I have tried measuring my room using a Radio Shack meter, graph paper and DVE "frequency-response" test. Only recently did I find out that the Radio Shack analogue meter is fairly inaccurate for measuring the full range of frequency response. (I did however know low bass was off by 2-3db) Case-in-point, whichever speakers I owned (no matter how bright or warm) seemed to make the meter jump 8-10 db's in the 10KHz area, almost always - but some more than others. Even after treating walls, and even after running Audyssey I noticed this.

Back to my question. When I see these graphs with everyone's frequency responses for their rooms, is this from the test of all speakers at once? For example, on DVE you run frequency responses for each of just the front 3 speakers, and then it does all of speakers at once.

In checking the frequency response of just the left speaker I see major peaks in treble and bass. However when I run all speakers together everything looks more flat and the peaks are not as high, and overall, it looks reasonably good. I'm guessing this is the average, and if I'm seeing some reasonably flat reponse for the average, is that more important than to be concerned with 1 or 2 of the speakers having more significant peaks?

Regardless, I'm still deducing that the Radio Shack meter must be inaccurate at both bass fequencies and high-treble. Ethan W's tests confirmed this a little.

Only recently did I find out that the Radio Shack analogue meter is fairly inaccurate for measuring the full range of frequency response.It is actually pretty damn accurate for what it is over the range that it can measure. Very reproducible.


(I did however know low bass was off by 2-3db)The amount that it is "off" at lower frequencies varies as the frequency becomes lower. It is off 7.5dB @ 20Hz. But only 2.5dB @ 40Hz.


Case-in-point, whichever speakers I owned (no matter how bright or warm) seemed to make the meter jump 8-10 db's in the 10KHz area, almost always - but some more than others. Even after treating walls, and even after running Audyssey I noticed this.Are you using pure tones when measuring this or frequency sweeps? Why are you measuring that high, anyway? Even if you could measure it very accurately there are going to be huge fluctuations at higher frequencies when you measure a speaker in-room. The published frequency response data you see for speakers is measured anechoically.


Back to my question. When I see these graphs with everyone's frequency responses for their rooms, is this from the test of all speakers at once? For example, on DVE you run frequency responses for each of just the front 3 speakers, and then it does all of speakers at once.In most cases it is either the 2 front speakers alone, the subwoofer alone, or the 2 front speakers plus the subwoofer. Most people are interested in EQing the combined FR of their 2 front speakers and subwoofer. With the usual software that is utilized, there is not a good way to send a sweep to all of the speakers + the sub. And, that is probably not the best way to do it, anyway.


In checking the frequency response of just the left speaker I see major peaks in treble and bass. However when I run all speakers together everything looks more flat and the peaks are not as high, and overall, it looks reasonably good.Yes, to a degree, that makes sense.


I'm guessing this is the average, and if I'm seeing some reasonably flat reponse for the average, is that more important than to be concerned with 1 or 2 of the speakers having more significant peaks?Thinking of it as an "average" is not really the way to think about it. It is the combined FR of all the speakers and the sub interacting with one another and the room.

The RS meters can't really be trusted over a few kHz... especially depending on which weighting you have it set to... but if you already have one, you can download REW from HTS and take some real sweeps and see exactly what you're dealing with, as REW allows you to load a correction file that will improve the accuracy...

Could be that your wall treatments aren't appropriate, could be that something about the room tricked Audessey, could be a number of things... but if you're concerned about 10kHz and above, I would suggest looking at the REW documentation and upgrading to a better mic to use with it... I use a mic/preamp combo that is very popular with REW and recommended in their docs at is about $125 to put together... (disclosure: there is just a little dissent on the suitability of even this mic above 10-15kHz, but it seems to work great for me...)

Are you using pure tones when measuring this or frequency sweeps? Why are you measuring that high, anyway? Even if you could measure it very accurately there are going to be huge fluctuations at higher frequencies when you measure a speaker in-room. The published frequency response data you see for speakers is measured anechoically.

Thanks sivadselim!

