How to use acoustic timing reference exactly??

[mamba76]

Hi. Been using REW for years now and always had loop back on my external Audiolink soundcard. Iv gone to a digital ouput now so can no longer use loopback.
How exactly do you use the acoustic timing reference? Do you take your first measurement from a tweeter with 'acoustic timing reference' active then measure (in my case) the other tweeter then the 2 door mid's without acoustic timing reference then check it out in overlays- impulse?
Im questioning the accuracy of my original loop back and think the acoustic timing reference may prove more accurate. I used to sweep upto 3000khz, not 20k when aligning tweets/mids as this is the crossover point and as far as I learnt the impulse response is only good for the END frequency of the sweep??

Id be glad for some clarification.

Thanks

 

[John Mulcahy]

Whatever you are using as the timing reference needs to be present for all your sweeps (and in the same position relative to the mic). For example, if left was the timing reference then it remains the reference whether you are measuring left or right. Full range sweeps are best for the acoustic timing reference.

 

[mamba76]

ok, I think I get it. So for instance if I leave the mic where I position it at my left ear do a sweep on left tweeter with acoustic timing on, then (without touching mic) do a sweep of the mid range only (timing reference can't be used on this driver because they roll off at 3khz and the signal is too quiet) then do a sweep of midbass speaker, again without moving the mic (acoustic timing reference obviously wont be heard on midbass) then I can align the spikes in impulse response? Then basically move the mic and repeat procedure for the other side?
Thanks

 

[jtalden]

That isn't a procedure that used for most objectives so please confirm what you are trying accomplish. Is the objective to determine the delay timing needed between tweeter and midrange drivers and also between midrange and midbass drivers in the same channel?

Are all 3 drivers located on the same baffle, or possibly the midbass is separated in a different box? Are the XO filters chosen and active? ​

I can probably help with your questions if understand the situation.

 

[mamba76]

Hi. Yeah I should have give more details but you guessed right. The tweets are in the A pillars, mid range in the stock front door locations and mid bass are in the rear door stock locations. '06 Audi A4 Avante. Including a 13 1/2 inch sub in the spare tyre location my system is 7way fully active using minidsp CDSP6x8. Midbass is JL audio ZR800 front components focal Utopia (2003 model TN51 tweeter). I will update my profile with full details soon. Im using these very capable midbass just to fill a small niche between sub and front components so unlike a typical 3 way set my midbass crossover lower about 140hz and 55hz on the low end, but it works really well. I have these in sealed 10l sealed enclosures s in the rear doors!
I always sit in drivers seat and have a ecm8000 mic on a swivel behind your neck (headrest removed) so you can swing from left to right ear. I want to align the 3 drivers at the left and right ear respectively. Im sure its set up fairly well but would be nice to confirm with an impulse response.

Usually I look on REW where a pair of drivers crossover play that tone on generator, invert one of the drivers and change time till its quietest, reverse it and your done.

I might as well add some pics of my install as well soon..

 

[jtalden]

There are too many reflections in an auto setup for my normal timing alignment procedure. I have no practical experience with auto setups and have not studied best practices for that application so my thoughts should be taken accordingly.

Using the impulse

to time align ​

in that application is very problematic. I suspect that your current procedure may be more suitable.

That said, in general, to align the 4 drivers in the L channel via Impulse location the mic is placed at the LP. The R channel is used as the reference channel. The R tweeter then will play the timing reference and the L channel will play the captured sweep for each driver. The sweep must be full range for all measurements of the 4 drivers in the L channel for this to work properly. The initial rise of the impulse of each driver provides a close indication the proper timing. A small timing adjustment from there to maximize the SPL in the XO range may be needed.

Narrow sweeps in just in the center of the XO range may work as well (or possibly better) in this case.

Similarly, the R channel driver alignment process uses the L channel as the reference. After each channel drivers are timed then the needed delay timing between the 2 channels can be determined.

 

[mamba76]

Ah I think I get it now. So I can play reference on left tweet/channel and play midbass only on right side at same time then play ref +right mid range then ref + right tweet, look at impulse and adjust accordingly. That makes sense now, was so used to using the loop back I never took it in properly.
I will experiment with narrow sweeps as results will differ depending on where the sweep ends.
Thanks for the help, much appreciated!

