Reference Speaker when using acoustic timing reference to measure delay

[MAD]

I'm using a UMIK1 USB microphone for input and an external DAC for output. I want to measure two drivers in a box (2x12" MID and 1x 90x40 degrees horn HI) and want to measure the delay beween the two drivers coming from their mounting distance (both on one baffle but the horn is deeper then the cone of the 12"). Am I right that I have to set "use acoustic timing reference" because I cannot use loopback as reference, then I set the output channel to "left" and connect the HI channel to left and MID channel to right? The sweep will then be supplied at left channel (HI) and the acoustic reference signal at right channel (MID), correct? So the mid channel will be my reference speaker if I understand correctly? Where can I see then the delay between HI and MID based resulting on that measurement?
I'm new to using REW so I'm not sure to having understood correctly the process of measuring delays with it :-)
Martin

One more thing, I should say that the microphone is directed to the center of the box for all measurements (spl, phase, delay) and will not be moved

 

[AudiocRaver]

Your description sounds like you are setting up for measurements properly.

 

[MAD]

That‘s good to hear but I still don‘t understand where to find the measured delay between HI and MID following my procedure :-)
Can you give me a hint?

 

[Richard Gilbert]

That‘s good to hear but I still don‘t understand where to find the measured delay between HI and MID following my procedure :-)
Can you give me a hint?

The delay will show up in the info window in each measurement tab on the far left of the screen. A new measurement tab comes up for each measurement you do and will have the date & time the measurement was taken at the top of the tab. Clicking on the date & time will allow you to give each measurement a name in place of the date and time.
The delay amount will be displayed in + or - an amount of milliseconds and + or - an amount of either feet or meters depending on how you have it set in the preferences to display distance.

 

[MAD]

Thanks, I will give it a try

 

[AudiocRaver]

Thanks, Richard.

 

[Richard Gilbert]

No problem. Feel free to hit me up if i can ever help with anything else. I'm by no means a REW pro but I can get through the basics ok. Its such a complicated and useful tool that I'm thankful for almost daily but even after 2 years it still intimidates me a bit. I just wish I could use it to its full abilities. Hopefully some day. Good luck with your project brother. Let me know if I can help again.

 

[John Mulcahy]

You can also see the delay in the Info window for the measurement, which appears when the Info button in the toolbar is pressed.

 

[MAD]

You can also see the delay in the Info window for the measurement, which appears when the Info button in the toolbar is pressed.

Thanks John - and let me say first of all: Thank you for your great work!
I also appreciate the immediate and helpful support in this forum!
Back to my topic :-)
I‘ve used the HT as acoustical reference, set a delay of .25ms for the MT based on the delay shown in the info window (in line with my expectation based on the legth of the HT horn. So delay (and group delay too) is now zero.
Next I would like to correct the run time from the speaker to the microphone (distance is 1.3m), how can I measure and correct that in REW? I‘ve tried to figure it out with the manual and my measurement but failed :-(

 

[Richard Gilbert]

Where else can you get advise from the creator/Master of REW. Second to none.Thank you, more than words can express, Sir John Mulcahy for this amazing program/tool(REW) and AV NIRVANA. You've certainly changed what myself and my system are capable of.

 

[John Mulcahy]

There is no other delay to correct. Using the HT as the reference means the transit delay from speaker to mic is automatically removed since the HT signal, which is acting as the reference, has travelled the same path.

 

[MAD]

I'm not sure to fully get it, so let me try to rephrase what I understand :-)
The transit delay to the HT (reference speaker) is automatically removed (including the measurement system delay because REW cannot distinguish between them), correct? And is this the reason why group delay is zero for the HT? And what is the reason for still having an estimated IR delay of -0.036ms for the HT (shouldn't it be a minium phase system)? And should I shift the IR response by that value to correct the results? Last but not least, is the phase graph also recalculated automatically, so (if measurement system delay is neglected) it will show the phase values "at the baffle"?

 

[John Mulcahy]

This post answers most of those questions. The phase plot is calculated relative to the timing reference, whether those are baffle figures depends on the whether the distance to the reference is the same as the distance to the baffle.

 

[MAD]

I've read the thread you are referring too before, but was still confused because I did two measurements (1m and 1.4m distance) of my Mid with the same acoustic reference (Hi) and found that the shown phase response was not the same, so I assumed I have to somehow subtract the distance. Redoing this measurement today it worked fine - same phase response shown for both distances. I can not recall what I did wrong before, but this is what I did today:
1) using Hi as the acoustic reference, measured Hi and Mid
2) using a window of 3.4ms (using "apply windows to all, keep ref time") to remove room influence, so measurement is only realistic >300Hz
3) Adding a delay of 0.25ms in my controller to the Mid too correct position (Hi is a horn, roughly 7cm behind the Mid cone on the baffle)
4) remeasure and repeat 2), delay is now showing in REW as 0.0ms for Hi and Mid
- Is this the way to get useful SPL and phase measurements (for setting the XO) for my configuration?
- If yes, one last question: When using "Estimate IR delay" after 4), it shows .5cm for Hi and 11.6cm for Mid. Why?

