Home Theater Forum and Systems banner

1 - 20 of 55 Posts

·
Registered
Joined
·
9 Posts
Discussion Starter #1
Attempting to perform t/a using a UMIK-1 & Rew. I've read probably a 100 post & spent hours researching the how's & why's. While they were extremely helpful, I'm still left a little unsure about the timing of my tweeters. So I'm attaching my file & if any one would be kind enough to take a quick glance & see if everything looks acoustically lined up per the measurements I'e achieved, I would appreciate it very much. And for a quick glance I'll attach some pics as well.
 

Attachments

·
Premium Member
Joined
·
1,562 Posts
I've not seen measurements like this for time alignment. Possibly there is a method, but I can't imagine how it can be done from this data. It does suggest there is no gross timing error from the positions of the impulses.

You swept 1000-1001 Hz for the MR and 3000-3001 Hz for the TW. Normally the sweeps are full range, or at a minimum the width of the XO range for both drivers.

There is a method discussed here long ago with a narrow range sweeps centered on the acoustic XO frequency, but I have too little experience with it to be sure it will work well in all situations.

The file is labeled as 'Truck' so it may be impossible to align MR to TW using my normal method. If you want me to attempt to check your settings new measurements would be needed. I can provide detailed measurement requirements if you identify the drivers and XO's employed.
 

·
Registered
Joined
·
9 Posts
Discussion Starter #3
I've not seen measurements like this for time alignment. Possibly there is a method, but I can't imagine how it can be done from this data. It does suggest there is no gross timing error from the positions of the impulses.

You swept 1000-1001 Hz for the MR and 3000-3001 Hz for the TW. Normally the sweeps are full range, or at a minimum the width of the XO range for both drivers.

There is a method discussed here long ago with a narrow range sweeps centered on the acoustic XO frequency, but I have too little experience with it to be sure it will work well in all situations.

The file is labeled as 'Truck' so it may be impossible to align MR to TW using my normal method. If you want me to attempt to check your settings new measurements would be needed. I can provide detailed measurement requirements if you identify the drivers and XO's employed.
The thread I ended up following for T/A is: http://www.diymobileaudio.com/forum/technical-advanced-car-audio-discussion/317537-measuring-impulse-response-ta-using-rew-umik-1-a.html

I'm attempting to align JBL components for my setup in my vehicle. There a set of 6.5" woofers located in the door panel & a 1" tweeter located in the dash. I can either run a HDMI to my headunit & send signals out that way or run rca's to my MiniDsp C-DSP. If you could provide any details I'd greatly appreciate any feedback eager to learn & if someone like yourself is willing to help. I'm all ears.
 

·
REW Author
Joined
·
6,711 Posts
Unusual approach. For making relative measurements on graphs in REW hold down the Ctrl key then click and drag the right mouse button.
 

·
Premium Member
Joined
·
1,562 Posts
Thanks for the link. I just gave it a quick scan. The logic of this method is still not entirely clear to me. It would seem to align the timing of the MR and TW at 2 different frequencies (1000 and 3000) rather than at the acoustic XO frequency. If those 2 frequencies are chosen the middle of bandpass range of each driver I thinking the timing result may be pretty accurate. I cannot be sure how close it is. In any case, it does provide a way to assure the delays are close to the correct timing.

My normal method has not been very effective in several auto applications due to the strong early reflections so it may not be much use here. A fallback is to get a reasonably close alignment as you have done already and then either just except it or possibly see if it can be fine tuned.

> If you want me to try confirm it is good as is, or suggest an adjustment:
Sweep each of the 4 drivers individually using full range sweeps (minimum sweep range is XO frequency ±2 octaves) with REW acoustic timing activated. Leave your current XO and delay timing in place.
[So, if your XO is set to 2k, set the sweep to a minimum of 500-8000 Hz.]
Sweep:
L-W (left Woofer)
R-W
L-T
R-T

I can work from that info if you post the mdat file.
> You can confirm timing yourself with this same set of measurements along with the 2 additional ones:
L-W + L-T (left channel)
R-W + R-T

For each channel use the REW SPL overlay chart to confirm that the gap in the rolloff of the W and T (XO range) is well filled by the W + T trace.
If it isn't then increase the delay of the W in small increments (~0.03 ms for a 2K XO) and remeasure W + T. Do that for 7 increments (8 different W delays) and chose the delay that provide the best SPL fill in the XO range. It is best to chose a W delay that is the same value for both channels so there may be an compromise W delay choice needed between the 2 channels.​

