DiyMobileAudio.com Car Stereo Forum banner

141 - 160 of 208 Posts

·
Registered
Joined
·
5,098 Posts
Put your sub on like a 6th order slope and the mid on a 4th order and measure again. See if there's a difference between 50-500 hz.
 

·
Registered
Joined
·
4,941 Posts
Discussion Starter #142
Group delay and DFR/TDA plots don't display the same thing. I don't know exactly what the difference is... TDA is something else and just like APL can't be compared to anything else I've seen.

I will post a detailed guide how to utilize TDA and how to set T/A properly. Idk, tape measure distance doesn't seem to work very well for a lot of people.


Sent from my iPhone 6 using Tapatalk.
 

·
Registered
Joined
·
1,999 Posts
Guys, is there a significant delay difference in using an analog or digital chain?

I mean, for simplicity and practical reasons I never used the exact same setup for measurements and listening, it was:
Phone > digit > dac > analog > apl > analog > ms8 (listening)
---------------- Laptop > analog > apl > analog > ms8 (measurements)
The difference was only the dac delay (if any)

But now the plan is to use full digital to the C-dsp:
Phone > digit > airport express and/or icon ido > digit > apl > digit > cdsp
------------------------------------------------- laptop > analog > apl > digit > cdsp

So bigger difference in chain, might have more impact, should I bother about that?
 

·
Registered
Joined
·
530 Posts
Put your sub on like a 6th order slope and the mid on a 4th order and measure again. See if there's a difference between 50-500 hz.
I have tried almost all possible combinations due to certain limitations in the present install: TA: 0-336cm - not too much for a sub in a trunk (sound is comming through two old holes in the backshelf).

Filter order: 0-4 (max 24dB/oct).

When the sub polarity (electrical) was turned to 180 degrees, the best match in phase was even more strange: LPF:40Hz, 1.order, HPF: 80Hz, 1.order. Now the sub is back to 0 degrees and therefore - another set of XO points. I need more capacity in TA. Will buy soon miniDSP for this purpose. Just to use as a time machine.

EQ of sub is also not really possible as I have only 2 frequencies in the 5 band PEQ (63 and 125Hz). Maybe another miniDSP will be needed.

Another question is about placement of the mic during delay measurements. I am not sure that during the last set of the tests, the mic was placed exactly at the same point as during the previous one. At that time I have concentrated in the area 100-20000Hz and have managed to find a set of parameters which gave me less than 1,5 ms in all this diapason. But now - after some readings, were authors stated that 1,5-2 cycles (periods) are just fine and are below the threshold for mid and high frequencies. So I was trying to make the LF range as flat as possible for group delay -by rather limited sources. Result: 2,5ms until 40Hz - is it audible?
 

·
Registered
Joined
·
4,941 Posts
Discussion Starter #145
All right, I promised I'd give a walkthrough how to setup a full system. The tape measure method isn't always accurate but can be used as a baseline to start from. There are many methods out there and people swear their method is superior...

This is what I find to be the optimal/easiest way of doing it, I ignore the difference between drivers on the same side to begin with. This is a combined "by ear"/measurement method.


1. First step is to make sure the left and right sides got an equal frequency response, this is paramount, if L/R differs more than 3dB the entire procedure will fail. If you use APL, the software automatically fixes the L/R balance. If you got a normal DSP you need to set it up manually with RoomEQ software or such.

2. Download this: Test Tracks. We're interested in the first track, this is correlated pink noise (mono recording).

3. In your DSP mute all speakers except midbass drivers, keep both drivers on (left and right).

4. Now we need to determine the acoustic center, if you unsure where this is, it's usually slightly left of the physical center of the dash in most cars. Put a piece of tape there or something else to mark this place.

5. Play the correlated pink noise though both speakers (left/right), delay driver side until you hear the noise coming from the tape marker on dash.

6. Now mute midbass drivers and turn on both midrange drivers (in a 3-way). Repeat step 5.

7. Mute midrange drivers, turn on both tweeters. Repeat step 5.

8. Now all drivers are tuned to the acoustic center. Turn on all speakers + sub. Run TDA, place mic in the middle and about 6 inch in front of the headrest. Don't sit in the seat, it's ok to be in the back seat though.

9. Now we should have something to work with, all areas where the speakers don't overlap should be flat (a black straight line). What we want to look for is offsets and other discrepancies around the crossover of the drivers. This means that the speakers are misaligned to each other by some amount.

*** The tweeter crossover is usually place high in frequency, the wavelengths will be small and therefore a very small delay shift is required to bring them back in phase around the crossover. The lower in frequency we go the bigger delay is needed for a given phase shift. There are multiple way of lining up speakers to eachother within a channel, I believe the "magnitude summing method" is easiest but I don't wanna bring this channel off topic***

10. Now that we already set L/R coherency and just want to set the delay within a channel (the delays between all left OR right speakers), it's important that we don't change the L/R relation delay numbers, we only want to offset it. Write down the actual delay between midbass drivers, between midrange drivers, between tweeters. This number might be offset, but not changed.