 

[MAD]

These are my two measurements, if needed to answer my question...

 

[John Mulcahy]

As you may have seen mentioned in the other thread, minimum phase calculation needs information that extends above and below the frequency range of the device being measured. Your measurements cover a relatively narrow frequency span, particularly the mid measurement, so the minimum phase estimate used in estimating the IR delay will not be accurate.

 

[MAD]

I've seen that, but I thought +1 octave below/above the frequency range I want use is enough for measuring Mid. So I've done the measurement again (Mid 20Hz-16kHz, Hi 250Hz-20kHz, both with Hi being the acoustical reference, fixed microphone position, Mid 0.25ms delay), and the estimated IR delay for Hi is now near 0 (1mm), but for Mid it's a value of 4,1cm, what is the reason for that? I would have expected that this is (near) zero!?

And one more question. Regarding acoustic timing reference you state in the other thread you are referring to "That should leave only constant delays, since the variable buffering delay is removed" and above you said "Using the HT as the reference means the transit delay from speaker to mic is automatically removed" - so what is the constant delay you are mentioning - processing delay?

 

[MAD]

These are the new measurements I'm referring to...

 

[John Mulcahy]

Constant delays are the acoustic transit time and any other fixed delays elsewhere in the system, whatever they might be. The measurement system has no way to distinguish where a delay occurs.

The system delay estimate has an implicit assumption that the systems being tested have similar HF bandwidth as it is comparing the locations of the IR peaks of the reference and the measurement. If the system being measured has a lower bandwidth than the reference, e.g. the mid driver, its IR will have a slower rise time and the peak will be delayed. In that case the system delay estimate relative to the reference would tend to position the measurement further back in time that it should be, with the IR rise starting before t=0. The minimum phase estimate detects that, hence the delay it suggests to apply. To check alignment it may be better to make a combined measurement of mid + hi and look at the IR and the step response.

 

[MAD]

John, thanks for the clarification in the second clause, understood. But regarding acoustic transit time being a constant delay (which is so far clear to me) in this context is confusing me. Looking at your answer above and your statement in the other thread ("That should leave only constant delays...) it means to me that with acoustic reference the acoustic transit delay is still in the measurement, but in #11 above you say that acoustic transit delay is automatically removed when using acoustic reference?

 

[John Mulcahy]

The delays on the way to the sources include a variable buffering element. Since the reference and the measurement signal travel the same signal path on the way to the sources they both experience the same delay, however it varies, and so do not vary relative to one another. After leaving the sources the remaining delays are due to the acoustic path lengths from each source to the mic. The relative delay when the signals arrive at the mic is due to the difference in those acoustic path lengths. If the reference and the system being measured are the same distance from the mic there would be no difference.

 

[MAD]

To check alignment it may be better to make a combined measurement of mid + hi and look at the IR and the step response.

Based on your feedback and clarification I've done some more measurements:
1) Mid+Hi w/o delay
2) Mid+Hi with .25ms delay to the Mid
Can you give me a hint where to see the (non-)alignment?

 

[MAD]

Measurements...

 

[MAD]

Regarding posts #17, #18 and my question on time aligment of Mid and Hi: I've did some further thinking and analysis - when accepting the proposed IR delay estimation (0.12ms / 4.1cm) for Mid, Hi and Mid IRs are in sync (regarding slope not peaks). So I've changed the delay for Mid from 0.25ms to 0.37ms and the combined IR now looks OK (before it was as shown in post #23 (2nd measurement).
Correct?

 

[EarlK]

Regarding posts #17, #18 and my question on time aligment of Mid and Hi: I've did some further thinking and analysis - when accepting the proposed IR delay estimation (0.12ms / 4.1cm) for Mid, Hi and Mid IRs are in sync (regarding slope not peaks). So I've changed the delay for Mid from 0.25ms to 0.37ms and the combined IR now looks OK (before it was as shown in post #23 (2nd measurement).
Correct?

This method ( of aligning using the beginning of the slope ) seems to be better ( IMO > though I do defer to jtalden on these matters ).

It's an interesting thread, but I can't help feeling you'd be better of finding another way of determining acoustic offsets ( for crossover design I assume ) .

I'd ( instead ) relegate this current timing methodology to the syncing of speakers in an HT environment.

A lot of network designers are now using "Curve Fitting" ( as espoused by David Ralph &/or Jeff Bagby ) since it offers a solid track record of "tried & true results".
- While his explanation is somewhat tortured ( & circuitous IMHO ) here's Dave's take on Finding Relative Acoustic Offsets.