Room response time and TrueRTA

Discussion in 'Speakers' started by Ilkka R, Jan 16, 2005.

  1. Ilkka R

    Ilkka R Second Unit

    Joined:
    May 19, 2004
    Messages:
    270
    Likes Received:
    0
    It is said that it takes about 250ms signal to get those room modes ringing. Here (bottom) they actually show a formula for this.

    Trise = 0.32 T60

    So it's about third from T60 time. If a good decay is say 600-800ms then that 250ms would be quite correct.
    Now we come to TrueRTA and it's Quick Sweep. QS takes about 1500ms and it goes 10Hz-20kHz. If it's linear each major third (1/3 oct) should take about 50ms or less (1500ms/31). Even less for each minor and so on. This indicates that QS would be too fast for room to response and therefor room modes should't be built. In the TrueRTA help John L Murphy says "There is gating but it is sufficiently wide that it does not exclude room reflections on the acquired sweep. Because of the wide gate, room reflections will influence measurements when testing indoors."

    I have measured with both ways, QS and pink noise and the results are the same, so Murphy is of course right. So is that 250ms correct or where is the problem? How fast signal we need to exclude those room modes? Am I confusing room modes and normal reflections?
     
  2. Ilkka R

    Ilkka R Second Unit

    Joined:
    May 19, 2004
    Messages:
    270
    Likes Received:
    0
    I didn't expect to get 100 replies, but I was hoping maybe Mark Seaton, Edward JM, Wayne P etc. could post few lines...

    I know this isn't an easy question.
     
  3. Edward J M

    Edward J M Cinematographer

    Joined:
    Sep 22, 2002
    Messages:
    2,031
    Likes Received:
    0


    The digitally generated signal sweeps logarithmic frequency intervals in equal time. That is, as much time is spent sweeping the interval from 20 to 40 Hz as from 10kHz to 20 kHz.

    If you want to spend even more time over the log frequency intervals, reduce the input sampling frequency to 8 kHz, which will limit the sweep from 10 Hz to 4 kHz. You'll notice the sweep is considerably slower, and will give better results below 20 Hz.

    Regardless, the proof's in the pudding - the Quick Sweep and pink noise give the same in-room results (as you already noted).
     
  4. Wayne A. Pflughaupt

    Joined:
    Aug 5, 1999
    Messages:
    6,061
    Likes Received:
    25
    Location:
    Katy, TX
    Real Name:
    Wayne
    I’ve never been able to buy into the idea that reflections unduly influence RTA readings.

    The problem is that the mic and RTA are “dumb.” The mic only “sees” the loudest sound pressure level it gets at a given frequency, and the RTA registers what the mic “sees.” Everyone knows that SPL dwindles with distance. By the time the reflected sound bounces around the room a few times and reaches the mic, it arrives at the mic at a lower sound pressure level. At the same time, the mic is also getting the full-volume signal from the speaker’s direct, as-yet unreflected soundwave. Since it’s louder than the reflected sound, the direct reading is what the mic sends to the RTA.

    Unlike the RTA, the ear can process and register both the direct and the reflected sound. And it’s a given that the reflected sound can affect and contribute to what we perceive to be accurate response.

    Thus, I don’t see “room reflections [influencing] measurements.” If anything, what can render the measurements inaccurate is the RTA not registering them.

    Standing waves are a different story, since their cumulative effect ends up increasing the SPL of the original signal. So naturally, a RTA will register standing waves.

    I know Mark has more practical and real-world experience at this than I do. Perhaps he can clarify some of this and/or correct me if I’ve missed it.

    As far as the proper ms of correction, I can’t help you there.

    Regards,
    Wayne A. Pflughaupt
     
  5. FeisalK

    FeisalK Screenwriter

    Joined:
    May 1, 2003
    Messages:
    1,245
    Likes Received:
    0
  6. BruceD

    BruceD Screenwriter

    Joined:
    Apr 12, 1999
    Messages:
    1,220
    Likes Received:
    0
    One of the reasons I like to use an MLS-based signal test program (ETF5 in my case) is due to the various measurement responses like RT60, Impulse Response, Energy Time Curves, etc. with gating control (windowing). I can even capture responses below 10ms to differentiate source signal from reflections.
     
  7. Ilkka R

    Ilkka R Second Unit

    Joined:
    May 19, 2004
    Messages:
    270
    Likes Received:
    0
    Thanks guys for your effort, but you actually didn't tell anything new. I know how and why standing waves appear, but I'm more interested about the time issues. How long the signal has to ring in order to standing waves to build? Usually 250ms is suggested.

    We know that the whole sweeps takes 1500ms (TrueRTA). If it's logarithmic as Ed says, then 10Hz to 100Hz would take a little less than 500ms, right? For every 5Hz piece it would take about 25ms. A little bit more at the low-end and little bit less when closer to 100Hz. Still following?