 

[jtalden]

I don't follow. REW plays the acoustic timing sound first and then continues on to play the sweep for the measurement.​

Have you tried making measurements yet?

To align the drivers in the left channel:

• Set the Right channel in REW as the measurement channel.​
• Set the right channel in REW as the reference with the right TW activated. [This will play the reference sound in the right channel TW each time a measurement is taken.]
• Set 'wait for timing reference' in the 'measure' popup. [This will let REW use the reference sound from the right TW to time the start of the sweep capture of the active driver in left channel.]
• Now take the following measurements of the left channel drivers.

1. TW
2. MR
3. MB
4. W

Those 4 measurements will have the same relative reference time as a result of using the right TW sound to time each measurement captuere. The 4 measurements can thus be used to determine how much the driver delay change is needed to provide good XO handoffs.

After the left channel delays are adjusted, then the right channel drivers can be measured and delay changes determined. To do that the left channel is used as the reference channel for the 4 measurements.

After each channel's drivers are properly timed then the delay timing needed between the 2 channels can be determined. To do that set the left channel as the reference channel and take a measurement of both channels. The overlay of the impulse chart will show the current difference in arrival time.

Note:
It's advisable to first confirm the acoustic timing setup is properly working. Set the right channel as the reference channel and left channel as measurement channel. Take ~5 measurements of the left tweeter. Use the impulse overlay chart to confirm the 5 TW measurements all fall on top of each other. Any significant spread in them indicates there is a problem.

 

[mamba76]

Thx Jt. I will try and have a go today, was way to hot in garage yesterday!
I use two measurement points as said before I think there are slight differences in the approach to tuning . I am using my ears a lot more these days to tune but REW is still essential Id say its 50/50. Its all about relating what you hear with what you see on graphs.
I know my cars audio very well after 2 years and am critical of any changes. I have almost nailed my process its just I lost the impulse tool when I went digital. To be honest I really only like the impulse ability for aligning the tweeters. 2 reasons:- 1) to see when the left tweeter arrives at left ear and when right tweeter arrives at right ear 2) to see when left tweeter arrives at left ear compared to left midrange at left ear (this is why I used a sweep up to 3khz-Xover point) Once this gets me in the ball park I can fine tune by ear. Ie a tambourine on loop may help align left tweeter and left midrange.
For reason 1) I think I need an absolute timing reference not a relative one.
Our tuning methods may be a bit different. Most people, like I use to, put the mic in center head position and are not sitting in the car when taking measurements. I always sit in the car as its more accurate. Then you have 2 ears. You have SPL and timing ques received by each ear so I found matching what is received at each ear location a better approach. If I could sit in the car and replace my head with a 3dio mic I probably would! But is was the concept of using acoustic timing reference I couldn't get my head round at first but I understand better now.

 

[jtalden]

Just to stress again concerning the delay between the TW's. We are only interested in the relative arrival time from the two tweeters. Both loopback and acoustic timing reference provide that. The absolute time of arrival is of no importance in determining the delay offset between the 2 channels. For acoustic timing the same TW must be used as reference for both TW measurements.

Your process may indeed work better than the one I outlined for your application. Mine was just intended to help you understand by giving you one possible approach in more detail.

Good listening!

 

[mamba76]

Hey you explained it spot on, I now have correctly timed left TW with left MR and right TW with right MR. I see what you mean about making 5 sweeps as sometimes it has a fit and is out like a whole second so if 2 or 3 in same place (give or take a few us) keep that and delete the rest, same with the speaker your measuring against.
I already knew for certain where the MR image is center so timing TW to these was the issue. Its the bass from the mid range (<500hz) that dictates where the centre image should be, everything above is relying on intensity more so MR only really have one place, these I always do first and they are the hardest (to put it mildly) because, like you said reflections etc.

What do you think of these 2 pairs of impulse responses?

I fine tune it with just TW & MR on one side at a time using white noise you can hear the image drift towards front of the car. About 0.3ms brings the TW round again to sound the same but with more/less delay relative to MR. How would you align the peaks in these responses? (sweeps upto 3khz)
Thanks

 

[jtalden]

Your current timing seems reasonable just based on the locations of the IRs. They are not spread enough apart to suggest there is a clear error. It does hint that the TW is delayed an little more than ideal, but that cannot be confirm with this data.

