SPL METERS????

[justin_tx_16]

I got the digital one because it measures averages for you, peaks, and other goodies.

[WMcD]

I'm not saying the situation is all bad.

I've used the RS with a test CD with 1/3 rd octave noise. This was "Sound Check II". Then put the data into a spreadsheet to graph it. In my case the speakers were in the garage and I used an HP-200 palmtop with Lotus 1-2-3. Excel running on anything would work too.

So, in about 20 minutes or less, you can get some reasonably accurate results for a small investment. It pretty much equivalent to an RTA.

Some perspective is necessary. In the very early days of audio, a decent mike and voltmeter with a dB scale would be considered a good starting point. It would be handy to have a chart plotter ganged to a 1/3 octave warble tone. You'd get a line graph.

The other issue about the room effects were fully appreciated. The goal was to test the speaker without the room. Maybe that is not our goal though.

Folks would drag the speakers outdoors. For a half space measurement they'd be layed on their back or for full space, put on a pole. The LMS system manual suggests pretty much the same. (They have the box standing up in the middle of a large driveway. But same idea.) The LMS is one of the more popular computer based measuring systems. You'll see their logo on many published curves.

When we visited Klipsch in Indianapolis, they did some test for us out on a concrete patio. They also had a three sided corner where an engineer (was his name "Dione"? maybe someone can confirm) ran demonstrations of sub woofer tests. The results were fed to an LMS.

And Trey showed us some tests in the anechoic chamber.

The point is, any microphone / meter is only part of the test set up. This is no fault of what we buy from RS. We also have to keep an eye on what we want to measure. (I'm getting to be my usual pedatic self here; sorry.)

I've read a bit about PWK's technique on measurements in rooms. I suppose he used a Klipsch "Logerator" plotter. But the issue was how to graph the combination of responses from three positions or more. He just ran them over on the same piece of paper, and then just took the highest levels of the combined trace.

Is that valid? Probably as good as anything.

The odd thing with constructive and destructive interference in rooms is that when when two equal signals arrive in phase, they combine to a 3 dB increase. But when they are out of phase there is a null, which can be 40 dB deep or more. You can hear it with a single tone if you walk around. It is difficult to detect otherwise with music.

The LMS and other computerized systems have gating which can allow a measurement of high frequencies. The microphone listens to the direct signal, and then turns off before the room echo arrives. So it can ignore room response. But that is not what we hear, in a room.

Another way of filtering out the nulls is to do 1/3 rd octave averaging of the raw data in any measurement set up. The deep nulls go away in the chart (and this is probably close to PWK's multiple trace). Warble tones do the same by having the input, rather than the output, averaged over a 1/3 rd octave range.

Is this fair? Yeah probably. The problem is that actual room response with a continuous plot is very, very ragged. It looks like what we see on TV when the guy on the polygraph has just lied, or the seismic graph shows the West Coast just got "the Big One". It is such a mess you can't see any trend.

But back to the subject, and IMHO.

One: The RS meter is not a bad piece of equipment. Rather, it is equivalent to two components of much more complicated set ups for measuring speaker response and room response.

Two: At best, if we are measuring room response, we wind up with the problem of a very jagged plot in one location; testing in multiple locations with several jagged plots; and then wondering how to massage the data so it tells us something more clear.

All for now.

Gil

[mace]

I've got the digital Rat Shack meter. It helps to measure peaks if you paint little white marks just below every 2 LCD blocks on the meter. It makes it easier to get the peaks from the LCD real time bar at the bottom.

I've noticed some funky stuff on the response curves I've generated (in my car, old boom box, KLF-30's, Cornwalls, Botson T-930's). I'll post more on that later. I used the 1/3 octave warble tones on the Stereophile Test CD 3. I do not know why this type of full spectrum testing is worthless compared to that frequency analyzer thing which someone posted a link to.

I found using both warble tones and single frequency tone analysis in my listening room very interesting. The room makes a huge, huge impact in the listening experience. My ear tells me even the warble tones can build up and from nodes.

One more thing... I don't think these radio shack meters have a fast enough response to find the real time peaks. At least PWK claimed that a VU meter could be up to 15 dB low for real quick bursts.

Later,

Mace

[WMcD]

Further to what Mace and PWK was talking about. The ancients were real geniuses. We have to rediscover it all over again. They were talking about actual headroom in the transmission system and what shows up on the analog VU meter.

It is my understanding that engineers at the time realized that bursts would be up to 18 dB or so over average on music or, perhaps, voice. And also that the moving vane meters could not indicate those quick rise times when they ocurred. The meters gave an average, only.

There was a defacto standard that 0 dB reference would be 1 milliwatt delivered by a telephone transmission line with an impedance of 600 ohms at 1000 Hz. This was the standard for moving audio around the country on copper.

