DiyMobileAudio.com Car Stereo Forum banner

861 - 880 of 928 Posts

·
Premium Member
Joined
·
2,418 Posts
Is there a reason why a tolerance of 60 degrees is acceptable?


Sent from my iPhone using Tapatalk


That’s just a generally accepted approximation that is used, based on how the physics work. Think of phase coherence like trying to read newspaper with a magnifying glass. In the center of your magnifying glass the words are in focus. As you move to the sides they get blurry but still legible. Outside the glass they get blurry and you can’t read them.

When we use raw delay to correct phase between drivers, it behaves similar to the magnifying glass. We can get good coherence over a certain range but not everywhere. The goal is to put the center of that magnifying glass over the middle of the crossing region, and get the sides to be just enough in focus to still read them.

Make sense?


Sent from my iPhone using Tapatalk
 

·
Premium Member
Joined
·
220 Posts
Thanks for the replys on the other post oabeieo, figured this is prob the best thread for the technical discussions...

So Dirac can boost no more than 10db.
So this sounded concerning to me as i have read here so many times you really dont want to boost much, if any, and when you need to only a few dbs.

From a strictly SQ standpoint, I just keep my target below the “0db line” and try to stay close to the bottom of my response. That will ensure the inversion will be all cuts. Like setting rew not to boost anything.
Then reading this, it sounds like if want to mostly cut and not boost much this is the way to do it! :D

Also I read on here before when someone asked but I could not find the answer, where can we see the corrections Dirac applies? I am using DDRC-24 as that's all I have, but the only thing I can see is the TA delay and gain on the Ch. 1-2 Inputs. Is it possible to see this, or is it "hidden" from the user? :(

Last question for now... Using the DDRC-24, I have Ch 1-2 out to L/R and Ch 4 out to SUB (Computer Desk setup), and I'm still working through various testing and setting scenarios. On my last tune, I noticed a tiny bit of delay for example on Ch-1 and a tiny bit of gain on Ch-2. However, that is going to impact my SUB since its using L/R summed for its signal, right? Is this correct, or should I change the setup? I know its a 2-ch Dirac, but it should still properly work with SUB.
 

·
Registered
Joined
·
4,167 Posts
Discussion Starter #863 (Edited)
subtwrFUSE is right , and I love his analogy.

Raw delay can only be used to align phase to a HPF it dosent work for a low pass. Meaning, you can use raw delay to align a sub to a midbass by changing the subs delay and that’s about it.

You can also use delay to line up the lowpass of a midbass to the highpass of a midrange or tweeter than use delay to align the sub to the midbass (in a 3 way)
 

·
Registered
Joined
·
4,167 Posts
Discussion Starter #864
Thanks for the replys on the other post oabeieo, figured this is prob the best thread for the technical discussions...

So this sounded concerning to me as i have read here so many times you really dont want to boost much, if any, and when you need to only a few dbs.



Then reading this, it sounds like if want to mostly cut and not boost much this is the way to do it! :D

Also I read on here before when someone asked but I could not find the answer, where can we see the corrections Dirac applies? I am using DDRC-24 as that's all I have, but the only thing I can see is the TA delay and gain on the Ch. 1-2 Inputs. Is it possible to see this, or is it "hidden" from the user? :(

Last question for now... Using the DDRC-24, I have Ch 1-2 out to L/R and Ch 4 out to SUB (Computer Desk setup), and I'm still working through various testing and setting scenarios. On my last tune, I noticed a tiny bit of delay for example on Ch-1 and a tiny bit of gain on Ch-2. However, that is going to impact my SUB since its using L/R summed for its signal, right? Is this correct, or should I change the setup? I know its a 2-ch Dirac, but it should still properly work with SUB.

It shouldn’t impact the sub that much. But it definitely could do a tiny bit. The sub will still sum both sides and play what it’s playing

When you run your measurements on the input matrix try to turn on all 4 checkboxes for sub so left and right RCAs play left and right simultaneously. That should correct any errors as far as the sub goes .


So 10db of boost. It can cut as much as needed.

If the entire target has let’s say 25db variance and the sub section is above the 0db line it will push down the levels. But it definitely depends on the response against it.