Below is an example of an overlay chart showing good SPL fill in each of the 3 XO ranges:
 

Attachments

·
Registered
Joined
·
9 Posts
Discussion Starter #6
Perfect! Thanks for your response. I'll get back out & take some more measurements. I see what you're getting at. The method described in that thread generated anomalies when attempting to measure as well. I would get different readings almost every-other measurement requiring to take several measurements & eventually you would see a pattern.
There is also this articl published my MiniDSP, this method was unsuccessful for me as well. https://www.minidsp.com/applications/auto-eq-with-rew/measuring-time-delay
I think if a loopback was used it would work, but as an acoustical reference is being use it produced results that were inconsistent, just like mentioned above.
 

·
Premium Member
Joined
·
1,562 Posts
That's a concern... I must have missed that bit. This is a problem for my normal analysis method.

The first requirement is that we should be able to remeasure a driver several times in a given setup and get the same impulse location on the chart. So, if the impulse position is moving position in repeated measurements it is a problem. You should have selected one of the 2 tweeters for the reference. If a driver's impulse moves using that reference then try using the other tweeter for the reference. If neither tweeter is stable as the reference then it is a problem for any of the common timing methods.

Can you show me several repeated measurements of the better tweeter for stability using it both as the reference and as the measurement driver? Your 'several measurements to see a pattern' method may work if there is only occasional outliers. If you have sorted through all this and are satisfied with your ability to selecting 'good' measurements for the timing analysis then we can just use those good measurements.

If we think it is too random then possibly only a loopback cable and XLR mic setup will work.

> Now thinking of the trial and error process that was the second bullet in Post-5:
That process will still work just fine so long as you feel the initial delay timing you have found is good. If those starting points (delays set) are good as shown in the final chart in Post-1 then the 8 increment process can be done with out even using the acoustic timing. We are just offsetting a known time increment from a known good starting point and checking the impact on SPL. The best XO range SPL fill is the best answer. Acoustic timing on, or off, is not an issue in this method.
 

·
Registered
Joined
·
9 Posts
Discussion Starter #8
@jtalden
I'm attaching my mdat file. I took measurements in the following order.
FRT as ref
FRM
FLM
FLT
I then ran a pass with the FRT+FRM & FLT+FLM.
Note: All t/a settings remained(the t/a settings I came up with using the method I posted to)
All X-over remained in place, I have the TWT's & MID's crossed at 3,000.
Please see if you can work some magic. I still have some level balancing & eq to do, but I want to get my x-overs & t/a done first.
 

Attachments

·
Premium Member
Joined
·
1,562 Posts
I analyzed this 3 kHz XO data:
> I noted that the timing is not the same as the initial charts you posted, but the difference is relatively small and probably the result of a slightly different mic position. Thinking through this, I presume that in many, if not most all, car/truck situations the midrange and tweeters are separated from each other by a significant distance. They are also often separated both vertically and horizontally and the left vs right channel geometry is very different. In this situation it takes only a small mic position difference to have a significant impact on 3 kHz XO timing. [This shows my lack of experience in car/truck setups as I am now just thinking through the impact of all this.]
> My normal timing method is completely unsuitable for this situation. The strong early reflections of the tweeter overwhelms its direct sound.
> I did adjust timing using a couple of different approaches and checked the impact on the SPL. Of course a minor SPL improvement is possible at the exact XO frequency when the timing is corrected for that exact mic location, but the improvement is not uniform over the XO range and is not likely to impact the sound quality.

So my analysis is:
> Your current timing is very good in setting the basic sound arrival time for the 4 drivers at the mic location.
> It is not helpful to fine tune the timing further due to the basic geometric considerations:
1. The 2 ear locations are significantly different from the mic locations regarding timing.
2. It takes only very small head position change to significantly change the effective timing.​

I thus conclude that I cannot suggest an alternate timing that is likely to provide a significant improvement to the current timing.
 

·
Registered
Joined
·
9 Posts
Discussion Starter #10
Thanks jtalden! The centering of the image does sound pretty good. I think what I'm gonna get back out today, do some level matching & adjusting x-overs to see if I'm ready to start eq'ing. Thanks for all your help!
 

