AVS Forum banner
  • Our native mobile app has a new name: Fora Communities. Learn more.

TrueRTA for optimal sub placement?

2253 Views 45 Replies 6 Participants Last post by  Ilkka
I was hoping that some of you could share your experience...


I'll start by saying that I am not an audio expert or a perfectionist, but I would like to find the best placement for my sub in my media room that will yield the best output with the flattest response. For the time being I am not interested in using a BFD as I am just getting started with all of this audio theory, and would like to make due with what I currently have.


That being said, I was thinking of picking up TrueRTA and use my RS SPL meter as the mic in order to find the best possible placement for the sub in my room. I am aware the the Behringer mic is more accurate, but as I see it, I am only looking for a 'delta' based on the first measurement I take. Any error in the mic will be in all measurements and therefore relative. Is it correct to make that assumption? Once I get more familiar with process I will probably pick up the Behringer mic.


Assuming what I said is correct, what is the best procedure for testing?


Should I be using a test cd with frequency tones from 10hz to 300hz in 1hz increments? If so, how would I setup TrueRTA to gather the data?


Or would it be best to use the pink noise generator on the receiver, setting TrueRTA to 1/24 Octave, and using 50 Sample Averages?


I've done much searching and reading over the past couple of weeks, but I am starting to suffer from info overload. So I am looking for a starting point to get me into the right direction.


Also, could I use TrueRTA (with a calibrated mic this time) to measure the frequency response of the other 5 channels to see how they measure up against each other? I have the Pio 1014 and used MCACC to setup the speakers, but I would like to use TrueRTA to check the results and fine tune if possible.


Any info you guys can provide would be greatly appreciated.


-patrick


I should mention that last week I used the test cd i spoke of (10hz-300hz, 1hz increments) and a calibrated Bruel & Kjaer to make some measurements from my primary seating position with the sub in its current corner location. I must say I was surprised at all the peaks and nulls in the curve. This was a very tedious process and I am hoping that TrueRTA can accomplish the same thing with a little less effort.


I came across this thread:
http://www.soundandvisionmag.com/id...=37152&pageNo=2


It was said by "IRRITATED - Guy (Crazy Old Fart)":

"Using specific frequency measurement (sine wave) will result in huge nulls and peaks. This is typical. You need to average the response when you do that sort of measurement. I use either 1/6 or 1/3 octave averaging.


The most accurate way to use TrueRTA to measure the bass response in the room is to use pink noise and set TrueRTA to use 50 or more samples, which averages 50 sampes together to get an accurate measurement. "


Is this good avice or was my original test method with the sine waves and B&K more accurate?
See less See more
Status
Not open for further replies.
1 - 20 of 46 Posts
TrueRTA can measure the FR quickly and easily and is a nice alternative to manually plotting sines.


The loop is basically mic, sound card input, TrueRTA, sound card output, audio device (sub, pre/pro, etc.).


TrueRTA generates the signal, the audio device plays it, the mic reads it, and TrueRTA displays the results.


If you are using the RS meter and want to know the actual unweighted FR, build a file (importable into TrueRTA) that corrects for the c-weighted filter in the RS meter.


The digitally synthesized log sine sweep (Quick Sweep) works very well and is fast. The pink noise will provide very similar results. 50 data point averaging is fine for the pink noise, or you can use more. Quick Sweep is certainly faster and more elegant than pink noise, but I've used both with good results.


You can vary the resolution of either the Quick Sweep or the pink noise from 1/24, 1/12, 1/6, 1/3, 1/1 octave increments.


An MLS based program like ETF-5 can help you identify room modes and room ringing as a function of time, and is a nice companion program to TrueRTA.


IlkkaR is getting into measurments and likes to chat about this stuff and is becoming an experienced TrueRTA user - seek him out for additional advice and he will also provide you with the mic correction file and some other user tips.
See less See more
Quote:
Originally posted by need4speed

Assuming what I said is correct, what is the best procedure for testing?


Should I be using a test cd with frequency tones from 10hz to 300hz in 1hz increments? If so, how would I setup TrueRTA to gather the data?


Or would it be best to use the pink noise generator on the receiver, setting TrueRTA to 1/24 Octave, and using 50 Sample Averages?
Definitely use TrueRTA's pink noise generator or Quick Sweep function. Quick Sweep is much faster, but if you want to get accurate reading below 100Hz with RS meter use max 22kHz sampling frequency. The meter is not sensitive enough and gives variable results (below 100Hz) with 44.1kHz or 48kHz sampling freq. Setting sf lower will result sweep taking longer time to sweep say 10-100Hz.


If you want to use pink noise, set averaging to 50-100. Play noise atleast 5-10 seconds.