The comments below are assuming the XO was engaged for the all measurements. This is a requirement for delay analysis.

Per the SPL chart t

he left TW is about 6 dB lower than the left MR. That indicates that ​

3 kHz is below the acoustic XO point. I would normally want to see a minimum of 2 octaves below and above the acoustic XO frequency for both drivers. If the acoustic XO is actually 3.4 kHz the minimum range needed would be 850-10.2k Hz. That would be enough to support my normal method of phase tracking analysis. As I indicated however that process doesn't work reliably in an automotive application anyway.

Given just this data I don't know of any reliable way to perform an analysis, but I tried to do that anyway. Below is a best guess based on

the left TW-MR XO​

.

I looked at it 2 ways and they both seems to agree. Assuming the XO frequency is actually very close to the 3 kHz provided here we can:

1. Look at the lower half of the XO for phase tracking and predict the upper half by based on experience.
2. Align
   timing based on ​
   the ETC as the proximate center of the XO is the measurement cutoff. The initial energy peak is the highest frequency that is measured so that is 3kHz for both drivers. The initial peaks should thus be aligned. [Just my own expectation/understanding.]

The finding:

• Reverse the polarity of the TW and reduce the TW delay by 0.6 ms.

I did not analyze the right channel in as much detail as the left, but it appears to benefit from the same change.

I suggest that your method is likely as good, or possibly more likely, to work well given your experience. I am not suggesting you should evaluate this change or that there is a high probably that it would be an improvement. It's likely that the SPL support and sound quality would not be significantly improved. It was more for my own interest to see if anything could be derived form this data.

Some charts:

 

[mamba76]

Thats definitely food for thought. The ETC alignment I'm interested to try seems much more precise and ultimately easier. Iv not delved into the other tools in REW as much as Id like. 2 years ago phase was new to me! but I wonder if ETC would work for MB xover at 130hz.

My latest EQing session has taken about 3 weeks as Iv been creating custom Targets and Measurements for each driver and tweaked my overall Target/housecurve yet again. Maximising input gains this time, not boosting more than 3db in EQ, just cuts. Without going into too much detail its the best Iv heard it yet. A tad hot somewhere in the TW, vocals are very sweet and so is the bass although Im not sure yet if the low bass decay is too short. Songs like Cormega-beautiful mind or straight outta compton-NWA have a purposful long decay on the kick. All Xovers are BW24 except MR/TW are LW24 (manufacturer spec) and sub high xover is BW48. I think you agree that the acoustic xover is what matters most.

Of course inverting the tweeters gives me that extra room for alignment didn't think there'd be a use for it at high frequencies but there is here too. My goal is just to optimize what I have installed in my car then maybe Ill be happy and can do something else!

Iv attached a review of my latest work if you want to have a look. I had problems uploading zip files so this is 29mb. The first 4 are target curves for each driver. I have matching house curves for these too but just included the overall one.The next 3 are L ear / R ear and average(all speakers playing) which I use to judge overall FR. The rest are individual responses (4 at L ear and 4 at right) all using verified acoustic timing method.

 

[jtalden]

Just to clarify:
I have not specifically investigated using the ETC for timing, but I expect it to work okay given the correct setup. It will not indicate the best polarity setting however as the phase relationship between the drivers is not apparent in that chart.

The recommendation to invert the TW was based on the phase tracking analysis of the most direct sound from the 2 drivers. You are correct that with a high frequency XO the driver polarity relationship is likely to be trivial relative to sound quality. In an auto environment with all the additional reflections, I would not even expect the SPL support at the XO to be significantly different with that change. Using ETC In a room setup at lower XO points I would expect it to be necessary to fine tune the delay timing or polarity setting to assure there is good SPL support through the XO range.

The file shows you have good timing and SPL is following your target cure very well. Well done!

 

[mamba76]

Thanks Jt. I will get round to trying the ETC alignment soon. Please have a look at my new thread. Im gaining some skills here just got to put everything together!

 

[Matthew J Poes]

If you can post a response measurement of the entire system, not the individual drivers separately, I can show you a couple methods of assessing the time alignment of the drivers in each channel as well as relative left to right.

