Register Latest Topics Chat
 
 
 


Reply
  Author   Comment  
watson

Registered:
Posts: 8
Reply with quote  #1 
Hi

Let's say that I read section F - Basics of Sound from TEF manual (in practice I know a little more - i've studied vibroacoustic University of Science and Technology in Cracow, Poland) and my boss said: please, take this brand new measurement system TEF25 and your job is to learn how to use it and start to make some measurements. During my studies I was using SVAN meters so this is my first contact with TDS measure system and I feel little dizzy about it. I read manual but sometimes it is difficult for me to understand english branch slang. Also I have DVD TEF Class level 1 but it is easier to read than listen especially when sometimes sweeps are very irritating (sometimes they mask "teachers" speech). So can I start with questions ?

We (me and my work mate) are trying to follow the DVD class measures and maybe everything was fine until the measures with low and high pass filters. We just don't understand what happens with etc figure, why the direct sound is so late? - see the attachment (low pass filter at 1 kHz - yellow - 24db/oct; violet - 48db/oct).

Hm... more questions we will ask later ;]



__________________
pozdro600
Doug

Moderator
Registered:
Posts: 71
Reply with quote  #2 
HI, Thanks for your post.  IM not sure how much training we can do via e-mail, but lets give it a try!   First thing I would like to know, is what are you trying to measure? What sort of things do you want to do?  For example, are you setting up sound systems?  Are you designing studios or working in large performance spaces?  The TEF can help you in all of these areas, but it might be useful to know what you are trying to do in order to find a place to start the discussion.  Also when you post a measurement, if you post the entire TEF file, instead of just the screen capture, I an others can look at the settings as well as the outcome.
Looking forward to hearing from you,

Doug

Charlie

Registered:
Posts: 12
Reply with quote  #3 

Watson,

There are a couple of things happening that make it appear as if the signal going through the low pass filter is arriving later than the signal that is not going through the LP filter.  The first is that the high frequency components of the signal have a shorter period (higher frequency).  Thus they appear first on the measurement when viewing the time domain.  When these higher frequency components are removed (LP filter) all that remains are the lower frequencies so it appears that they are arriving later (they are to some degree & I will get to that in a moment).

 

Try remeasuring with & without the LP filter but confine your sweep to go from -1 kHz to +1 kHz.  This way the test signal will have almost the same frequency content with & without the LP filter.  The true arrival time difference should be much more apparent.

 

The second thing that is happening is the signal passing through the LP filter is being delayed slightly due to the phase shift caused by the filter.  This delay is frequency dependent and can be seen very clearly by viewing the Group Delay.


__________________
Charlie Hughes
watson

Registered:
Posts: 8
Reply with quote  #4 
OK, thx, but... :/

Quote:
the high frequency components of the signal have a shorter period (higher frequency).  Thus they appear first on the measurement when viewing the time domain

but why?
Let's say that we making Time Response Test and we have sweep time set for 3 s = 3000ms;  frequency span = 23999 (start = 1Hz; stop = 24kHz) Hz and time span 9 ms.
So, TEF starts sweeping frequencies with 23999/3000 = 7,9 Hz/ms and as I understand in first milisecond we have 1Hz to 8,9Hz, so in 4,5 milisecond loudspeaker will play 4,5*7,9 = 35,55Hz and 4,5ms is the time of direct arrival of sound ? so what is happening there ? where are those high frequencies which You telling me that appears on the graph when loudspeaker is not playing them. I understand that during one period of 100Hz sin wave we have 10 periods of 1kHz wave but the speed of sound is the same for all of the frequencies. And if 100 Hz sin wave was emitted first, and 1kHz is emitted some X miliseconds later So microphone will receive 100 Hz and X ms later will receive 1kHz wave?
I just don't understand this TEF sweep, the mechanism of measurement
I know that I'm thinking in very different way that I should think, so please, show me the right way of thinking

__________________
pozdro600
Charlie

Registered:
Posts: 12
Reply with quote  #5 

Watson,

My comments about the higher frequencies appearing first on the measurement may have been a poor choice of words on my part.  This could perhaps be better stated that the HF components appear earlier in the Impulse Response of a system than do the LF components.  The measurement method used to obtain the IR is not the issue here.  It doesn't matter if you sweep from 1 Hz to 24 kHz or sweep backwards from 24 kHz to 1 Hz.  The result of the measurement (the IR) will be the same.

 

The IR is the time domain representation of a system's response to a stimulus.  It is possible to measure this directly by using a unit impulse (Dirac) to excite the system under test (SUT).  However, the signal to noise ratio of this type of stimulus is not very good because it has an extremely high crest factor.  To get better SNR a sine sweep is done in the frequency domain.  An Inverse Fourier Transform (IFFT) is then performed on the measured frequency sweep data to yield the same info in the time domain.  This is the IR.