    So for every 5Hz there would be ~25ms sound coming out of the speakers. And yet room modes have enough time to build. So that suggested 250ms is not corrent.

    Actually what Wayne said about reflections, it is not true. Yes, microphone and RTA are dumb, but they don't measure max spl, they measure total power. So when gate is long enough (about 10-20ms), reflections will always affect. The lower the frequency measured, the longer the gate needed.

    Here is a picture showing normal two-way speakers in-room response. The gate is very short, only 5ms. The mic records only the direct sound and not the reflections as they will take longer time to arrive. Curve is very similar if measured in anechoic chamber. Only frequencies below 200Hz are missing because the whole wave doesn't fit in that short gate. Lower the freq, longer the wave. There is a 1/3 oct. smooth, but that doesn't affect very much.

    [​IMG]

    The second picture is showing the same speaker but now using gate of 180ms. As you see reflections have now enough time to arrive and to bring their own flavour to the mix. Also room modes have developed (42Hz, 160Hz). This is what we get with quick sweep or pink noise.

    [​IMG]

    This whole conversation began on the Finnish audioforum when we discussed about BFD and equalisation ganerally. Someone pointed out that what if you have a room mode (a hill/bump, not valley) which builds after 250ms or similar and you tame it with you BFD/EQ. Let's say it's a 10dB bump. When you are measuring with QS or pink noise, your FR seems flat. Then you put your favourite cd in and start to listen let's say music with normal bassdrum in it. The signal that bassdrum produces is very short, less than 250ms for sure. Now the room mode doesn't have enough time to build and you have a huge 10dB dip in your FR. With longer type of signals that isn't a problem, but there are a lot of material with short and quick punches, kicks etc...and with music transiets are always there.

    Here is one solution for this problem. Not cheap though. [​IMG]

    So if you want a perfect sound, first you have to find out your rooms rise time for every frequency and then measure it's decay (RT-60). Then apply a custom made correction that is time and frequency relevant... Ok, maybe I went a little bit too far, but you got the picture? [​IMG]
     
  8. Edward J M

    Edward J M Cinematographer

    Joined:
    Sep 22, 2002
    Messages:
    2,031
    Likes Received:
    0
    I think there is a difference between the amount of time it takes for standing waves to develop, vs. the amount of time room decay typically lasts.

    My room starts to show room gain at around 30 Hz. 30 Hz is a 38 foot wavelength. My room is about 18 feet long, so this makes sense.

    With sound traveling at 1150 fps, it only takes about 30 ms for a wavelength to leave the subwoofer, hit the far wall, and return to the source.
     
  9. Ilkka R

    Ilkka R Second Unit

    Joined:
    May 19, 2004
    Messages:
    270
    Likes Received:
    0
    Yep, that's true. I already posted a formula for that.

    Trise = 0.32 x T60

    So it's about third from T60 time. T60/RT60 means decay time.

    That 30ms could be closer to truth than 250ms. Of course it depends of frequency. The lower the freq. the longer time it needs to "reflect".
     
  10. Edward J M

    Edward J M Cinematographer

    Joined:
    Sep 22, 2002
    Messages:
    2,031
    Likes Received:
    0


    All sound waves travel at the same speed; what varies is the wavelength, which is freqency dependent.
     
  11. Ilkka R

    Ilkka R Second Unit

    Joined:
    May 19, 2004
    Messages:
    270
    Likes Received:
    0
    True, I meant the same thing, waves are longer and they take longer time to reflect fully and start building gain.
     
  12. Wayne A. Pflughaupt

    Joined:
    Aug 5, 1999
    Messages:
    6,061
    Likes Received:
    25
    Location:
    Katy, TX
    Real Name:
    Wayne
    Ilkka,

    First, you won’t often find a kick drum signal that short, generally speaking, since they have considerable decay and after-ring, especially the drum’s lower frequency resonance. The sharp instantaneous attack of the petal striking the head is probably shorter than 250 ms, but it’s a fairly high frequency, well above virtually any room mode.

    Second, unlike a sine wave, which is a pure tone, a note from an instrument is complex, with lots of harmonics. You might loose the absolute fundamental, but you’ll still hear the kick drum. It’s not like it’s going to be completely sucked out because that frequency was notched.

    Third, kick drums are seldom accurately recorded anyway. [​IMG]

    Regards,
    Wayne A. Pflughaupt
     
  13. Ilkka R

    Ilkka R Second Unit

    Joined:
    May 19, 2004
    Messages:
    270
    Likes Received:
    0
    No, you are so wrong. If you add 80dB and 90dB, you still get 90dB, that is true. But reflected sound can be higher (intensity) than the direct sound. Of course it depends of materials in your room etc. If you don't believe try this: grab one spl-meter, play some broadband pink noise, say 400Hz-2kHz for example. Adjust volume that meter shows for example 80dB, from 2m/6ft or similar. Then move you gear outside, don't touch the volume. Play the same noise and check your meter. Does it show 80dB? No, it shows less.

