peaks, and not continuous, then this would correspond to transient program signals (typically < 200 ms) and an amplifier's peak undistorted transient output watts, which for highly regulated amps might be the same as RMS, and for other loosely regulated amps might be another 3dB (2x) of watts.

If your calculations of tens of kW in a home were true, then we would also never reach even 95dB in the home without thousands of watts per channel. I can reach 95dB from one speaker @ 10ft with a continuous sinewave, let alone a peak.

Quote:
gsmollin wrote: View Post
Now, in your example, you were quoting a 20 dB crest factor. I only subscribe to 10 dB, which I have justified above. However, the broadcast industry's program material model may be too compressed for the movie industry, and 20 dB may be closer to the truth. If that were true, then add another 10X to amplifier, and get 20 kW required power.
Unfortunately, for music, one must guess, average, or survey for different types of music and their crest factors (average to peak ratio). Highly compressed pop music can be 3dB. Very dynamic music like certain classical will be over 20dB.

Fortunately, for films, you don't have to "subscribe" to any crest factor, as there are standards. 20dB is coming from 85dB average level with peaks of 105dB per channel (excepting LFE and summed bass, which can be 115 to 118dB).

Last edited by eyleron; 03-19-12 at 04:25 AM. Reason: Added film 20dB crest factor
eyleron is offline
Old 03-19-12, 04:43 AM
Shackster
David

Join Date: Dec 2009
Location: Minot ND
Posts: 29
Re: What is "reference level"?

Quote:
Sir Terrence wrote: View Post
Since nobody has really address this question, I will. Reference level is not 75db or 85db. It is actually 0 digital reference, or the loudest digital signal before system overload. So reference levels for the cinema would be 105db for each main channel, and 115db for the LFE based on 0 reference. This would be continuous output, and nobody can really stand this kind of output continuously.
While you're right about reference level referring more accurately to 0dBFS, you're mixing up average dialogue level with peak level. Peak is 105dB. Average is 85dB.

You're right that people can't stand 105dB, and fortunately, they never need to. They can withstand 105dB quite readily.

Speaking of tolerance, I didn't see a mention of how it's difficult to relate one's experience with 105dB peaks, due to the varying methods that people employ to test, and the varying qualities of home theater systems. It's fairly difficult to achieve 105dB peaks undistorted in the main channels, and it is VERY difficult to achieve the accompanying LFE peaks of 115dB-118dB!

For main channels, we need real high power handling speakers, and the amps to feed them hundreds of watts per channel, or speakers that are sensitive enough to achieve that output on less power. This usually means sensitivities in the low-to-mid-90s, at least.

To test, one cannot simply play a 105dB sine wave, as that's a continuous signal, which is much more challenging on the amp and speakers. While speaker and amp manufacturers tried to quote spec's based on 80ms peaks in the past, I see 200ms quoted more often, and I read that in an AES paper recently.

There are short burst tones that one can use to simulate a program peak. I think a lot of SPL meters also may not be able to log a max spl of such short duration, but I don't know. Does anyone know what the limits are of the Radio Shack and Galaxy meters, and a common rig of EMC8000 & REW in recording brief peaks? Ideally, one could output burst tones that could be monitored for THD.

You can also find out the recorded program peak of a film, like "-2dBFS @ 200Hz @ 1:19:04" and monitor to see if you achieve that, and how clean that sounded. It's more difficult to discern distortion in the brief peaks, but at some point one's speaker will make bad noises instead of good sound, or it just simply won't achieve that desired level.

Clean, undistorted, unclipped peaks are revelatory. It doesn't sound so "loud," but rather more "real" and effortless. Soundtracks often sound "loud" due to the distortion from the amp or speakers. Or from the reflections from an untreated room.

What are y'all's thoughts on what the threshold should be for achieving reference level LFE? 20Hz@115dB? 10Hz@115dB?

For those interested, I have a list of reference level speakers. It contains some classic high sensitivity speakers, some uber-sensitive ones, and a lot of pro speakers. The latter range from "colored" stage speakers, to studio nearfield, midfield monitors, and control room speakers (for big rooms), to "audiophile" designs to \$20,000+ designs. There are also some near-reference-level speakers to educate using edge cases.
The list is sorted by output level, and I think it's interesting to see one model stick out like a sore thumb (due to the color grading), and then I see that while it had lower sensitivity, it has high power handling (which means it's a reference level speaker as judged by its own specs) and just needs to be run with a high power amp.