Both ways give same results if performed correctly.

Quote:
I've done much searching and reading over the past couple of weeks, but I am starting to suffer from info overload. So I am looking for a starting point to get me into the right direction.


Also, could I use TrueRTA (with a calibrated mic this time) to measure the frequency response of the other 5 channels to see how they measure up against each other? I have the Pio 1014 and used MCACC to setup the speakers, but I would like to use TrueRTA to check the results and fine tune if possible.
Of course you can. Actually you can do that with RS meter also. Of course it's not as accurate as Behringer, but up till 10kHz it's quite accurate. Of course you have to use a correction file.

Here is for example one of my friend's FR measured with RS. 1/6 octave smoothing was applied.

http://personal.inet.fi/private/ilkk...sd2_system.png

Quote:
I should mention that last week I used the test cd i spoke of (10hz-300hz, 1hz increments) and a calibrated Bruel & Kjaer to make some measurements from my primary seating position with the sub in its current corner location. I must say I was surprised at all the peaks and nulls in the curve. This was a very tedious process and I am hoping that TrueRTA can accomplish the same thing with a little less effort.


I came across this thread:
http://www.soundandvisionmag.com/id...=37152&pageNo=2


It was said by "IRRITATED - Guy (Crazy Old Fart)":

"Using specific frequency measurement (sine wave) will result in huge nulls and peaks. This is typical. You need to average the response when you do that sort of measurement. I use either 1/6 or 1/3 octave averaging.


The most accurate way to use TrueRTA to measure the bass response in the room is to use pink noise and set TrueRTA to use 50 or more samples, which averages 50 sampes together to get an accurate measurement. "


Is this good avice or was my original test method with the sine waves and B&K more accurate?
IG's advice is good. Of course your *real* FR is not smooth at all above ~100Hz. Below it depends how good your room is, how many standing waves it creates etc. But it's recommended that you apply some smoothing and averaging to it.


Here is a graph from that same friend, first no smoothing applied. Measured with pink noise using 50 averaging.
http://personal.inet.fi/private/ilkka/svs/pb12_isd2.png


The second pic is same FR using 1/6 octave smoothing.
http://personal.inet.fi/private/ilkk...sd2_smooth.png


I'm happy to help you regarding TrueRTA. Just ask! :)


edit: Ed, thanks. ;)
See less See more
Ilkka..would you mind exchaning email addys via private message?
Quote:
Originally posted by need4speed
Ilkka..would you mind exchaning email addys via private message?
You've got PM. :)
"50 data point averaging is fine for the pink noise, "


What's being averaged over what?


"Ilkka..would you mind exchaning email addys via private message?"


How about keeping it public for the benefit of others? I'm going to get TrueRTA as well and could use the education.


Thanks
See less See more
>How about keeping it public for the benefit of others? I'm going to get TrueRTA as well and could use the education.
>"50 data point averaging is fine for the pink noise, "

What's being averaged over what?
Quote:
Originally posted by need4speed
>How about keeping it public for the benefit of others? I'm going to get TrueRTA as well and could use the education.
as promised, here is the corrospondance via email:

> 1. Since my PC running TrueRTA is in another room, I think I will use

> the pink noise tone from my Pio 1014. I dont imagine this will be a

> problem?


My computer is also in another room. You just need long rca cables. Using pink

noise tone from your receiver is a problem. Usually (almost always) those tones

are band limited around 400Hz-2000Hz. They are used to calibrate channel

levels, not measure FR. AVIA has same tones. You need atleast 20Hz-20kHz pink

noise if you want to measure your speakers FR. You can get it from test disc

(AVIA etc.) or from TrueRTA's generator.


> 2. Lets say I did want to use TrueRTA as the generator as well...since

> the output is stereo, and the receiver is 5.1, how do I send that signal

> only to the LFE channel? I assume I will have to turn off all the other

> speakers. But since the truerta signal is not in the LFE channel, will

> it actuall play on the sub...this probably shows my lack of

> understanding how the LFE actually works :)


Yes, the receiver is 5.1 but you are sending stereo signal in, same as you were

listening normal cd's. You can't send signal only into LFE, because then the

signal would had to be digitally encoded and it's not, because it's just normal

analog stereo signal. So you have to set your receiver to normal stereo mode,

not PL2 or DD etc.


If your speakers are set to small (as they should), I assume that x-over is set

around 80Hz. The only speakers playing the noise are your mains and subwoofer.

If you want to measure for example your right surround, you have to hard wire

it so that it is the only speaker connected to your receiver at the time. If

you want to measure only your mains then just turn your sub off. If you want to

measure only your sub, then disconnect the minus wire from your speakers. And

so on.