 

Let's use your example of one period of 100 Hz occurring in the same time span as 10 periods of 1000 Hz.  If we look at this in the time domain the first peak of the 1000 Hz component occurs at 0.25 ms (one-quarter period).  The peak of the 100 Hz component occurs at 2.5 ms.  The screen shot of the ETC you posted shows this same type of thing happening.  When the higher frequency components are filtered out you are able, to a certain extent, see the arrival time of the lower frequencies more clearly.


__________________
Charlie Hughes
watson

Registered:
Posts: 8
Reply with quote  #6 
Thank You Charlie.
Thank You very much...
You know, TDS is the technique I haven't seen/heard before, everything was new, so I thought that Time Response is a name of some new kind of measure and so on... and when I read the words - "Impulse Response" of a system - that was so easy and I hit myself in head with laugh
Than we made a big step forward and we found new questions to ask ;]
Now the thing that we would like to ask for is: how to measure frequency response of a reflections. We made some measurements with and without 4 pieces of Ecophon on the floor (12 cm high of construction - foamed polystyrene and ecophon) between loudspeaker and microphone to observe differences in first reflection - from the floor. See attached files. Difference is very clearly to perceive. Reflection is a little bit faster when Ecophon lies on the floor, but is 11dB down relatively to reflection without Ecophon. This is very logical. But how to measure frequency response of the reflections to gather information about frequencies that was cut becouse of use of Ecophon.
Can anybody help us with this subject ??


__________________
pozdro600
Charlie

Registered:
Posts: 12
Reply with quote  #7 

To measure the spectral content (frequency response) of a reflection set the Receive Delay to the arrival time of the reflection instead of the arrival time of the direct sound from the loudspeaker.  The time resolution will need to be fairly short (approximately 2-3 ms) to not let the direct sound into the measurement.  This will limit the frequency resolution to between 333 - 500 Hz.

 

For the measurements you posted the Receive Delay should be approximately 7.21 ms without the absorption and 6.5 ms with the absorption.  If you don't want to change the Receive Delay between measurements you can split the difference and use 6.85 ms.  Good Luck.


__________________
Charlie Hughes
watson

Registered:
Posts: 8
Reply with quote  #8 
Just like I thought, it is about Time Delay, but of course I've forget to set short enough time resolution to receive nothing more than spectrum of the reflection... ehh... :/ but now it seems to look very good. The acoustic in our room have changed a little bit so there are new etc measurements attached. Three files of tds with spectrum of measurements: clear, absorb with 500Hz and 2ms resolution and absorb2 with 1000Hz and 1ms resolution. And the site of Ecophon product we used in measurements: http://www.ecophon-us.com/templates/SystemPage____4732.aspx . See the characteristic of absorption (black dotted line). The biggest αp is in 1000Hz, but our measurements seems not to show this. How to explain/interpret this measurements. Which part of graphs we can trust; Is the data below 500 Hz, when we have freq res 500Hz, trustful and so on... We would like to hear/read the professional interpretation of our measurements. If You have enough time of course :]
And than we will be asking about 3D measurements...
Thanks for Your help


__________________
pozdro600
watson

Registered:
Posts: 8
Reply with quote  #9 
Oh... of course I know that we haven't properly installed Ecophon, so value of maximum absorption is not so important...

__________________
pozdro600
Charlie

Registered:
Posts: 12
Reply with quote  #10 

You may need to modify your measurement set-up so that the arrival of the reflection is occurring at a later time compared to the arrival of the direct sound from the loudspeaker.  The mic needs to be closer to the loudspeaker and they both need to be farther away from the reflecting surface.  Try the following.

 

Place the measurement microphone close to (just to the side of) the loudspeaker.  Both the loudspeaker & the mic should be pointed at the reflecting surface.  This should be the only reflecting surface within the time resolution window you plan on using.  The frequency resolution (and LF limit) you want will define the smallest time resolution you can use.

 

Measure the reflection without and acoustical treatment.

 

Measure the reflection with acoustical treatment applied to the entire reflective surface (or a very large portion of it).

 

Difference these two measurements.  This will give you a fairly good idea of the level reduction of the reflection due to the application of the acoustical treatment.

 

Keep in mind that this level reduction may be valid only for this angle of incidence of the sound relative to the reflecting surface.  The loudspeaker and the mic can be separated while keeping them aimed directly at the surface to measure other angles of incidence.


__________________
Charlie Hughes
watson

Registered:
Posts: 8
Reply with quote  #11 
Hello again.

Now my question is about sound equipment required or preferred to make measurement. Loudspeakers.
What You can recommend us to perform measurements.
For example I was using a spherical sound source (all-directional) to measure RT60 in echoic chamber, do TEF25 requires this kind of sound source or it needs only one loudspeaker. Also to measure STI, what kind of sound source You are using?
And how powerful should be the sound source? 75 or 100 WATS? I know it depends of ambient conditions, like noise... but...

Waiting for any advice any empirical suggestions, we'll be very glad for any answer...
 Thanks

__________________
pozdro600
Previous Topic | Next Topic
Print
Reply

Quick Navigation:


Create your own forum with Website Toolbox!