• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Speaker Linearity with Volume Change

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
7,473
Likes
4,189
Location
Riverview FL
#1
Ok, sometimes I have a little too much time on my hands and I find yet another way to waste it.

One of my neurons developed a need to know how linear (if at all) my system is at the listening position at reproducing relatively "correct" levels. The next thing I knew, some other neurons got all agitated and this measurement ensued.

I took 24 sweep measurements decrementing the beginning attenuation at my DSP by 2dB per step for 24 steps and 46dB, to see if there was anything to note.

The lowest sound pressure level is chosen around the point the signal rises out of the noise floor of the room and the measurement tool at 1kHz and above. The bass frequencies are still "noisy" at the low measurements but the higher frequencies are pretty clean.

Using 3kHz as an anchor, I see a spread of 46.0dB to 91.9dB, for a range of 45.9dB rather than the expected 46. with 1/3 octave smoothing. That result might be closer than I expected.

Looks like the air conditioning was on for a few of the lowest traces - hump at 18Hz.

1/3 smoothing

upload_2016-10-23_13-48-42.png


1/24 smoothing

Ambient and tool noise obvious at low level and low frequency.

upload_2016-10-23_13-49-38.png


For those unfamiliar with a 46dB volume attenuation range - The daily blow-by-blow of the Trump vs Clinton Celebrity Death Match is on at a fully satisfying level (86dB peak, 66dB average), then attenuate by 46dB, the result is a just barely audible something with nothing intelligible, the sound not even recognizable as voices.
 
Last edited:

Blumlein 88

Major Contributor
Joined
Feb 23, 2016
Messages
7,069
Likes
7,644
#2
Good work Ray. The only published results of such things I know of are at soundstage.net in their speaker measurements. They will show a difference in response between a couple loudness levels. Not as wide a level difference as you did here. Most modern speakers of any quality do quite well in their measurements. Again though not over a 46 db range of loudness. So I guess that level linearity is one less thing to worry about.
 

fas42

Major Contributor
Joined
Mar 21, 2016
Messages
2,818
Likes
142
Location
Australia
#3
Yes, I agree - excellent job, Ray. Particularly that you show 1/24 smoothing, all the little wriggles form neat tram lines. As regards 46dB range, people should try experimenting with 60dB gaps - I did this 30 years ago, with a test CD ... very enlightening, at -60dB signal, maximum volume, ear pressed hard against the treble driver I could just make out the tune.

Of course, I could be nasty and suggest that you push somewhat higher in SPLs, to see at what point significant anomalies start appearing ... :p :D.
 

tomelex

Addicted to Fun and Learning
Patreon Donor
Joined
Feb 29, 2016
Messages
759
Likes
282
Location
So called Midwest, USA
#4
The last two comments moved us to dynamic range and room ambient. This is where we start talking about that first watt and the nearly 30db of dynamic range in that first watt, and then to get the next 30db of dynamic range you need be up to 1000watts. Its crazy stuff.
 

fas42

Major Contributor
Joined
Mar 21, 2016
Messages
2,818
Likes
142
Location
Australia
#5
This is where we start talking about that first watt and the nearly 30db of dynamic range in that first watt, and then to get the next 30db of dynamic range you need be up to 1000watts. Its crazy stuff.
Sorry, Tom, I don't quite follow your logic here ...
 

tomelex

Addicted to Fun and Learning
Patreon Donor
Joined
Feb 29, 2016
Messages
759
Likes
282
Location
So called Midwest, USA
#6
on your speakers, if the first watt gets you about 30db of dynamic range, then to get 30 more db of dynamic range you need 1024 watts. All the low level details, ambience, and a lot of clarity are in that first 30db. The first watt, with efficient speakers gets through to you, the first watt with inefficient speakers is mostly lost trying to move through the passive crossovers and sluggish motors of the inefficient speakers. that's why headphones are so dynamic, fast, and once you get on board, its hard to get off unless IMO you go the route of super efficient speakers.
 

fas42

Major Contributor
Joined
Mar 21, 2016
Messages
2,818
Likes
142
Location
Australia
#7
Yes, if 1 watt gets you 90dB at 1 metre for average sensitivity speakers, then theoretically you need to pump that 1,000W into the same speakers for a 120dB peak. But one never needs 120dB, unless you want headbanging listening! An orchestra playing live only generates about 110dB peak for a close audience member - such a live situation gives one enormous dynamic range, and the power requirements for the electronic equivalent are much reduced.