LFE-track is an additional bass track. So it has to be in the disc before you

can even play it. You can't create it by yourself. Most DD and DTS encoded

movies have it. Often receivers let you choose where to direct it. Mains, sub,

maybe both. Often it's better to direct it only to sub. Then your mains don't

distress and the sub can those loud low frequencies better. Analog signal can't

have LFE-track.


> 3. I've gotten mixed reviews in regards to averaging the readings. Some

> say that it hides nulls and peaks that should be there. They are big

> proponents of using sine wave tones to test the response. Others have

> recommended averaging. So its a toss up. Let's assume I wanted to use

> discrete sine waves tones to compare agains the pink noise or quick

> sweep, how would i go about setting that up?

>

> Now as far as the RS meter is concerend...I found the following at the

> ETF site, which says the exact opposite of everyone else...ie they say

> the meter is very flat below 500hz and requires no corrections there,

> here is the quote:


"(1) Microphone:


If only low frequencies are to be measured, we recommend the Radio Shack

SPL meter. This unit is very flat - much more so than you will be able

to fix your room. It is therefore quite accurate for its intended

purpose. We supply a low frequency calibration file for this unit on our

support page on the web site under "Support". In this case, get a camera

stand from a photography store that will allow the mic to be mounted.

This stand may be expensive and the option below may be more desireable

in terms of total cost and performance. This microphone should be used

at levels below 95 dB for distortion measurements if the distortion

performance upgrade is used. Distortion measurements should be made with

a good high SPL musical instrument microphone when higher SPL's are to

be used.


Radio Shack SPL meter


ETF supplies a high performance calibrated electret condenser

microphonefor use in sensitive applications. The RS analogue or digital

SPL meter may be used for ETF as well. This will not supply the same

accuracy as the unit sold here but is recommended for applications where

critical adjustment of frequency response is already done or not necessary.


This unit has weighting curve adjustments that confuse a lot of users.

This adjustment does not control the output, it only damps the meter

movement.


Download the calibration file below and open it as a microphone

calibration file when using ETF with the RS unit. Please note that this

is not a true "calibration file" because it is not calibrated for your

unit. It will remove the gross response errors associated with this

device. Typical differences between this unit and any random unit are

within ± 2 dB below 8 KHz, typically. This unit is virtually useless

for accurate high frequency measurements.


Other calibration files exist on the internet for this unit, these

calibration files apply when reading the meter movement, not when taking

data from its RCA output jack. These units are very constant gain below

500 Hz (flat frequency response) and do not require further low

frequency calibration"



I have to say that they are SO WRONG! Weighting does apply also to output jack.

I can easily prove that.

http://personal.inet.fi/private/ilkka/svs/weighting.png


Look at that picture. It shows same system measured with weighting knob set to

c and a. If weighting would NOT affect output jack, those lines would be

identical. Are they? NO.


Also check this page:
http://www.gti.net/wallin/audio/rsme...0/33-2050.html


This guy knows everything about this meter and he agrees with me. Scroll down

and there is a question about this problem.


Another proof is that I have calibrated my TrueRTA's SPL reading with 500Hz

tone, where meter does not have any error (or very small). As you know TrueRTA

can show SPL reading (total sound power). If weighting would NOT effect to

output jack, that reading and meter needle would have to show different values

below 100Hz (where meter has errors). I played 20Hz and 30Hz tones at 60dB and

70dB SPL's and checked the reading in TrueRTA and then in meter itself. They

were almost IDENTICAL (maximum ±1dB). Those general correction factors show

that at 20Hz meter has 7.5dB error. So if weighting would not effect to output

jack, there would be that error or even close to that between those readings.


I don't believe that those general correction factors are accurate. I have made

my own correction file using that ETF file above 500Hz and general c-weighting

corrections below it. Those general CF's show that meter is almost flat to

20kHz. Everyone who knows something about microphones knows that it's not true.

ETF people know it too. Thats why there is almost 13dB correction at 20kHz in

their file. If the person who measured those general CF's did that big error at

higher end, how can we be sure that lower end is correct? I don't believe it

is. I believe that general c-weighting corrections are much more accurate.

Actually the difference is not so big. :) At 20Hz only about 1.2dB. Below it a

little bit more. But above 1kHz that ETF file is much more accurate.


So use that correction file that I linked in that thread.


If you have any further question about TrueRTA or measurements don't hesitate

to ask!


-Ilkka
See less See more
Quote:
Originally posted by need4speed
>How about keeping it public for the benefit of others? I'm going to get TrueRTA as well and could use the education.
second part of the email:


I realised that I didn't answer to your third question, sorry about that. Here

it comes.