The variance in the target could also be considered boat if it’s extreme enough.


If you stay below the 0db line and put your correction at the bottom of the response do it only cuts and than it’s a non issue

It’s okay tho to be above the response a little when passing over the big dips
OTOH if the dips are that big it might be better to take the gain losses and put the target down even lower. You’ll know if it dosent sound good up high


Or make a v shape in the target over the big dips. That also sounds good and saves gain
 

·
Premium Member
Joined
·
97 Posts
Thanks for making this video. It answered some of the q's I had.

I just watched this video as well. Thanks for making it.

U used the forward mic with the 0 degree mic file in this video. I recall that is what Minidsp instructions were to use. However, somewhere in this thread, many mentioned that you get better results with the 90 degree mic file and pointing the mic upwards.... Which is better and whats the difference between both for the end result?

I also noted that you stated in the video that you preferred the sofa setting measurement instead of the chair. Is there actually a difference in the algorithm or its just a visual show of the points for measurements?
 

·
Registered
Joined
·
4,167 Posts
Discussion Starter #867
I just watched this video as well. Thanks for making it.

U used the forward mic with the 0 degree mic file in this video. I recall that is what Minidsp instructions were to use. However, somewhere in this thread, many mentioned that you get better results with the 90 degree mic file and pointing the mic upwards.... Which is better and whats the difference between both for the end result?

I also noted that you stated in the video that you preferred the sofa setting measurement instead of the chair. Is there actually a difference in the algorithm or its just a visual show of the points for measurements?

So in Dirac 1.7 the algo is the same. It’s just different measurements points as suggesting good points if that was your seating arrangement.

If you do chair on sofa you get chair and vise versa.

So basically what I was saying is do the chair measurements on the left side and part chair part sofa on the right .

Meaning, experiment with your car a little, but definitely try getting one or two measurements in the passenger area.

I like chair on Left and on right do the chair “box” except the top forward measurement point I put where passenger seat is .

Getting at least one measurement point helps. You can hear what’s going on over there indirectly. Especially if the passenger side has any horn loading or off axis behavior that can affect the stage all over.

When it was first released they said use the 90deg file. The general consensus among users is that the normal 0deg works better. You want the direct sound as much as possible. That way Dirac can differentiate between reflected sound vs direct sound. It’s only the last oactave that it makes a difference. Do what you like , it will still work with either mostly the same.
 

·
Registered
Joined
·
4,167 Posts
Discussion Starter #868
in that video I also talk about moving the mic forward or back to get the center right by adding or decrease path length difference.

I also want to say, depending if speakers are low or high mounted moving the mic lower can also add PLd and the correct phase information needed to get that center nailed down.


If your speakers are low in the doors by moving the mic down will also add PLD and move the center to the right.

Just don’t move the mic to the left or right....(yeah it’s not good). Keep it parallel with the physical center of your body , just try different spots around your head or chest until you get the center. You will find it eventually and it will work every time.
 

·
Registered
Joined
·
28 Posts
So in Dirac 1.7 the algo is the same. It’s just different measurements points as suggesting good points if that was your seating arrangement.

If you do chair on sofa you get chair and vise versa.

So basically what I was saying is do the chair measurements on the left side and part chair part sofa on the right .

Meaning, experiment with your car a little, but definitely try getting one or two measurements in the passenger area.

I like chair on Left and on right do the chair “box” except the top forward measurement point I put where passenger seat is .

Getting at least one measurement point helps. You can hear what’s going on over there indirectly. Especially if the passenger side has any horn loading or off axis behavior that can affect the stage all over.

When it was first released they said use the 90deg file. The general consensus among users is that the normal 0deg works better. You want the direct sound as much as possible. That way Dirac can differentiate between reflected sound vs direct sound. It’s only the last oactave that it makes a difference. Do what you like , it will still work with either mostly the same.
Quick Question - in your video you had a front Left Right setup but what would you do if you have rear fill? still keep the mike pointed to the front with a 0 degree calibration file? BTW - like the others have stated - appreciate you taking the time to make the VID
 

·
Registered
Joined
·
4,167 Posts
Discussion Starter #870
Quick Question - in your video you had a front Left Right setup but what would you do if you have rear fill? still keep the mike pointed to the front with a 0 degree calibration file? BTW - like the others have stated - appreciate you taking the time to make the VID

Try it .