After thinking about what you've said, I think I'm starting to get this after haring all of these suggestions. Echoes and a speaker's natural response need to be addressed separately. If I see a peak of 10 decibels in the treble frequency in my room, setting an equalizer to lower it isn't going to solve the problem accurately. It might bring the level down, but it's still not creating the right imaging at the listening plane because the speaker is interacting with a wall still. Equalizers are meant to adjust the natural response of the speakers, and the only way to measure this would be in an absorption chamber etc.

Once the speakers are ideally flat in this environment, you then address the room. You turn down those peaks by 'turning down the room' through treatment and/or speaker placement. So assuming all speakers are naturally flat-response, turning down room peaks through equalizing might lower the peaks, but you lose the accurate imaging and interaction with other speakers. If you "turn down the room," you still have the correct pitch and imaging from the speaker at the listening plane, but the excess volume created by the ceiling or wall is now gone.

I think I might understanding this more. Is this all correct?

Having said all that, how do I first address the speaker issue in making sure I've got a flat response? I assume it would be to measure it in a acoustically-dead room. But in addition to that, if I were to even still measure with a Radio Shack meter, which test would I use?

Are you using pure tones when measuring this or frequency sweeps? Why are you measuring that high, anyway? Even if you could measure it very accurately there are going to be huge fluctuations at higher frequencies when you measure a speaker in-room. The published frequency response data you see for speakers is measured anechoically.


I measure my room using the DVE frequency sweep and a Radio Shack meter. I go through the L, C, R and all 5.1 together. I watch the needle to see what it does. You said "even if I could measure it very accurately." You said that the Radio Shack meter is "pretty damn accurate for what it is over the range it can measure," so does this mean that treble and bass are going to be a bit inaccurate EVEN in an acoustically dead room? As I said, when I measure my room (I'm using some treatments, and speakers calibrated with Multi-EQ) I see flat except for bass and treble - in particular a significant peak at @ 10KHz. This particular peak I see when running all 5.1 speakers in frequency-sweep, the bass peaks curiously are reduced much more when doing all 5.1 at once. Concerning the treble-peak, I have to think this is a combination of room, but partly the inaccuracies of the Radio Shack meter.

Aside from this, is using DVE Freq-sweep with a Radio Shack meter not a useful to measure my room? Since the RS meter is reasonably accurate, should I use better tones other than a frequency-sweep to compliment this.

Thanks again sivadselim! Hopefully i got the earlier hypothesizing right.

Could be that your wall treatments aren't appropriate, could be that something about the room tricked Audessey, could be a number of things...

I think it's the speakers and the inaccuracy of the RS meter combined. I have observed this peak in multiple HT rooms. I even have swapped out receivers and speakers and it still does it. I briefly owned the Onix x-ls speakers and in doing the same test, I still got that strange spike at about 10KHz. Again, this is JUST from the frequency sweep on DVE using Rat Shack SPL. Using my sub-par Sony tower speakers, I get it as well. Audyssey does do some correcting of speakers, but it doesn't do 100% in all cases. The curious part was the Onix x series are naturally flat with extended frequency response, so there is no naturally peaks for them. In that case, that irritating peak must have been the room and possibly my method of testing causing it.

I think it's the speakers and the inaccuracy of the RS meter combined. I have observed this peak in multiple HT rooms. I even have swapped out receivers and speakers and it still does it. I briefly owned the Onix x-ls speakers and in doing the same test, I still got that strange spike at about 10KHz. Again, this is JUST from the frequency sweep on DVE using Rat Shack SPL. Using my sub-par Sony tower speakers, I get it as well. Audyssey does do some correcting of speakers, but it doesn't do 100% in all cases. The curious part was the Onix x series are naturally flat with extended frequency response, so there is no naturally peaks for them. In that case, that irritating peak must have been the room and possibly my method of testing causing it.

It sounds like the only constant here (I think you say that you've tried multiple rooms and multiple speakers and multiple receivers) is the meter... that's the only thing it sounds like you havne't tried swapping, and you always get the same strange peak... I think it's the meter. no ifs, ands, or buts about it. Add to that, most people don't think the RS meters are reliable at that high a frequency (no dis to sivadselim here)...