We see this 0 dB reference on VU meters on tape decks. This is the highest steady tone without distortion. If you look at the scale, max is 0 dB and everything lower is, for example -3 dB, - 10 dB, etc.

However, in the best of circumstances in transmission, it was assumed that the dynamic range of the transmission line would stand dynamics 18 dB greater than average, and also assumed the VU meter would only show the average, 18 dB below.

Boy, they were SO smart. It is shame that their great accomplishments are sometimes lost on us.

Best,

Gil

[Al Klappenberger]

Gil, Mace,

There's quite a few ways to do response measurements. The easy way that most

people try to do first, as I did, was to get an SPL meter designed to measure

sound levels in industrial areas for safety concerns, then connect a simple

sine-wave audio signal generator to your amp to make a series of measurements

at a bunch of points to later use to draw a curve. In my case, it was a

Hewlett Packard 200CD generator and an HH Scott SPL meter I bought at a

"hamfest". It took me about 5 minutes to discover that I simply couldn't do

it! Moving the signal generator just the slightest bit would make the meter

jump wildly as much as plus or minus 20 dB! I wound up selling the SPL meter

and gave up.

The next step, for me, was building a log-amp similar to what PWK called the

Logerator. The one I built came out of a text book for op-amps. I combined

that with an active precision rectifier/detector. It has a nice 50 dB range.

This along with a good calibrated mike, an HP AC voltmeter to use as an amp

and an X-Y plotter I got from eBay. This is similar to what PWK was using

years ago. I also got a used Wavetek function generator that would do a sine-

wave log sweep from 20 Hz to 20 Khz. This was still sine-wave CW though. I

connected all this up and ran some plots. Again I got a curve that looked so

rough that it was nearly useless. It was obvious that something had to be

done to smooth out the curve.

My next step was to built a little circuit board that would "warble" the CW

tone generated by the function generator up and down in frequency slowly over

a range of 1/3 octave so area as it swept from 20 to 20 KHz. NOW it made the

kind of curve I expected to see! I used that scheme up until just a few weeks

ago. 1/3 octave warble tone generators are available.

Lastly, I bought a used "dynamic analyzer" (Spectral Dynamics SD375) that was

intended for analyzing transient noise from aircraft and such. This beast

weighs 55 Lbs and is a dual channel "fast fourier transform" (FFT) spectrum

analyzer combined with all sorts of other bells and whistles including

transient capture and the ability to actually edit the captured data. I also

bought a white noise signal generator (eBay, of course). This scheme allows

the entire spectrum from 20-20 KHz to be curved on the screen in real time.

White noise contains all frequencies at an equal level. This scheme uses

white noise, not pink noise. The use of pink noise requires a instrument

capable of computing the amount of power within each 1/3 octave band. It has

a response that rolls off at a rate of 3 dB / octave. This is much easier on

your tweeter than white noise and approximates how the brain perceives music.

Your brain tends to average what it hears and the 1/3 octave bands has been

found to simulate this.

All of these methods have a drawback though. The response of the room is

included with the response of the speaker. This is fine if you simply want to

equalize the entire response with an equalizer though.

There are several methods (that I know about) for measuring anechoic

response, that is, the response of the speaker alone. The obvious one is to

build yourself a multi-million dollar anechoic chamber! That was a bit beyond

my budget limits though! There are two other ways to do it too.

1 - Assuming that every sound travels from the speaker to the wall and back,

you can filter the mike frequency such that it tracks the sweep frequency of

the signal generator, but at a frequency that is, in time, delayed by the

propagation time from the speaker to the mike. This way, the mike will be

tuned to a frequency well above the reflection that arrives at the mike from

the wall. Reflections are then two low in frequency to get through the mike

and mess up the measurement. This method takes a lot of fancy equipment and

has a low frequency limit. I have never seen this kind of equipment available

and wouldn't know where to get it.

2 - Another method uses a single spike or pulse from the speaker. A single

spike, if it's narrow enough, is made up of a full spectrum of energy. If you

have an instrument that will capture this spike as it comes to the test mike

and then shut off, it can be used to actually compute the response of the

speaker using fourier transform math. This way, the instrument captures the

spike and shuts off before the reflected energy can get back from the wall.

The SD375 analyzer I have will do this and I can testify that it works! I

have attached a screen photo of this type of test. The upper trace is the

impulse picked up by the mike. The bottom is the "computed" frequency

response. This was one of my first attempts at do this. I was confused by the

double spike in time and "zeroed it out" using an edit feature of the

instrument thinking it was the reflection. To my surprise, the frequency

response dropped between 500 and 6000 Hz. Then I realized the second spike in

time was from the squawker, not a reflection, which is delay in time because

it is mounted on the back of the K500 horn. The woofer and tweeter are

mounted on the front of the speaker, in the same plane (this is the "Heresy

on Steroids" speaker in my den) with the MOUTH of the squawker horn and

arrived at the mike first. I also learned later that the rise in response a

low frequency shown on the bottom trace was because I didn't have the "spike"

loud enough. It is actually ambient noise picked up from the house. That's

stuff like traffic noise from the road outside and the house heat-pump

running etc.