Last edited by eyleron; 03-19-12 at 10:10 AM. Reason: added clarification about list.
eyleron is offline
Old 03-19-12, 04:50 AM
Shackster
David

Join Date: Dec 2009
Location: Minot ND
Posts: 29
Re: What is "reference level"?

Quote:
GranteedEV wrote: View Post
However that doesn't factor in thermal compression. Even if your amp can drive your speakers to 105db, your speakers may lose their compusre at those peaks.

Most people's systems can't do reference levels properly. Regular tweeters just don't have the thermal power handling to do so.
While I agree that a common design to achieve these higher levels is using a horn, or a horn/compression drivers, I don't think thermal power handling comes into play for the peak levels being discussed. From what I read, and what speaker designers say, thermal compression is for longer signals. The brief transient peaks are limited more by other mechanical compression and magnetic compression effects.

Power compression (thermal) appears at even 1/10th the speaker's rated power handling, as a fraction of a dB reduction in output at certain frequencies. At 1/2 power, I have seen 1.5dB, and at full rated power, 3-4dB variances. This now only reduces output, but it changes the frequency response of the speaker.

A few manufacturers like JBL Pro publish these numbers. I think we all wish that more did!

Last edited by eyleron; 03-19-12 at 04:51 AM. Reason: clarified that 'no thermal compression' is in regards to peaks
eyleron is offline
Old 03-19-12, 12:32 PM
Elite Shackster

Kick logic 2 da curb

Join Date: Aug 2010
Location: Calgary
Posts: 1,745
Re: What is "reference level"?

Quote:
eyleron wrote: View Post
While I agree that a common design to achieve these higher levels is using a horn, or a horn/compression drivers, I don't think thermal power handling comes into play for the peak levels being discussed. From what I read, and what speaker designers say, thermal compression is for longer signals. The brief transient peaks are limited more by other mechanical compression and magnetic compression effects.
How long is brief? How short is brief?

This is something I was told by a driver designer:

Quote:
Paul Appollonio wrote:
Music definitely has peaks, and averages, but they move about so quickly without a quantifiable definition we are waving at flies with our hands. Let us assume a peak with a duration of 0.1 second. That is a reasonable amount of time to call a peak a peak. We could integrate it longer or less, but I would argue integrating less than 33 milliseconds is likely useless, as that seems to be the approximate integration time of our ear brain mechanism. So, regardless of the duration of the peak, the VC will eventually get hot. What you find in a speaker is two time constants, one for the VC, and another for the motor surrounding it. You could count the box as a third, but this is a tertiary effect. The VC time constant of a small woofer like this is probably on the order of 3 seconds. Now, if you look at music which is typically played or recorded today, you will find the actual dynamic range QUITE limited. This means the difference between peaks and Rms is now very small. That makes a difference. Todays recordings have a dynamic range which is very limited relative to for example, a recording of a symphony orchestra.
...
As for heating and compression, this is not difficult to measure. ANYONE who can come up with \$15 and find their way to Radio shack can buy a meter that will measure DCR. Measure the speaker at the terminal with the amp disconnected, and the woofers cold. Play your speakers for an hour, as loud as you want to, then quickly disconnect the amp, and measure the speaker again. As long as there is no loud noise after you turn off the speaker, or as long as you don't have a fan or wind blowing at the cone, you should be able to get a somewhat stable DCR reading. That will tell you just how far off center the heat has pushed the device. I can measure it other ways, but I've thousands of dollars worth of gear to do that. You can calculate the heat in the VC based on the difference between the starting and stopping DCR (assuming any series choke presents a negligible amount of DCR in addition to your VC). All speakers heat up. My point was if I want 105 db EITHER peak or Rms in the room (RMS would be loud too) then having 8-10 db more sensitivity to begin with is a big advantage.

Don't underestimate us! Whether it be time, space, or alternate realities, they mean nothing to us! We will pierce the path and methods you've chosen! That is who we are!
GranteedEV is offline
Old 03-19-12, 06:39 PM
Shackster
David

Join Date: Dec 2009
Location: Minot ND
Posts: 29
Peak Duration, How to Test

Great point...how long is a peak? It'd be nice to know how long THX thinks it is, in regards to their amp and speaker testing.

I've read 200ms a few times, or the typical length of a quarter note. Here's an AES paper I read about it, where 200ms was considered a typical peak duration in many genres of music: The Dynamics of Recorded Music. The paper quotes another study saying that music peaks range from 8dB (rock extra compressed by the FM radio station) to 20dB (probably classical, and would apply to films) above average levels.