·
Registered
Joined
·
22 Posts
Hi jaltan,

Could you help me look at my left and right tweeter measurements? I am trying to see if REW can help me TA my car 3 way active. I will do a full set if that is required. But just want to make sure that my tweeter measurement is correct.

Thanks!
 

Attachments

·
Premium Member
Joined
·
1,562 Posts
I understand correctly you are only asking if the 2 tweeters are delayed properly for sound to arrive at the mic location at the same time?

If so:
> Acoustic timing is on as it should be so this data can be used.
> You have measured 1000-1001 Hz.
[This seems a strange method and frequency for tweeters. This is below the XO frequency for most tweeters and the sweep range is too narrow.]
We can try to use this data anyway:
> For tweeter to tweeter timing alignment (2 identical drivers in different channels) we can just look at either the Impulse chart or the ETC chart.
> For car timing situation the ETC overlay is probably the easier chart to use.

So looking at the ETC chart:
It appears the ETC of L-TW is leading the R-TW by about 2.8 ms.
ETC-1.jpg

Shifting a copy of the L-TW by -2.8 ms (an additional delay) results in better alignment of the peak energy from the 2 TWs.
ETC-2.jpg

> The issues here are:
  • This is at 1000 Hz and we don't know what is happening at any other frequency. The arrival at another frequency can vary greatly in this type of situation.
  • The TW peak energy arrival is influenced by all the bouncing around of the sound that occurs in a typical car application. This may be as good or better then using the Impulse to align speakers in a car application.
  • It would be more common/logical/traditional to align the initial rise of the 2 impulses instead of the 2 ETC's. Doing it that way the needed L-TW delay would be 3.1 ms.
  • Looking at the Impulse chart the polarity of the 2 tweeters appears to be reversed. That is; we would need to invert one of the TWs to agree with the polarity of the other one. The problem is again that this may only be the case at this one frequency. The other frequencies may provide a different story. The ETC chart is no help in polarity determination as it is always plotted as an absolute value.
Suggestions:
  • If the polarity is wired correctly then that is the best indicator. This data is too narrow to really determine polarity with any confidence. The polarity reverses at different frequencies due to reflected sound in the car.
  • If the geometry (difference in measured distances from the Mic to the 2 TW's supports a 2.8-3.1 ms change to the L-TW your delay settings then this is an appropriate adjustment. If not, the measured distance difference is the better method. Your notes seem to indicate the R-TW speaker is currently delayed 1.46 ms which seems backward to me. If the R-TW is reset to 0 and L-TW is delayed the 1.46 ms then that is a difference of 2.92 ms or about the same range as my recommended adjustment. Possibly you just delayed the wrong TW?

If you want to confirm the delay setting it is better to measure as:
> Acoustic timing on.
> XO set on.
> Sweep set full range (or a minimum of XO frequency to 18k Hz).
> For timing analysis it is helpful to have a reasonably smooth frequency response. It is thus helpful to do some rough EQ work on the TW response if it is currently very irregular.
 

·
Registered
Joined
·
22 Posts
Thanks jtalden!

I will try to do a full measurement of Tweeters/Mids/Woofers later. I guess part of the confusion could be that you are thinking this is a LHD car. Its a Right Hand Drive car. So the right tweeter would be delayed more then the left.

I will try to capture all the data with the reference timing on. I seem to be having issues getting it to run once I enable the option. It seems to hang after reaching 19% or I get a waiting for reference signal prompt. So I was hoping that I get a sense of how good or bad my TA is by just capturing the signal without timing reference since the plots seems rather consistent. In the meantime, thanks again.

[/LIST]
[/INDENT]
Suggestions:
  • If the polarity is wired correctly then that is the best indicator. This data is too narrow to really determine polarity with any confidence. The polarity reverses at different frequencies due to reflected sound in the car.
  • If the geometry (difference in measured distances from the Mic to the 2 TW's supports a 2.8-3.1 ms change to the L-TW your delay settings then this is an appropriate adjustment. If not, the measured distance difference is the better method. Your notes seem to indicate the R-TW speaker is currently delayed 1.46 ms which seems backward to me. If the R-TW is reset to 0 and L-TW is delayed the 1.46 ms then that is a difference of 2.92 ms or about the same range as my recommended adjustment. Possibly you just delayed the wrong TW?