As you can see from all this, a little SPL meter simply can not measure the

response of a room or speaker without the use of other equipment. You CAN use

one along with a test CD or signal generator that provides 1/3 octave pink

noise bands or 1/3 octave warbel tones to measure response. To try to do it

with only a CW signal generator is pointless. Of course, using an SPL meter

to set equal levels from the channels of a home theater using full spectrum

pink noise is fine too. Just don't expect to be able to do useful point-to-

point response plots!

Al K.

[Frzninvt]

Damn I feel spoiled now, my DBX 14/10 can flatten a rooms response in 15-45 seconds automatically. I place the calibrated omni-directional microphone at ear level in my seating position, press "auto-eq", adjust the pink noise level volume slider on the unit and bam in under 60 seconds, done! This can be repeated for 10 different seating positions and stored for instant recall in memory. In addition you can average all the memory banks, or choose which ones you would like it to average. In addition it measures SPL and line voltage in real time. It has a HFR (High Frequency Rolloff) button in the event the room is bright. This unit takes all the guesswork out of it, as it is computer controlled so measurements are extremely precise. Very rare and not very cost effective for alot of folks but if you come across one cheap - snag it!

[formica]

Al K, have you ever experimented with the software based RTA's such as ETF5 ? I'd be curious on your or anyone else's opinion / experience with them... though my intentions are more for room optimization and not speaker response.

Frzninvt, is the DBX 40/10 similar to (perhaps a higher end version) of a Behringer Ultra-Curve Pro DSP8024 ? Can it be used as an RTA without the EQ function?

thanks...

Rob

[Al Klappenberger]

Frzninvt

Yes.. It's quite logical to combine a 1/3 octave RTA with a 1/3 octave equalizer. It could be automated like that very nicely! I do suggest that you only equalize with the mike in front of each speaker and average them. I have tried to equalize at the listening position several times and found that it does not work right. The room upsets the settings to the point that the stereo image goes totally to h___! I do it by putting the mike in front of my center speaker (about 3 feet) and adjust both left and right for a flat response there. The center speaker is Left+Right through a resistor netwrok like PWK suggested in the Dope from Hope series. I then set both channels of the equalizer to the average of the two and forget it. All manually, of course, which takes a lot more than a few seconds!

Al K.

[Al Klappenberger]

Rob,

Yes, I did download a software based RTA program once. This was before I had a laptop computer and calibrated mike. To make it practical you need both of these. If you have both, it is a very good deal. I may try it again someday. I have just been using the new SD375 analyzer instead and haven't gotten back to that idea. Also, the laptop I have belongs to my wife and I would need to do some beging!

Al K.

[Frzninvt]

Here is a photo of the unit. It is an older DBX consumer group product designed and built before their demise in 1990. It is a 2/3 Octave Equalizer with Constant "Q". The second you plug the calibrated mic into the front or rear mounted port the equalizer display switches over to SPL/RTA and monitors the room in real-time. Unplug the mic and it measures line voltage and provides a real-time response on the RTA. It is extremely sensitive and adjustments can be made in 1db increments. I am reasonably certain the EQ portion of it can be bypassed but would have to check. Unfortunately, it is tough to find these things and they usually do not go cheap. It sold in the neighborhood of $1300 new from '86-'89 and is far superior to its predecessors the 10/20 and 20/20. I have it hooked to my Program Route Selector so that it can be easily bypassed when not called for. I love the thing, I got really lucky and got a great deal on it.

[artto]

I have the analogue version but thinking of getting the newer digital one. Don't forget the old analogue meters have a 'ballistic' charateristic. And there is no 'peak-hold' fuction on the RS SLM. So you never really know how high the peaks are. I've heard estimates that the real peak value can be as high as 12dB above peak value readings from the meter.

[formica]

I was hoping on making due with long wires as my girl friend sold her laptop ... and my PC is in an adjoining room.

I've done the "sine wave and RS meter" thing and it is quite time consuming if you'd like to experiment... so I'm really looking for an easier and more precise way of experimenting with room placements and treatments.

I'll test a demo version when I get around to making some wires... I seem to remember someone using the RS meter's mic (as it's corrections are known) as a start, and if I like the software I look into getting a calibrated mic and preamp.

Later...

