How to get sub' delay

[dima1stg]

Hello folks. Happy New Year!

Looking for the best way of finding sub's delay. On attached picture, considering that L is left main speaker, is the circled delay the one I'm looking for? Many thanks!

 

[dima1stg]

On another hand, I get this circled delay very different when using LPF. I have XO set to LPF @ 80hz and got delay of 13.6578 ms vs. 3.3160 ms without XO. Can such XO really introduce 10 ms delay? Or am I looking at a wrong thing?

 

[sm52]

13 ms is a more correct figure. But we don't know others. To tell you anything, we need to look at the mdat file of the measurements of the three speakers. Right, left and sub.

 

[dima1stg]

Many thanks for prompt reply.
At this point I only measure sub. Left main is used for a timing reference only. I can provide mdat for the sub, but my question is whether delay mentioned in comments of a measurmenet "Delay 29.3209 ms (10.057 m, 33 ft) using IR start time relative to Acoustic reference played from L with no timing offset" is the right place to get a delay of a measured speaker relative to "L" speaker.

 

[sm52]

It’s correct if the sub is 3-5-7 meters behind the main speakers in time. That is, in order to time align with the main speakers, the sub must be made earlier at this distance. This is a negative delay.

 

[dima1stg]

Front of the sub is 90cm on left of reference speaker (L) and 15cm deeper. There is no way I can have it further behind. So where can I find in REW the delay between them?

 

[sm52]

Front of the sub is 90cm on left of reference speaker (L) and 15cm deeper.

If you don't have a receiver or audio processor that can apply delays, you won't be able to align the sub in time with the main speakers. There is a way to use a software player on a computer with a plugin that processes the signal from the player the same way an audio processor does.

 

[dima1stg]

I do have miniDSP, but in order to apply delay I need first to figure it out, hense my initial question - how can I figure it with REW, if at all possible.

 

[sm52]

From post #3:

To tell you anything, we need to look at the mdat file of the measurements of the three speakers. Right, left and sub.

The left one will be in all measurements as a reference.

 

[dima1stg]

Here is mdat with all channels. L is left channel as well as a reference. So what's the delay of Sub1 relative to mains, Sub2 relative to mains, and Sub1 relative to Sub2? Many thanks!

 

[sm52]

You have not selected an output device in any measurement, no output device output has been selected. Therefore, two devices can participate in the output signal at once. Or how the sound system of the computer OS decides. Both L+R channels are included in the subwoofer measurement. If you measure the sub L, then the output channel will be L. And if the sub is R, then the output channel will be R. In this case, there should always be one reference output channel (you have L). If you are measuring the left speaker, the measurement will be correct. First a whistle from the left speaker, then a measuring sweep, then another whistle. To measure the right speaker, you should specify the output channel R, the reference L. But if you measure the left sub, then the whistle from the left speaker will pass through the left speaker, and the measurement sweep will pass through the left speaker and the left sub simultaneously. Or do you have four output channels? The delays cannot be determined from your measurements because there are no distinct peaks in the measurement impulse. There are many of them, but there should be one. Also measure all speakers and subs at the full range, from 0 hertz to 48 kHz.
When selecting an output device, specify a line with EXCL.

 

[dima1stg]

As for output - it is set to default, but default is set to the right path. Otherwise the sound wouldn't be heard at all.
I was always wondering what EXL stands for? Can you enlighten me please...

As for the selection of reference speaker, please see https://www.avnirvana.com/threads/trace-arithmetic-question.12916/.
As John said, "To meaningfully compare phase between measurements they must have a common timing reference. You used the left speaker as reference when measuring the left, and the right speaker as reference when measuring the right, so the results do not have a common reference. Pick a speaker to use as the reference and stick to it."

Since my goal is to find out the relationship between various speakers in a system (vs. measuring each speaker on its own), as per John (and common sense) there should be a single reference.

I'm using OpenDRC having 2x8 matrix, each driver (main speakers are bi-amped) is driven via it's own channel. During the measurement all channels, except the one being measured and L's tweeter being the reference, are muted.

"Both L+R channels are included in the subwoofer measurement" - how did you figure this out? If this would be the case, sub graphs wouldn't roll-off.

This I don't get: "The delays cannot be determined from your measurements because there are no distinct peaks in the measurement impulse. There are many of them, but there should be one". I'm not really familiar with impulse response, just simply measuring with timing reference. Should I do something else?

Having no problem to make new full range measurements, however what's the point of measuring the full range if we're talking sub' delay relative to main? Lets say we have full range measurements of mains and 30-100hz range of subs. What's the advantage vs. having same 30-100hz range of both subs and mains?