If you want to confirm the delay setting it is better to measure as:
> Acoustic timing on.
> XO set on.
> Sweep set full range (or a minimum of XO frequency to 18k Hz).
> For timing analysis it is helpful to have a reasonably smooth frequency response. It is thus helpful to do some rough EQ work on the TW response if it is currently very irregular.
 

·
Registered
Joined
·
22 Posts
Not sure why. I just cant seem to get the reference timing to work. it just hangs at 19%. So I just measured my 3 way active in sequence of LW RW LM RM LT RT in my RHD car. I inverted both my tweeters just to see if it made a difference compared to yesterdays tweeter only measurement.

I am looking at my graphs now to see if it made any difference at my crossover of 3.5k
 

Attachments

·
Premium Member
Joined
·
1,562 Posts
Confused? Me? Hmm, RHD... I'll try turning around to face the other direction and see if that helps me make any better sense of this. :R

2.5 ms + 3ms would then seem to be too much. Maybe you drive a large mobile home... :)

More seriously, I agree there appears to be a problem with the acoustic timing. I will look at he new data to see if it helps us in any way. It may take a couple off day as I am tied up with other matters at the moment.

If you can't get repeatable measurements using either TW for the reference, then there may not be much we can do beyond setting the delay to the measured difference. I would expect that to be a good setting for a car application.
 

·
Registered
Joined
·
1,298 Posts
Hi I am trying to use the Acoustic timing.

Do I leave the "Acoustic Timing" selected in the preferences for all measurements when trying to time align speakers?

Thank you
 

·
Registered
Joined
·
22 Posts
I am still not getting if I am using this function correctly. Please let me know if I am getting this right.

After connecting my headphone/output from my notebook via a 3.5mm to 3,5mm input on my car, I will start REW.

1. Go to Preferences > Analysis. Choose "Acoustic Timing Reference" from the drop down menu

2. Go to Preferences > Soundcard for the "Output Device" choose my soundcard in the drop down menu.

3. For the "Output" choose speaker. The drop down box that contains "Left" "Right" or "Both". I choose both. Is this correct?

4. Make sure Mic is recognized. Close Preferences

Now go to Measurement, and this is where I am also not sure what I should be doing. I can mute any of my 3 way active speakers. The furthers speaker is my left mid (RHD car) but I understand the reference speaker should be a tweeter. So Left tweeter should be the best reference.

So in the measurement, I should use 5k to 12k as the sweep? Speaker output as both and reference speaker as Left. I then mute all except both my tweeter and click on wait for reference speaker?

After this, whats my next step? How do I align say my right mid to the left tweeter and what sweep frequency should I use? I see some instructions which as that we disable all crossovers on the DSP. Is this neccessary?

Thanks!
 

·
Premium Member
Joined
·
1,562 Posts
> The Reference channel can be either channel L or R.
That channel TW has to be active - not muted. This same TW is used as reference for all measurements no matter which measurement channel or driver is being measured. Pick the TW that provides repeatable results as the reference TW. Either one should work in most cases. Assume We select L-TW for discussing below.

> The measurement channel contains the driver(s) we are measuring.

So, when are measuring the L-TW:
Reference = Left
Measurement = Left

So when are measuring the R-TW:
Reference = Left
Measurement = Right

There is never a reason to select 'Both' channel as 'measurement' when doing time alignment work. We would get the both TW's measured together if we did that in this example.
 

·
Premium Member
Joined
·
1,562 Posts
After this, whats my next step? How do I align say my right mid to the left tweeter and what sweep frequency should I use? I see some instructions which as that we disable all crossovers on the DSP. Is this neccessary?
> All XO's should always be active or any measurements are useless. Also, the drivers may be damaged. [XO's impact the timing alignment.]
> I just sweep full range for all measurements. I know of no advantage to reducing the range.
> To time align the L-TW to the L-MR switch the reference the R-TW:
Reference channel = Right
Measurement Channel = Left
> Measure L-TW (mute the left MR and W)
> Measure L-MR (mute the left W and TW)

If one of the TWs is not repeatable as the reference then it is a little harder. We can cover that if needed.
 
1 - 20 of 55 Posts
Top