Rob

[artto]

----------------

On 10/9/2003 11:53:06 AM Frzninvt wrote:

Damn I feel spoiled now, my DBX 14/10 can flatten a rooms response in 15-45 seconds automatically. ----------------

That's a falacy my friend. In a small room (the size we domesticate in), what ever an equalizer does for one specific location, it makes worse everywhere else. The only way to truely control this, especially to obtain a larger listening area, is with acoustics.

[artto]

IF, you have some software like Sound Forge, AND a DAT machine, a mixer/preamp, and a calibrated mic, you can use the free downloadable NCH software to create your own test CD (pink noise, 1/3 octaves, sines, whatever)(or your test recording of choice), record it to the DAT, download the DAT tape to your computer, & then use the real time analyzer in Sound Forge to see what the response looks like. I believe Formica has the link address for NCH. Its on the Forum somewhere. If you can't find it & need a copy of the software, email me.

[artto]

----------------

On 10/9/2003 6:27:54 PM formica wrote:

I've done the "sine wave and RS meter" thing and it is quite time consuming if you'd like to experiment... so I'm really looking for an easier and more precise way of experimenting with room placements and treatments.

I seem to remember someone using the RS meter's mic (as it's corrections are known) as a start, and if I like the software I look into getting a calibrated mic and preamp.

----------------

Rob, trying to measure sine waves in a reverberant room the size most of us have to listen in is basically futile. There are just too many reflections, and too many standing wave problems. Moving your head a few inches can change the results significantly. Thats why random broadband noise is most often used to take measurements from. And it resembles music much more than pure sine waves.

The RS meter's 'corrections' are not known. What is known is the variation that can be expected from unit to unit, which is quite substantial. That is NOT a calibrated mic in the RS SLM. Only when the individual mic's response is 'calibrated' (measured & recorded graphically & nuermerically) do you know actually know how much to make adjustments in the recoreded room response.

[cluless]

love_hertz...

ask a simple question...see what you get...LOL

I don't know how many people you can catch that can give you an honest answer to the reliability of the RS analog vs RS digital because most of us own one or the other but not both. But if you think about what you are trying to do, the answer becomes a little easier. What you want is a consistent reading on the output levels of your speakers with a minimum of boosting or damping. What you are looking for is consistency. Anyhow, bother AVMAN, make him earn that sale...I'm sure that he'll give you a hand in setting up your system.

[formica]

----------------

On 10/9/2003 7:20:18 PM artto wrote:

Rob, trying to measure sine waves in a reverberant room the size most of us have to listen in is basically futile. There are just too many reflections, and too many standing wave problems. Moving your head a few inches can change the results significantly. Thats why random broadband noise is most often used to take measurements from. And it resembles music much more than pure sine waves.

The RS meter's 'corrections' are not known. What is known is the variation that can be expected from unit to unit, which is quite substantial. That is NOT a calibrated mic in the RS SLM. Only when the individual mic's response is 'calibrated' (measured & recorded graphically & nuermerically) do you know actually know how much to make adjustments in the recoreded room response.

----------------

I realise how variable the results can be in using sine waves... lol... it wasn't long before i ended up with a dozen or so charts when I experimented with meter's position and only two variables. Pink/white noise testing is a reason a RTA is so appealing to me... I don't have a DAT, calibrated mic, or soundforge... and I'm guessing just soundforge would cost significantly more than something like ETF.

You're right that I shouldn't have used the term 'corrections' in referring to the RS meter's mic... as that would imply that the 'actual' mic was tested... when in reality it's a ballpark of how far off they are. No worse, I figure, than the RS/sine-waves combo...

Later...

Rob

[love_hertz]

thanks on input guys and gals (gotta be politically correct)

[artto]

The Radio Shack SLM's (Sound Level Meter) are OK, but not very accurate. There is a relatively wide variation in the response of the integrated microphone from unit to unit (in other words, the one you buy may be +2dB at 1000Hz while mine may be -2dB).

The newer digital ones are more useful, have a few more features. The older VU meter types will not really show peak levels accurately do to the inherent slowness of the meter 'ballistics'. You can expect the actual peak levels to be 10 or 12dB higher.

Either one is fine for monitoring average SPL levels, and to some extent, peak levels (the digital one is better at peak readings-no meter ballistics to deal with & it has a peak-hold indicator) . They might even be used to 'get you in the ball park' for equalizing a sound system but I wouldn't recommend them for that. For fine tuning a room or accurately EQing a sound system you still need to use a calibrated mic with a known response curve in conjunction with a real-time spectrum analyzer.

AND NEVER use sine waves to tune the room acoustics. As Al K quickly found out above, all you have to do is move something, anything, in the room a few inches, and the results will be very different! White noise & pink noise are better 'music simulators'.

[roy]

i own both analog and digital radioshack meters...they both work great but digital is more accurate and faster.....also can do a lil more with it and its easy to read too..i did comparisons side by side digital wins slightly...both very useful!