The thesis was that for typical program material < 200ms, on a budget one could use an amp that had enough dynamic headroom for planned peak levels, with the RMS being a fraction of peak output.
The worst cast peak duration in some pieces still suggests that to cover those worst cases of over 200ms, one should consider the amp's RMS rating and not the dynamic rating, as some peaks were up to two seconds long.

Here's another AES paper (that I only read the abstract on) that says 80-200ms: A Musically Appropriate Dynamic Headroom Test for Power Amplifiers.

So, if one accepts the above and wants to apply to peaks for speakers, the questions are, "How much heating does the voice coil undergo in 200ms," and "How much heating in 2 seconds?" I definitely can't test the 200ms per se, unless I download some shaped tone burst files and have some means to script their playing, but as you quoted, it looks easy to test resistance after playing some program material like music or a film soundtrack.

I suppose a 2 second sinewave could simulate the worse case.

Last edited by eyleron; 03-19-12 at 06:42 PM. Reason: added how on a budget they suggested dynamic headroom in amps
eyleron is offline
Old 12-23-12, 04:37 PM
Shackster
David

Join Date: Dec 2009
Location: Minot ND
Posts: 29
Re: What is "reference level"?

I have since learned that it's not heating of the voice coil we have to worry about for short term peaks, but rather Xmax, mechnanical excursion, and magnetic flux effects. Is the magnetic field linear as the driver is forced into greatest excursion?

Drivers that are loafing along using a small fraction of their peak-handling capability are less likely to exhibit distortion.
Those that are at or beyond their limits will distort those transients even more.

The distortion may be an altered frequency response, as some of the drivers handle the signal differently, and it can be as benign as simply "lopping off the tops of the waves," which means you've lost that visceral immediacy from losing dynamics. Instead of a short 1/5 second rise of 25dB, you may only get 20dB, for instance.
eyleron is offline
Old 01-02-13, 06:33 PM
Shackster
David

Join Date: Dec 2009
Location: Minot ND
Posts: 29
Re: What is "reference level"?

Quote:
eyleron wrote: View Post
While you're right about reference level referring more accurately to 0dBFS, you're mixing up average dialogue level with peak level. Peak is 105dB. Average is 85dB.

You're right that people can't stand 105dB, and fortunately, they never need to. They can withstand 105dB quite readily.
Hmm I see well after the fact I meant, "They can withstand 105dB for a fraction of a second quite readily."
eyleron is offline
Old 02-22-13, 05:03 AM
Banned!

Join Date: Feb 2013
Posts: 5
Re: What is "reference level"?

really great information for that concept i am totally unaware about that.
georgeallen is offline
Old 03-20-13, 02:23 PM
Shackster
Nid

Join Date: Feb 2013
Posts: 58
My System
Re: What is "reference level"?

Quote:
tonyvdb wrote: View Post
Reference level is the volume at 75db (decibels) in your room using full band pink noise. When you set up your system and you run pink noise through your speakers one at a time setting each level using an SPL meter to 75db you will (after all channels are done) and your volume control is set to 0db you will achieve "reference" when playing back movies or music. Movies will tend to reach peaks of 115db with your system set up properly.

This is not always easy to do as most home theatre systems can not reproduce this without distortion and is usually louder than most people go. Unless you have fairly expensive gear and larger speakers you likely wont achieve this without distortion.
Just a question, when you say set the volume at 0dB, do you mean the AVR volume ? My Yamaha AVR's volume display reads in negative values at low volume( it starts at -80, then goes up to +15 ) so the closer you get to 0, the louder it gets, and if you go louder, the numbers become positive values up to +15. At 0 volume, it is excessively loud, and past half volume, -10 is very loud. I find this negative volume display very confusing on Yamaha AVRs..can anybody shed some light on why it's a negative value on the loud end ? How do you read it ? ( I also find the dBFS negative reading confusing on dB meters, but I'll start a thread somewhere else for that ) Thank-you in advance.
Nitrofreakman is offline

 Bookmarks

 Tags "reference , level"? , meant , reference level

Message:
Options

## Register Now

Random Question
Random Question #2

User Name:
OR

## Log-in

Human Verification

In order to verify that you are a human and not a spam bot, please enter the answer into the following box below based on the instructions contained in the graphic.