> 3. I've gotten mixed reviews in regards to averaging the readings. Some

> say that it hides nulls and peaks that should be there. They are big

> proponents of using sine wave tones to test the response. Others have

> recommended averaging. So its a toss up. Let's assume I wanted to use

> discrete sine waves tones to compare agains the pink noise or quick

> sweep, how would i go about setting that up?


Averaging does NOT hide peaks and nulls. Smoothing does. You can not get a

reliable reading with pink noise without using atleast 10 averaging. I prefer

averaging around 50-100. So do many others. Smoothing does hide peaks and nulls

if you smooth too much. 1/6 or 1/12 smoothing is alright. It can hide a small

peak/null but not a big one. Just look at those images I posted to AVS.


I only smooth FR when I post it to Internet. It looks a little bit better that

way. ;)


If you want to use sine waves you'd have to set TruRTA into measuring mode

(press Go button), set the peak hold on. Be very very quiet and using your cd-

player etc. play those tracks one by one. Play atleast 4-5 seconds per track.

If you are using 1Hz increments then use 1/24 oct. mode, but if 2Hz or more,

use 1/12 oct. mode. If you are measuring too accurately and only playing sines

with 5Hz increments, you get a very bumby FR. :)


Happy measuring!


-Ilkka
See less See more
Ilkka...your comments have been very helpful. I will post back my results as soon as I get them.
Quote:
If you want to use sine waves you'd have to set TruRTA into measuring mode (press Go button), set the peak hold on. Be very very quiet and using your cd-player etc. play those tracks one by one. Play atleast 4-5 seconds per track. If you are using 1Hz increments then use 1/24 oct. mode, but if 2Hz or more, use 1/12 oct. mode. If you are measuring too accurately and only playing sines ith 5Hz increments, you get a very bumby FR.
There will be an error introduced by TrueRTA using this method. The Peak Hold feature only holds the highest amplitude FFT bin value in the bin cluster represented by the fundamental. It will not hold the actual sound pressure.


At the lowest frequencies, the fundamental is represented by several bins, none of which represent the total acoustic output. This phenomenon can be minimized or exacerbated by adjusting the input sampling frequency, but it nevertheless exists, even at the 8kHz sampling frequency.


There is a work-around for this, but it is beyond the scope of new TrueRTA users. Generally you are much better off using TrueRTA as the signal generator and the signal reader (i.e., closed loop mode).
See less See more
Thanks Ed, thats a little beyond my audio knowledge, but I will certainly take your word for it. If I have the chance I will run that test knowing that it will have errors, and compare it to the data that I collected manually with the B&K.


Also, should I be using 8kHz sampling for all of my sub measurements?
Quote:
Also, should I be using 8kHz sampling for all of my sub measurements?
It's not really needed - the 8 kHz SF has a deathly slow ramp up time. It provides better mic pick-up and data aquisition below 15 Hz on the Quick Sweep if the line input gain is low and the signal from the subwoofer is very weak.


For most applications the 48 kHz mode works very well with Quick Sweep, even down to below 20 Hz, provided the line gain is adequate, and the sub output at the very lowest frequencies is healthy enough to overcome ambient noise levels.
Quote:
Originally posted by Ed Mullen
There will be an error introduced by TrueRTA using this method. The Peak Hold feature only holds the highest amplitude FFT bin value in the bin cluster represented by the fundamental. It will not hold the actual sound pressure.


At the lowest frequencies, the fundamental is represented by several bins, none of which represent the total acoustic output. This phenomenon can be minimized or exacerbated by adjusting the input sampling frequency, but it nevertheless exists, even at the 8kHz sampling frequency.
Thank you for this clarification. I've never done a measurement with this method and therefor I cannot recommend it. I believe that it can give you close results (using low sf), but not as accurate as QS or pink noise.

Quote:


There is a work-around for this, but it is beyond the scope of new TrueRTA users. Generally you are much better off using TrueRTA as the signal generator and the signal reader (i.e., closed loop mode).
What about experienced TrueRTA users? ;) Not that I need it.
See less See more
Quote:
Originally posted by need4speed
Thanks Ed, thats a little beyond my audio knowledge, but I will certainly take your word for it. If I have the chance I will run that test knowing that it will have errors, and compare it to the data that I collected manually with the B&K.


Also, should I be using 8kHz sampling for all of my sub measurements?
I use 22.05kHz sf for subwoofers. 48kHz for full range speakers. Best combination between speed and accuracy.
thnx...picture is worth a thousand words.


should the RS meter be plugged into the mic or line-in?
1 - 20 of 46 Posts
Status
Not open for further replies.
Top