PA Audio Delay lines

[addoaddo]

I need to install a PA system in a large thin room. The stage will mainly be used for speech and a small amount of music. As the room is long and thin I was planning to put three speakers evenly spaced down each long wall. This wall is 37m long.

Should I look at getting delays for these speakers or will they be OK. Currently I do not have a delay unit but might be able to source one if needed. As this is for a charity I do not want to waste money on equipment that is not needed.

 

Calculating the delay I think is OK as you use 1msec per foot between speakers?

[Shez]

Sound will take about 100ms to get from one end of the room to the other. That will be very audible as an echo which will cause major intelligibility problems so I definitely would put delays in.

[Bobbsy]

It's actually a bit more, closer to 1.1 ms per foot depending on temperature and barometric pressure. However, for back of the envelope stuff, I use 1 ms per foot as well.

 

If you go with your scheme, I assume the third speakers will be at about the 25 metre point. This would give a delay of about 80ms from the front speakers and 40ms from the intermediate ones. This is enough to give definite problems with clarity/intelligibility. However, the thing to consider is that any calculation will only be valid for one specific spot--there's no "one size fits all" delay.

 

Frankly, rather than three speakers and delays, my first instinct would be to try and find a different solution for speaker placement. Of course, this depends on what you're trying to do in the long, skinny room. For some applications you need a near equal level throughout--others (for example a wedding venue) actually benefit from a loud end and a quiet end.

[Simon Lewis]

1ms per foot is about there, but you then want to add ~ 10 to 15ms on top. This allows you to invoke the Haas Effect:

 

secondary sounds arriving within 30 milliseconds or so of the initial or primary sound fuse with it to produce one apparent sound of increased loudness. Furthermore, the secondary sound may be up to 10dB louder than the primary sound before being judged equally as loud.

 

You will also want to invoke the Precedence Effect through this delay:

 

Listeners tend to lock on to the first arriving sound and its direction, ignoring short term delayed secondary sounds or reflections. This maintains correct localisation

.

 

Despite what the textbooks say, the two aren't the same ;-)

[Al M]

I thought that current thinking was that invoking things such as the 'precedence effect' and the 'Haas effect' are generally advised against nowadays, as they introduce unnecessary comb filtering for not much benefit.

 

 

 

 

The likes of Bob MCarthy in his book Sound Systems: Design and Optimisation advise against them, and suggest setting the delay time as accurately as possible where the coverage area of the two speakers meet.

 

 

 

 

Granted, in this post we are not talking about using SIM 3 or SMAART to tune Carnegie Hall, and maybe there would be some benefit in a long thin room like this one.

 

 

[addoaddo]

Thanks for the advice, I need to delay the second and third row of speakers. I am a bit confused about the precedence and Haas effect and if this is important for a one off event or is it more considered for permanent installations.

[Simon Lewis]

I thought that current thinking was that invoking things such as the 'precedence effect' and the 'Haas effect' are generally advised against nowadays, as they introduce unnecessary comb filtering for not much benefit.

 

True, but there is a (as you suggest) a difference between a high end system that might be operating in true LCR and trying to stop a 100ms late delayed reflection destroy the last semblance of intelligibility in a 37m long thin room....

[mackerr]

I thought that current thinking was that invoking things such as the 'precedence effect' and the 'Haas effect' are generally advised against nowadays, as they introduce unnecessary comb filtering for not much benefit.

 

True, but there is a (as you suggest) a difference between a high end system that might be operating in true LCR and trying to stop a 100ms late delayed reflection destroy the last semblance of intelligibility in a 37m long thin room....

 

To get that semblance of intelligibility you will want the delay time as accurate as possible. Over delaying just decreases intelligibility without adding any appreciable benefit.

 

For someone who does not have the measurement tools available to accurately measure those delay times I find that tapping on a mic and listening while increasing the delay till the image moves to the stage is a pretty accurate substitute.

 

Mac

[J Pearce]

Its all good until a sweaty rain soaked audience sit down and change the humidity of the room...

[Simon Lewis]

Its all good until a sweaty rain soaked audience sit down and change the humidity of the room...

 

Humidity has only a very limited effect upon the speed of sound...

temperature, however....

[J Pearce]

The audience are sweaty and can therefore be assumed to be hot :P

[mackerr]

Its all good until a sweaty rain soaked audience sit down and change the humidity of the room...

 

Only the temperature matters, and at 65ºF the speed of sound is 1122ft/sec, and at 85ºF it is 1144ft/sec. At 75' that would be a time of flight of either 66.84ms, or 65.56ms. If you have "over delayed" the error will be greater in a warmer room, but by an insignificant amount.

 

Off axis of the line between the main speaker and the delay speaker the time difference between them will be less, so the maximum delay would be with the 2 systems in line with each other.

 

Mac

[Simon Lewis]

Mac,

 

I'm aware of the quite rational "delay can do more harm than good" reasoning, but find it interesting that you suggest delaying until the image moves to the stage. This suggests that precedence is taking place. I do not have the data to hand, but believe that precedence occurs when the secondary sound arrives between 2ms and 50ms after the first wavefront. If it is sooner, it sums, if later it tends to be perceived as an echo. Your suggestion may not be fully invoking the special case of precedence (Haas Effect peaks at around 15ms) but it does still provide the necessary directional clues.

 

Simon

[mackerr]

Mac,

 

I'm aware of the quite rational "delay can do more harm than good" reasoning, but find it interesting that you suggest delaying until the image moves to the stage. This suggests that precedence is taking place. I do not have the data to hand, but believe that precedence occurs when the secondary sound arrives between 2ms and 50ms after the first wavefront. If it is sooner, it sums, if later it tends to be perceived as an echo. Your suggestion may not be fully invoking the special case of precedence (Haas Effect peaks at around 15ms) but it does still provide the necessary directional clues.

 

Simon

 

The image moves to the stage immediately after the accurate delay time is achieved. The fact that the image may hold up with as much as 50ms of excess delay is unrelated to getting maximum intelligibility. You can verify this with measurement with 2 speakers in even a small room. The exact measured delay time will move the image to the main speaker by making the delay speaker invisible.

 

If you are able to use a measurement tool like Smaart or SysTune you will also be able to look at polarity of both systems. Changing the absolute polarity of the delay speaker can make it "appear" and "disappear".

 

Both of these demonstrations are part of the Smaart training process.

 

Mac

[waster]

1ms per foot is about there, but you then want to add ~ 10 to 15ms on top. This allows you to invoke the Haas Effect:

 

secondary sounds arriving within 30 milliseconds or so of the initial or primary sound fuse with it to produce one apparent sound of increased loudness. Furthermore, the secondary sound may be up to 10dB louder than the primary sound before being judged equally as loud.

 

 

NO! Why do people bother to measure delay times and then mess them up with adding an arbitrary amount of 'hass effect' delay. I'll take the time aligned option please. Especially if you are going on the 1ms is thereabouts... and then adding more delay.

 

To the OP... Do you have a long tape measure or a laser measure? I'm assuming you don't have a measurement program. This is what I would suggest as a quick and dirty way to get the delays. Stand where you are getting coverage by the main speaker and your first delay measure the distance between the main and you and then the delay and you. Subtract the little number from the big number and feed the answer into your delay processor. Then listen.

 

Though I will give MacKerrs method a go sometime.