On the systems I did with rears I was doing a front calibration simultaneously so I used normal 0deg

But definitely try it .

It works fine for me as I attenuated the HF anyways on rears
 

·
Registered
Joined
·
2,411 Posts
Got a question for some of the experts here, I have gotten a preset set up now that I am happy enough with, until I turn the volume up past about 25 on my head unit (loudest I typically go is 30, think the head unit goes to 40). Anyway, at that point some vocals and instruments become painfully harsh.

The problem I am having is, that I can't see anything obvious when looking at the RTA that tells me where that harshness is coming from.



I took that at volume 30, so the harshness with music would be as bad as it gets. Nothing really obvious jumps out at me, the only thing I see that could be the problem is around 1.5-2KHz there is about a 2dB peak. Assuming that is the root of the problem, can I use PEQ after Dirac to knock it down? If I do that, does it not then mess with the overall Dirac corrections?

If I look in DLCT, the measured response is above my target curve, so it should be pulling it down anyway. So, I can't remove the peak before running Dirac since it does not exist.
 

·
Registered
Joined
·
728 Posts
Take down a lot of 2-5khz

Like 5db worth


Sent from my iPhone using Tapatalk
 

·
Registered
Joined
·
3,275 Posts
Got a question for some of the experts here, I have gotten a preset set up now that I am happy enough with, until I turn the volume up past about 25 on my head unit (loudest I typically go is 30, think the head unit goes to 40). Anyway, at that point some vocals and instruments become painfully harsh.



The problem I am having is, that I can't see anything obvious when looking at the RTA that tells me where that harshness is coming from.







I took that at volume 30, so the harshness with music would be as bad as it gets. Nothing really obvious jumps out at me, the only thing I see that could be the problem is around 1.5-2KHz there is about a 2dB peak. Assuming that is the root of the problem, can I use PEQ after Dirac to knock it down? If I do that, does it not then mess with the overall Dirac corrections?



If I look in DLCT, the measured response is above my target curve, so it should be pulling it down anyway. So, I can't remove the peak before running Dirac since it does not exist.
Look at distortion charts for any spikes.

Is there anything DL is boosting

Sent from my XT1710-02 using Tapatalk
 

·
Registered
Joined
·
2,411 Posts
Take a look at the decay times via the REW waterfall graph. Take one at a good volume and at volume 30 and comparing those would be a good start.

https://www.gikacoustics.com/understanding-decay-times/
Thanks, will have to figure out how to read them, but at least just comparing at say volume 20 and volume 30 may be a good idea and might show something obvious.

Is there anything DL is boosting
Just looked at DL and there is a dip on the right side, almost exactly at 1.5dB that Dirac is boosting back up to the target. The dip is not there on the left, only on the right. There are a couple other smaller dips it is likely boosting as well, those are up around 7KHz. Going to take a measurement of the right side mid only, see if I am remove the dip before DLCT and then see if that makes a difference.


Edit: Quick update, went out to the garage and measured the right mid with REW, no dip shows up on the mid with REW, so there was nothing to fix. Played both left and right mids together, still no dip like shows up in DLCT. While out there I added a high shelf filter to the tweeters at 7.5KHz, Q of 0.9 and -4dB to match where the dip around 7KHz is (it starts at around 5KHz, pretty consistently until 7KHz).

Back into DLCT and took a new set of measurements, this time there is no dip at 1.5KHz, not sure if before it was a weird measurement or what exactly, but it's not there now, so DLCT is not having to boost it. The dip from 5-7KHz is also gone, so no need for it to boost there either. I ended up cutting my entire target curve down by 2dB, meant I lost a little overall volume, but the entire target curve is now above or pretty much on my measured response.

A quick listen, and I really do mean quick, the harshness appears to be gone. I'll have a better idea in a couple days, but so far so good. Annoyingly, the center sounds like it may be just right of center. It's an easy fix, but I am not going to mess with it just yet, have a head cold and all kinds of congestion so it could be just fine. So, fingers crossed it's resolved. I have 4 presets saved now, all just slight variations on each other, assuming the harshness is gone from this latest one, then I'll move it to slot 1 (from slot 4) and use the other 3 to tinker around with different settings, configurations etc. I'm hosting a meet here next weekend, so will be good to have some others listen and give me some feedback as well.
 