Only two ways to accurately measure the speakers by themselves (without room interactions): 1: anechoically... that means, either in an anechoic chamber, or in an open field... I'm guessing neither is available to you... 2) in a room with a known response.... this is somewhat misleading though, different speakers with different radiation patterns can interact with the room differently... ok, I'll throw in a third... sort of a poor mans anechoic chamber or open field... set the speakers up in the middle of your largest room, place the measurement mic 1 meter away... what you're doing is measuring them near-field, giving a bias towards their direct sound field and diminishing the effects of reflections... you won't get an accurate response, but you can see how the peak changes wrt the rest of the band, and get an idea of whether it's coming from the speakers or the room...

Also, try measuring just one speaker, it could be the two summing at some frequency ...

but most of all, download REW, take sweeps that you can then save and post for others to look at... you can voluntarily make a donatoin to support its development, but you don't have to... that makes it free, and is an improvement over the method you're using... you can still use the RS meter you have, and this will make it more accurate over the whole band, but it still won't be really accurate at this frequency...

Check your rcvr and make sure you don't have any silly effects on...

I don't find my measurements taken with an RS meter plotted as suggested hear look anything like measurements I take with REW, ATB PC Pro, or Speaker Tester, all programs I trust much more. I can post a comparison, even with the appropriate correction values added, to show the differences if you like. It gives very rough idea's, but I wouldn't trust peaks in the upper range as being accurate. I don't even like using it to setup and match levels, since peaks in the response can trick the meter, and cause you to misbalance the speakers.

Running more and more speakers, especially in the low frequencies, excites more and more modes, and will even out the response due to the more even room interactions. This is why it looks smoother. It's also why multiple subwoofers spread throughout a room will always have a smoother bass response than one or two, especially if all placed in generally the same location.

While a lot of people might disagree with this, room treatments, especially velocity absorbers, don't do a lot to fix response problems as discussed here in a room. They will have near zero effect on the low frequency modes, and the most they can do at higher frequencies is reduce the level of the reflected waves. If they manage to reduce it to a level of inaudibility, then they will be effective at reducing these modes, echos, and flutter. However, they typically can't, and if you look carefully at measurements of the effects of velocity absorber wall panels, you see they not only don't smooth the response out much, but really just lower the level in the room, across the board. EQ's such as Audissey really can't fix a lot of these problems either. You can eq a fairly smooth response by eliminating some of the peaks, but you can't typically get rid of the troughs, which are typically caused by antiphase cancellation. Often the cancellation is so large that it would exceed the maximum eq boost, and even if you could, the amplifier would have to produce so much more power at that frequency to fight the effects of the room. Much better off fixing this with room treatments and better placement, then just using the eq to remove some small peaks here and there.

The 60 setting on the RS meter exaggerates the low end also so if one is measuring for a 65dB - 64dB target and measuring the low frequencies it is going to be more off even when using corrections.

http://farm4.static.flickr.com/3329/3213707572_6506111189_o.jpg

It is actually pretty damn accurate for what it is over the range that it can measure. Very reproducible.

The amount that it is "off" at lower frequencies varies as the frequency becomes lower. It is off 7.5dB @ 20Hz. But only 2.5dB @ 40Hz.


AFIK all RS SPL meters roll off the highs. This is clearly shown in Ethan Winer's comparison graphs (http://www.realtraps.com/art_microphones.htm). RS says this-

"C-weighting has C-curve (flat) frequency characteristics and causes the meter to respond mainly to frequencies ranging from 32 to 10,000 Hz."

It is however a consistant product within model type and the correction tables work well enough.

Audyessy will normally roll off the highs a bit.

I hav not tried this with the newer style Audyessy mics, but the older Onkyo pucks were actually fairly flat out to >15kHz. It might be worth a try.

The specs for the RS meter make it clear that it is not intended for full range measurement.It is actually pretty damn accurate for what it is over the range that it can measure. Very reproducible.I think that the claimed upper limit is 10kHz, but I am not certain.