The 3 sweep averages you included will not work as the timing information is missing that I need. Just send the raw data.

 

[Vladimir_1953]

What ended the system setup?
What was the result?

 

[Yan]

If you can post a response measurement of the entire system, not the individual drivers separately, I can show you a couple methods of assessing the time alignment of the drivers in each channel as well as relative left to right.

The 3 sweep averages you included will not work as the timing information is missing that I need. Just send the raw data.

HI, I'm in the process of tweaking my speakers and room.

I'm having a hard time tofind a way to timealign my speakers drivers. (Horn + Woofer).Your method would be appreciate.

I attached my measurments : Both speakers + each speakers separatly.

Thanks

 

[hbap349]

hi members,

Interesting thread to me here but I've tried it in my car but failed to do so. I wish any experienced members here could further guide me in this matter.

My vehicle is right hand drive. In my case, Left midbass (passenger door) is the furthest speaker so i choose that as timing reference output. I set the L channel in REW as timing ref output. Attached was my setting make to DSP IO for the process (i not sure if i am doing right).

"wait for timing reference" was check in measure pop up dialog box. I started with the right Mid Range so i muted all other driver in helix dsp except for the left mid bass and right mid range. i also set my start freq from 1000hz and end 1001hz.

my first problem is i didn't hear any reference timing sound come out from the reference channel before the sweep sound on right mid range.
second problem was at some point, it just stopped at around 19% and indicated ....waiting for timing reference.... (i have to cancel)

i not sure what was set wrong.

I am using UMIK-1 usb connection from laptop and laptop bluetooth to car player for the REW measure process.

 

[John Mulcahy]

"wait for timing reference" was check in measure pop up dialog box. I started with the right Mid Range so i muted all other driver in helix dsp except for the left mid bass and right mid range. i also set my start freq from 1000hz and end 1001hz.

The timing reference is a high frequency sweep from 5 kHz, the reference speaker needs to be able to reproduce that. Why would you configure a sweep with a result covering just 1 Hz?

 

[Matthew J Poes]

As John noted, what you are doing doesn’t sound like it will work. You should sweep a much wider area. If you want to look at a 1khz pure tone, don’t use sweeps. That is a different test.

for using timing reference, sometimes DSP devices have muting circuits that cause the timing reference to be missed. I also find some devices simply won’t play it. The fix is often to change the buffer size in the output to the largest possible value. I find that helps with such devices.

it is also important the reference can actually reproduce the timing beep. So I think in your case it needs to be a tweeter if the car is fully active.

 

[hbap349]

Thanks John & Matthew for your input. I am in the learning stage, been reading from other forum about this but i can't make it happen to my system, hence i decided to write here.

Yes, i truly didn't get the timing beep from the reference speaker. May i ask, in doing time delay process, do we not take the furthest speaker to be the timing reference ?

at what sweep frequency range should i set to get it correct? was my DSP IO setting correct for this process?

 

[John Mulcahy]

You said you chose left midbass as the timing reference. The timing reference signal starts at 5 kHz, it does not have any content a midbass driver can reproduce. It has to go to a tweeter or a speaker that includes a tweeter.

 

[hbap349]

During my first few attempt, I did bypass the low pass filter on the midbass output channels in DSP. I thought the midbass supposingly able to produce 5kHz?. I will definitely try it to the tweeter accordingly in my next tune session.

 

[Matthew J Poes]

Thanks John & Matthew for your input. I am in the learning stage, been reading from other forum about this but i can't make it happen to my system, hence i decided to write here.

Yes, i truly didn't get the timing beep from the reference speaker. May i ask, in doing time delay process, do we not take the furthest speaker to be the timing reference ?

at what sweep frequency range should i set to get it correct? was my DSP IO setting correct for this process?

The timing reference just provides a point of reference for REW to measure the start of the impulse against. I am not really sure why it would matter how far away the timing reference is. Maybe John could speak to how critical that is.

not seems odd to me that it would “need” to be the farthest. The beep is replacing a loopback as a timing reference. A loopback is the opposite, it’s the closest possible signal. It’s immediate. There is no fly time since it’s a direct connection.