DiyMobileAudio.com Car Stereo Forum banner
41 - 60 of 236 Posts

·
Registered
Joined
·
462 Posts
The individual drivers post Dirac measurements did match the Dirac targets very closely but the relative levels between them did not match the relative levels between the targets set in the filter design screen. It set the sub level (as viewed in the Dirac tab in the plug in) to -12db or so... Which killed the sub level and messed up the xover point / phase relationship between it and the midbass.
Ah, I see. I suppose you could go adjust up the gains in the DSP (or at the amp) after Diract. I know you already know this :), and it's obviously not desirable.

I've noticed that the Dirac tab in the MiniDSP software shows what gains it introduced on each channel, and those can't be modified. It's pretty annoying because, even if you turn Dirac off, those gains are still in effect, which seems odd to me. It's documented in the manual that way as well (I.e. the expectation is that Dirac gain adjustments are in effect irrespective of whether you have Dirac on).

That actually forced me to start a tune over again once, once I realized it was doing that.
 

·
Registered
Joined
·
4,980 Posts
Discussion Starter · #42 ·
I've been experimenting with multi channel for a while. And applying the xover filters after Dirac correction is something I've been planning to try.

The only problem I see with this is grouping the stereo pairs in separate groups in the filter design screen. In my experience, no matter the relative level difference between the groups, Dirac will level out to the default curve (that first shows) killing the low freq rise. It even shows that way when switching between the measurement and filter design screens.

Here is the Dirac measurements from a pretune done in the plug in. In this case I did not defeat the sub LP as directed in the manual. Notice the relative levels between the sub/midbass and a xover point of 70hz:

View attachment 312429


Then when continuing to the filter design screen - notice the relative levels have now "normalized" to follow what is the default target in Dirac and xover has now moved to 50hz:

View attachment 312430


Here is a shot of the measured, optimized, and curtains used for each stereo group. Notice the grouping on the right side. Target files were created with Jazzi's spreadsheet, modified with extra data required and loaded into Dirac. Also notice the optimized targets relative to each other accounting of for typical LF rise & xover at the planned 70hz.

View attachment 312432

So one would think that this would measure somewhat like the targets were set at with their levels relative with each other intact but it does not. I do not have a screen shot proving it on my work computer but the measured corrected response basically follows the default curve that initially loads on the filter design screen.

I have plans to experiment with keeping the Sub & midbass in the same group. Hopefully that will prove a remedy to this issue.

What I did is different. I had ALLof the crossovers OFF during the calibration... yes A cap for the tweeter would be great, youll want to leave it on because it will become part of the measurement. and make sure its a hi-volt polypropeylne cap that rolls off way out of band. Like a 20-50mfd would be good


yeah I get your rise thing, With this youll want to make everything flat so the crossoverws will sum perfectly. If you add tilt to the target I am not sure how they'll sum. All I know is this is what I did and it sounded killer....Had plenty O midbass and the sub I had at -20db before dirac so plenty of gain from sub to ajust after.

youll have to try it but put tilt in your targets.... I would imagine as long as both speakers in a crossover have the same tilt it should work....maybe... idk...
 

·
Registered
Joined
·
4,980 Posts
Discussion Starter · #43 ·
Ah, I see. I suppose you could go adjust up the gains in the DSP (or at the amp) after Diract. I know you already know this :), and it's obviously not desirable.

I've noticed that the Dirac tab in the MiniDSP software shows what gains it introduced on each channel, and those can't be modified. It's pretty annoying because, even if you turn Dirac off, those gains are still in effect, which seems odd to me. It's documented in the manual that way as well (I.e. the expectation is that Dirac gain adjustments are in effect irrespective of whether you have Dirac on).

That actually forced me to start a tune over again once, once I realized it was doing that.
Yeah it lowers everything to the lowest measured source.... like if the rears measure an average of 7db less then fronts... guess what.....lol

so what I do is pre dirac in that case would just add back the 7db as a output boost to the rears, run dirac, then lower them back after,
 

·
Registered
Joined
·
4,980 Posts
Discussion Starter · #44 ·
Ah, I see. I suppose you could go adjust up the gains in the DSP (or at the amp) after Diract. I know you already know this :), and it's obviously not desirable.

I've noticed that the Dirac tab in the MiniDSP software shows what gains it introduced on each channel, and those can't be modified. It's pretty annoying because, even if you turn Dirac off, those gains are still in effect, which seems odd to me. It's documented in the manual that way as well (I.e. the expectation is that Dirac gain adjustments are in effect irrespective of whether you have Dirac on).