At the end, I guess it would be much easier to explain the procedure of using REW to calculating relative driver' delays in general, somethink like: get such and such measurement with such and such parameters, look into here or/and there, substract this from that, etc.

Thanks!

 

[sm52]

I was always wondering what EXL stands for?

From help, :

Getting set up for measuring​

"...when using the default setting REW cannot determine the specific capabilities of the device such as the bit depths or sample rates it supports, so it is strongly recommended to select the specific devices, inputs and outputs you wish to use for measurement in the Soundcard Preferences panel and do not leave the selection as default."
If you select an output device whose name begins with EXCL, REW will use WASAPI exclusive drivers, which prevent other programs or devices from using the computer's internal sound card. Then no system sounds from events in the system, or sounds from other programs will be mixed along with the measurement.

Since my goal is to find out the relationship between various speakers in a system (vs. measuring each speaker on its own), as per John (and common sense) there should be a single reference.

Right.

During the measurement all channels, except the one being measured and L's tweeter being the reference, are muted.

This is good. But if you select the output channel L+R, then the sound will go to two channels at once.

I'm not really familiar with impulse response, just simply measuring with timing reference. Should I do something else?

It is from the impulse graphs that the timing relationships between the speakers can be accurately determined. And in order to see and coordinate the phases later, you need the 'Alignment tool'.

what's the point of measuring the full range

If your measurements are cut off at 100 hertz, you won't see the phases of the two speakers you're trying to match after 100 hertz. If the subwoofer's cutoff frequency is 80 hertz, then its influence on the main speaker can be up to 200 hertz until the subwoofer level drops enough. If the two are out of phase, you will get pits. To get accurate data within the range you need, John recommends always measuring the full range.
I can't spend too much time explaining. But I can give my answer on what delays are needed to coordinate two speakers and two subwoofers.

Also measure all speakers and subs at the full range, from 0 hertz to 48 kHz.

Specify the left tweeter channel as the "timing reference output" channel, and change the "output channel" in turn.

 

[dima1stg]

I figured that LPF XO @ sub1 channel adds about 13-16 ms delay. I also have LPF+HPF XO set @ sub2 channel and HPF XO on main channels. Should I measure with these XOs or without?

 

[sm52]

To avoid correcting something that might be wrong, and thus stacking wrongs on top of each other, it is necessary to turn off all crossovers and measure each channel separately, but with a common synchronization in time. And then think about what ways to use to bring all the channels into the overall picture.

 

[dima1stg]

Here it goes. All channels are measured separately. Subs are fed from L+R source. No XOs, only exception is HPF on L @ 4,500hz when measuring subs since L is a reference channel. All channels are with timing reference. Mike is at 90 degrees, though I'm not sure if this is correct for higher frequencies. REW' max frequency is 24,000 hz. Thanks!

 

[sm52]

Can you set negative delay for any channel? For example, - 4.5 meters, or -13.2 ms. I'm not sure miniDSP allows this.

 

[dima1stg]

No, it does not. However it is easy to normalize accross the channels, so that channel with most negative delay would have 0 and other channels would add positive delay equal in value to that most negative one. And miniDSP delay is measured in ms, though it's easy to comvert knowing speed of sound.

 

[sm52]

According to your measurements, it turns out that each of L and R must be delayed by 12.8 ms to match the SVS sub. And to coordinate with the YSW sub, they must be delayed by 14.8 ms and the polarity of the YSW sub must be changed by 180 degrees. The polarity is easy to change. And try making the delays for L and R 13.8 ms. This is if you don't use crossovers. At what frequency are you planning to cut the crossover?
More. I previously lowered the YSW level to -9 dB, and SVS to -3.4 dB.

 

[dima1stg]

Lots of thanks!
Most likely I will use attached house curve and YSW and SVS volumes are set to accomodate it. Sorry, forgot to mention it before.
Planning to have LR 48 db/oct @ 80hz crossover on SVS and mains, and HPF LR 48 db/oct @ 43hz + LPF LR 48 db/oct @ 66hz with YSW just to cover the SVS' dip around 54-55 hz, all these will introduce delays and whole picture will be different though.

 

[sm52]

The picture will change, but not much. It seems that the measurements of the subs are normal, but to be sure that there are no pitfalls, it is better to take measurements as for the main speakers. That is, when measuring a subwoofer L, indicate the output channel L and the reference output channel L. And when measuring a subwoofer R, the output channel R and the reference channel L. And let the measurement range for subwoofers also be 0-24 kHz. Subs will not be harmed. Make the same measurements with all the filters and my delays. If you have a calibration file for 0 degrees, it's best to use that and point the mic at the center of the stage at ear height.