What the real problem is that an audio system will often fail to produce that 110dB situation cleanly, and also fails to get the silences in between the loud bits right. Hence, poor subjective dynamics ... one of the first things I do experiencing an unknown system is to try a recording which immediately points to whether the rig is capable of reproducing decent dynamics - if not, which is the usual result, then I scale back my expectations dramatically - it's inherently limited, no point in trying to thrash an underwhelming car ...
 

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
7,473
Likes
4,189
Location
Riverview FL
#8
The first "watt" as defined by 2.83Vrms into 8Ω gave me 87dB which would be 50dB above my A-weighted ambient noise floor.

That's two watts with a 4 ohm speaker, so, for me, at the listening position, something like...

87dB- 2W
97dB - 20W
107dB - 200W

That pretty much covers my needs, I see 107dB on an occasional peak when playing something dynamic a little too extra loudly. Right now, peaks are about 97dB,

The amps are rated 350/700/1400W into 8/4/2Ω and 49rms or 69pk volts out. The speakers call for 20-500W.

---

Listening right now, I saw 102.2dB peak, and my trusty Fluke measured a max of 24.16V...

upload_2016-10-25_3-45-29.png


Nice ballpark numbers...

Heat sinks 120F and 122F.

--

Now 104.8dB peak and 29.32V at the meter...

upload_2016-10-25_3-56-6.png
 
Last edited:

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
7,473
Likes
4,189
Location
Riverview FL
#9
Looking at the Impulse Response overlay of the series of measurements in the original post:


upload_2016-11-11_15-54-42.png


And zooming in on a later time period, at 25 ms:

upload_2016-11-11_15-57-49.png


I didn't expect to see what would appear to be an indication that "softer sounds travel more slowly than louder sounds". Ok, it's a very slight delay creeping in, but still...

Is this true?
Not true?
Some measurement artifact?

after 82 ms:

upload_2016-11-11_16-3-19.png


after 163 ms:

upload_2016-11-11_16-13-58.png
 
Last edited:

Blumlein 88

Major Contributor
Joined
Feb 23, 2016
Messages
7,069
Likes
7,644
#10
Nice pretty pictures oooh!

Hey the diaphragm hasn't moved as far on softer sounds. So it has to travel further to the microphone. Well on the front half of the wave. I guess it travels less on the back half of the wave.

That is an interesting artefact though. Will have to think about it some.
 

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
7,473
Likes
4,189
Location
Riverview FL
#11
Hey the diaphragm hasn't moved as far on softer sounds. So it has to travel further to the microphone.
That doesn't seem to affect the initial impulse. They are all nicely stacked. Then I thought "Well, the acoustic timing reference is related and would experience the same effect blah blah blah" but then I remembered that the quick HF timing reference tone is the same volume prior to each trace whatever their attenuation was, so the measurement timing shouldn't slip with attenuation of the test sweep tone.

---

My thought going into this was to see how long it took for "recognizable" reflections to devolve into noise, either by diffraction chopping them up (?) and making a mess, or just attenuation making them disappear (which is mostly what I got) but saw this interesting cascade in time.

The "ladder" of amplitudes would (and does seem to) identify reflections of the sweep tone vs ambient or random noise.
 
Last edited:

Cosmik

Major Contributor
Joined
Apr 24, 2016
Messages
3,075
Likes
1,929
Location
UK
#12

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
7,473
Likes
4,189
Location
Riverview FL
#13
Maybe. The test only spanned 6 minutes though

The Air Conditioner could have made a difference.

Ok, need to try it again with stable air temp...
 

Similar threads

Top Bottom