In the cinema EQ is done in the projection booth this is to keep any disturbance like someone moving around because it will affect the graph. Last month I think I was using the SPL db couldn’t be asked to use REW at the time.

A simple 100Hz sine wave was disturbed by my movement as I walked closer the right front corner the SPL db reading changed with an increase and this in a way told me to placed a few simple empty boxes in the corner which fixed the furthest rear left sidewall surround as well as the other few lined along the sidewall.

So it should all be done outside of the room with the computer or laptop, place the microphone in the room and GET OUT!

I don’t have the luxury at the moment since the room is cluttered with audio equipment on the left side of the room and the pc is in the room as well.

But I could try and get the keyboard or mouse near to the door and run the cables underneath and then press the start button on REW.


When I worked as projectionists for Warner Bros back (1998) the sound engineer asked for my assistance to place the microphone a few feet forward and slightly to one side and then get out!

Also it took 2 days for the guy to EQ each of the 12 screens with the Dolby CP500, now then.

Also the little black box that he used I found a site that was 90% close to what I saw back around (1998) that displayed a red dotted X-Curve ISO2969 in led form and all was needed was to get the red dots peaks and dips lined up to the X-Curve.

I agree with JBL on this point. I posted this file a few times over in the REW forum and have not gotten any complaints yet. This program I wrote waits 15 seconds then presses the left click button the mouse for you while you are away.

Enjoy, and be sure not to move the mouse over a self destruct button on accident!

I agree with JBL on this point. I posted this file a few times over in the REW forum and have not gotten any complaints yet. This program I wrote waits 15 seconds then presses the left click button the mouse for you while you are away.

Enjoy, and be sure not to move the mouse over a self destruct button on accident!

That’s rather noble a remote countdown timer.

Every time I try and open a zip file, I don’t have any luck, but I’d just get the get the computer outside the room if necessary and monitor the functions outside.

Cheers for the zipfile, I’m sure someone else would find it useful.

http://i279.photobucket.com/albums/kk123/IndianaJones34/InoutEQtesting.jpg

Here’s something I suggested a few days go about room EQ.

“In the cinema EQ is done in the projection booth this is to keep any disturbance like someone moving around because it will affect the graph.

When, I worked as projectionist for Warner Bros back (1998) the sound engineer asked for my assistance to place the microphone a few feet forward and slightly to one side and then “get out”!

Also it took 2 days for the guy to EQ each of the 12 screens with the Dolby CP500, now then

Since I rather close to the door and I don’t know why I didn’t do this early, even thou I remember my days at Warner Bros.

So the title of this thread came to me out of the closet LOL. Yeah that’s right the (Kevin Kine, In & Out 1997) and no, I’m not gay.:D

So I’ve been In & Out the room like a yoyo, testing it first with me in the room and then out of the room, out of sight while REW performs its frequency sweep.

The difference looks appears so tiny when glancing at it first time, but there are some small variables.

Centre channel while being inside the living room.
http://i279.photobucket.com/albums/kk123/IndianaJones34/InoutFrequencysweepIN1centre.jpg

Centre channel while being outside of the living room.
http://i279.photobucket.com/albums/kk123/IndianaJones34/InoutFrequencysweepOUT1centre.jpg

Ran the centre though a few more times depressed the LF slightly and added a slight EQ in at 1Khz on the DCX2496. The left and right at present are miles out and I’ll work on them later.

Test was done while being outside of the room and good thing too because the high frequency is loud on the outside and I was inside the room I’ll be waving my arms around because of the pain of the high frequency, bending me ears.:(

http://www.avsforum.com/avs-vb/attachment.php?attachmentid=133758&stc=1&d=1234822832

You’ll get a good workout with “In & Out” it feels good too. :D

Well its looking okay for now I’ll run a few more though then focus on the left and right and see how close I can get them with the centre, sigh this is going to take all-night and into the early morning hours.

Centre channel JBL control 5

http://www.avsforum.com/avs-vb/attachment.php?attachmentid=133759&stc=1&d=1234824703