    If you have a properly damped room, difference might be low as 1-2dB. If you run a RTA you can get even more specific info. Some frequencies might get boosted more than others. If you have lots of hard surfaces, mirrors and glass, there can be huge peaks and valleys at some frequencies.

    Here is a picture of speakers FR measured from 15cm/6". QS was used.
    [​IMG]

    Then the same speaker measured from 1.8m/6'.
    [​IMG]

    Don't tell me that reflections don't affect. :p)
     
  14. Lev-S

    Lev-S Second Unit

    Joined:
    Jun 29, 2003
    Messages:
    324
    Likes Received:
    0
    My brain hurts. [​IMG]
     
  15. Ilkka R

    Ilkka R Second Unit

    Joined:
    May 19, 2004
    Messages:
    270
    Likes Received:
    0

    This is true, but I'll bet if you eq a rather wide bump, you will definetely hear it, less punch etc. Gladly room modes are often quite narrow.
     
  16. Wayne A. Pflughaupt

    Joined:
    Aug 5, 1999
    Messages:
    6,061
    Likes Received:
    25
    Location:
    Katy, TX
    Real Name:
    Wayne
    Once again, you can’t support it with charts like these, taken under different conditions.

    For instance, some of the differences in the second chart – the lower SPL, the sag at the highest frequencies and the added extension at the lowest - are solely the result of the increased distance. Also, you didn't indicated if both speakers were playing. If so, that would easily explain the ragged response in the second chart: Comb filtering, as a result of phase issues between the two speakers.

    If you want to give us a pair of charts to support your contention, they would have to be two readings with the mic at the same distance. The first would be an instantaneous “snapshot” of the first sound wave taken as it arrives at the speaker, before any reflections can arrive. The second would be a time-lapsed reading. The former method is what the major audio mags use these days in lieu of anechoic chambers.

    In any event, I’m not altogether sure we’re on the same page here. When I said I didn’t think “reflections unduly influence RTA readings,” I was referring to the mid-and-upper frequencies that normally get absorbed (or attenuated) in a reasonably-dampened room (as would be the case with any decent sounding HT room). Naturally, below the midrange, soundwaves aren’t absorbed by normal room furnishings, and only moderately so with room treatments (bass traps notwithstanding). Of course, the room influences a speaker’s response below 500 Hz, and that influence will show up with a RTA reading. Technically speaking, the room’s effect on response at mid-to-lower frequencies is caused by reflections, but you seldom see it labeled as such. You usually see it discussed as the room’s “influence,” “cabin gain” or “room gain,” “standing waves,” “room modes,” etc.

    So basically, I think we’re talking about the same thing here, only from a different approach.

    Regards,
    Wayne A. Pflughaupt
     
  17. Ilkka R

    Ilkka R Second Unit

    Joined:
    May 19, 2004
    Messages:
    270
    Likes Received:
    0
    I agree, mostly. But room does influence also above 500Hz. And now I'm talking about normal rooms, not dedicated HT rooms which are built and measured carefully. Those rooms are built to absorb most of the soundwaves. Normal rooms are not. Even on those rooms below 500Hz frequencies are diffucult to handle, beacauce of the longer waves. They need huge traps ect as you said.
    I can show you ten measurements from normal Finnish living rooms, they ALL show differencies between nearfield and listening place measurements, at range 20Hz-20kHz. Not just below 500Hz.

    Room modes only affect at lower frequencies. At higher they become so frequent that you can't notice them anymore. For example if you have 5m long room, standing waves are at 34Hz, 68Hz, 102Hz, 136Hz, 170Hz, 204Hz. After that they become so frequent that their affect will be cancelled by each other.
     
  18. Max F

    Max F Second Unit

    Joined:
    Jun 26, 2004
    Messages:
    250
    Likes Received:
    0

    Ilkka,

    Maybe its your English (since you're from Finland), but you are coming across as being rather rude. I think Wayne has shown alot of patience in trying to answer your questions. I hope you wouldn't talk to him this way in person.
     
  19. Ilkka R

    Ilkka R Second Unit

    Joined:
    May 19, 2004
    Messages:
    270
    Likes Received:
    0
    Max, We Finns are more straightforward so don't take it personally.
    This has been the most informative post to this subject and I believe this is quite correct. At 30Hz it takes about 30ms to build up room gain and room modes to appear. At higher frequencies even less. So I believe we can safely say that this 250ms what is suggested (before) is way wrong. Room modes doesn't take that long to build up.
     
  20. BruceD

    BruceD Screenwriter

    Joined:
    Apr 12, 1999
    Messages:
    1,220
    Likes Received:
    0
    Ilkka,

    As I indicated previously, RTA measurements cannot distinguish between direct and reflected sound, but MLS-based measurement software can.

    Here is a link to a discussion about the influence of early reflections (
     

Share This Page