In other words left midrange might have 1,5ms delay and right midrange 0ms delay. If you set 2ms delay on left and 0,5 on right, there is still 1,5ms delay between these two drivers and the center will not shift. This is what I refer to as an offset.



---------------------------------------------------------------------


1. After running the sweep in TDA. It might look something this now;



Here we see that midranges are slightly delayed vs tweeters and the midbass is about 6ms away from midranges. DFR will show the numbers with precision. In this situation we want to delay midranges 6ms and tweeters 6,5ms.

2. Enter numbers + the relative delay between L/R in your DSP and do another sweep.



Should look something like this now. The shift at 50Hz are due to subwoofer group delay, that is a whole other topic...

3. Done!
 

·
Premium Member
Joined
·
1,148 Posts
Rhhaaa demo expired! :/


Yeah mine did too. I was pretty bummed. I even tried uninstalling, re-downloading, and reinstalling but it still recognizes my computer and says demo expired.
 

·
Registered
Joined
·
711 Posts
I played with TDA and my car a bit today. I was able to get this:



My system isn't complete yet, I'm missing midbass drivers. At the moment the crossover between my mids and subs is 200hz @48db. Once I have the midbass drivers in I'll only take the mids down to the 350-450 area. The horn to mid crossover at 1200hz @48db.
 

·
Registered
Joined
·
530 Posts
Yeah mine did too. I was pretty bummed. I even tried uninstalling, re-downloading, and reinstalling but it still recognizes my computer and says demo expired.
TDA is exceptional tool and it worth buying. Besides it is all the time Boeing upgraded and you might be able to use the same license for different program modifications.
 

·
Registered
Joined
·
5,666 Posts
TDA is exceptional tool and it worth buying. Besides it is all the time Boeing upgraded and you might be able to use the same license for different program modifications.

Worth buying for how much? Saw the pricing in pounds and got frightened and left. Hehe. Is there a Diyma student diy'er discount?
 

·
Registered
Joined
·
530 Posts
Worth buying for how much? Saw the pricing in pounds and got frightened and left. Hehe. Is there a Diyma student diy'er discount?
Yeah, savings will be forgotten soon, but this program will serve you a long time ahead.

One of the advantages is you could run both front sides together and to see how they play simultaniously. For doing this one should select "mono" as the output signal. This feature had helped me to optimize the TA settings in order to avoid mutual cancelling by respective drivers.

Another "phat" feature - final table with the data measured, which might be derived from the graphs.

Most probably other consumers will find something else, which will be interesting in this soft, as distortion levels, delay FR, 3D presentation of the test results etc.

Recently I have used the TDA to check how the APL makes time alignment. With the purpose I have delayed one of the front channels by 15ms (need more actually in order to connect the front to the subwoofer). Single test with both fronts running has proved that delay made by APL is quite accurate and corresponds to the desired value.

My next little project with the TDA will be determination of the car transfer function, for which I intend to use AFR function. Interesting to exchange some experience with those who tried it before. APL Worshop could be the alternative solution for this, maybe little bit more complicated, but also more precise.

What do you think guys?
 

·
Registered
Joined
·
530 Posts
TDA is awesome :D

Great curve! What is the scale range on the X-axis?

I was not able to get so clean response with my passive front and the subwoofer playing from the trunk through two 17cm holes in the back shelf (sedan). It means that the whole trunk plays as a ported enclosure with uncontrolled ports, similar to unpredicted bandpass.
 

·
Registered
Joined
·
1,999 Posts
My next little project with the TDA will be determination of the car transfer function, for which I intend to use AFR function. Interesting to exchange some experience with those who tried it before. APL Worshop could be the alternative solution for this, maybe little bit more complicated, but also more precise.

What do you think guys?
I suspect APL would be the right tool there no? Since you'll need many many many points. Or what else did you have in mind?



Great curve! What is the scale range on the X-axis?

I was not able to get so clean response with my passive front and the subwoofer playing from the trunk through two 17cm holes in the back shelf (sedan). It means that the whole trunk plays as a ported enclosure with uncontrolled ports, similar to unpredicted bandpass.
Probably the default 30ms.
I usually start with 30, then go to 10ms, with sometime -1 as a reference.
This was left or both I don't remember. Left is always easy to get clean, right not so.
For sub and woofer I actually quite like the ease of use (biggest marge of errors on ms), but for midranges and tweeters I have hard time to not mess everything up!
 
141 - 160 of 208 Posts
Top