·
Hillbilly
Joined
·
173 Posts
My thinking is that boost... and you don't have enough slop from about 2K up... as tonynca stated... pull it on down. That 2-5kHz range can be brutal on the ears. You may have to pull it on down even more... say start at about 1kHz with a 2db cut and slope it down out to 10kHz with a 7-8db cut. I have to do this same thing even in my home theater, and with my dash speakers firing straight up into the windshield... mine looked similar to yours and ended up having to pull it down more.
 

·
Registered
Joined
·
728 Posts
My thinking is that boost... and you don't have enough slop from about 2K up... as tonynca stated... pull it on down. That 2-5kHz range can be brutal on the ears. You may have to pull it on down even more... say start at about 1kHz with a 2db cut and slope it down out to 10kHz with a 7-8db cut. I have to do this same thing even in my home theater, and with my dash speakers firing straight up into the windshield... mine looked similar to yours and ended up having to pull it down more.


Yeah the reflections in the car are just too powerful with all the glass and hard surfaces. Those freqs will hurt.

From what I noticed 7khz and up tends to be a bit less harsh and if you like that sparkle boost 10khz and up to get that silky sparkle HiFi sound.

But people should start rolling off starting at 1khz and riding the slope back up from 5-6khz.


Sent from my iPhone using Tapatalk
 

·
Registered
Joined
·
1,273 Posts
Got a question for some of the experts here, I have gotten a preset set up now that I am happy enough with, until I turn the volume up past about 25 on my head unit (loudest I typically go is 30, think the head unit goes to 40). Anyway, at that point some vocals and instruments become painfully harsh.



The problem I am having is, that I can't see anything obvious when looking at the RTA that tells me where that harshness is coming from.







I took that at volume 30, so the harshness with music would be as bad as it gets. Nothing really obvious jumps out at me, the only thing I see that could be the problem is around 1.5-2KHz there is about a 2dB peak. Assuming that is the root of the problem, can I use PEQ after Dirac to knock it down? If I do that, does it not then mess with the overall Dirac corrections?



If I look in DLCT, the measured response is above my target curve, so it should be pulling it down anyway. So, I can't remove the peak before running Dirac since it does not exist.
From my experience with the standard 8x12, those maximums (max input 4v, maximums displayed in the vu meters, 0 db levels, 4v output) are hard limits, they are absolute, exceeding any of those by the tiniest amount will result in digital clipping. If I need boost, I reduce the level by an equal amount before that boost is applied. Just make sure you never exceed 0 db anywhere, or you'll get that hard to identify digital clipping. Oh, by the way, I think that harshness is digital clipping.
 

·
Registered
Joined
·
2,411 Posts
I'll listen to it for a few days and then maybe change things around a little. I don't have anything over 0dB, master volume at -3dB, most speakers are at around -5dB on the output sliders etc.

On another preset there is no harshness, I think on that one though the target curve does drop more after about 2khz, so we'll see. Listened a little earlier but only quietly as the kids were in the car with me and I don't want to damage their hearing!! I don't want to damage mine either, but really don't want to damage theirs.
 

·
Registered
Joined
·
1,273 Posts
I'll listen to it for a few days and then maybe change things around a little. I don't have anything over 0dB, master volume at -3dB, most speakers are at around -5dB on the output sliders etc.

On another preset there is no harshness, I think on that one though the target curve does drop more after about 2khz, so we'll see. Listened a little earlier but only quietly as the kids were in the car with me and I don't want to damage their hearing!! I don't want to damage mine either, but really don't want to damage theirs.
Loud rock music is good for the little ones! Just roll off the top end.
As for your harshness, it's obviously not digital clipping with those levels, maybe it's just over compensating somewhere for a dip that isn't audible. My 80prs does that really well, boosting the piss out of an inaudible dip, that's why I bought my 8x12.
 
861 - 880 of 928 Posts
Top