That actually forced me to start a tune over again once, once I realized it was doing that.
in Dirac 3.1.1 you can offload filters now. There’s an x next to the filter in Dirac app.
once a filter is loaded the TA and gains are always on weather Dirac is on or off.... makes things very hard for us that use upstream Dirac boxes and doing a pre Dirac setup.....

Now you can offload the filters and then it will be a raw signal only in that slot/configuration
 

·
Registered
Joined
·
462 Posts
in Dirac 3.1.1 you can offload filters now. There’s an x next to the filter in Dirac app.
once a filter is loaded the TA and gains are always on weather Dirac is on or off.... makes things very hard for us that use upstream Dirac boxes and doing a pre Dirac setup.....

Now you can offload the filters and then it will be a raw signal only in that slot/configuration
You're saying there are X buttons in the last step of the Dirac app (where you pick which presets to output to)? And if you X that out, it will clear the Dirac tab in the MiniDSP for that specific slot? Hoping your answer is yes.
 

·
Registered
Joined
·
33 Posts
For the sub bass, I had the best luck using 2 channels for the front stage, in Dirac, and then using a 3rd channel for the sub, but not using Dirac for that (just REW). This was per OP's suggestion. The reason putting the sub into the 2 channel Dirac didn't work for me (I think), is because Dirac was introducing some delay on one of the two channels...so when channels 1 and 2 mixed to the sub, it would be little funky with that extra delay on one half of the signal. I think it would also work if you used a third channel for the sub and ran that through Dirac, and draw your target (similar to Truthunter's approach).

I'm surprised Truthunter's 7-channel method didn't work as he expected it to. I would expect Dirac to match whatever curves he drew, especially knowing that they were achievable curves.
Thanks! All of this info is really helpful. I wish I had better knowledge on tuning.
 

·
Registered
Joined
·
279 Posts
You're saying there are X buttons in the last step of the Dirac app (where you pick which presets to output to)? And if you X that out, it will clear the Dirac tab in the MiniDSP for that specific slot? Hoping your answer is yes.
This is the way I understand it to work, though I've not tried it myself.
 

·
Registered
Joined
·
279 Posts
I've thought about trying this method too, but I run a pretty good tilt to my target curve so I'm not sure how that would work EQ'ing everything flat.

One concern I had was that doing it this way eliminates the benefits of having Dirac adjust phase through the xover region. True, if you get things perfectly flat you don't need any phase correction, but would you rather have Dirac just doing simple EQ (granted its using FIR filters instead of IIR), or use it to help with phase? Not sure one way is better than another, but something to consider. The other option would be to use the EQ in the Mini to achieve a flat response and then have Dirac match that to your target curves from Jazzi for the multi-way.
 

·
Registered
Joined
·
4,980 Posts
Discussion Starter · #49 ·
You're saying there are X buttons in the last step of the Dirac app (where you pick which presets to output to)? And if you X that out, it will clear the Dirac tab in the MiniDSP for that specific slot? Hoping your answer is yes.
Yes
You can now delete Dirac filters and have Dirac truly disabled for a real pre Dirac tune
 

·
Registered
Joined
·
4,980 Posts
Discussion Starter · #50 ·
I've thought about trying this method too, but I run a pretty good tilt to my target curve so I'm not sure how that would work EQ'ing everything flat.

One concern I had was that doing it this way eliminates the benefits of having Dirac adjust phase through the xover region. True, if you get things perfectly flat you don't need any phase correction, but would you rather have Dirac just doing simple EQ (granted its using FIR filters instead of IIR), or use it to help with phase? Not sure one way is better than another, but something to consider. The other option would be to use the EQ in the Mini to achieve a flat response and then have Dirac match that to your target curves from Jazzi for the multi-way.
Yeah , the 2ch Dirac will fix the step, and subsequently correct for crossover phase in the sum...

this method is for a multi-way tune with Dirac, knowing there will still be crossover phase shift...

The thing is , if they sum textbook it’s just an allpass.... it’s like turning on an allpass filter on a point source. It’s the minimal amount of phase shift, but more importantly they sum to be in phase. So your ear can not detect the phase change.
so it sounds (dam near) linear phase... but it’s not. It’s actually an allpass.

in a car it’s very difficult to get good phase behavior from crossovers.. like really good is what I’m talking about.. this method, while I was in a rush and didn’t have time to really test any of it , it sounded very good.. I was blown away how well things were summing and how coherent the left and right and all the speakers were playing with each other.

this would definitely be the way to do it if someone had a ddrc22 upstream of a 8x12
 

·
Registered
Joined
·
4,980 Posts
Discussion Starter · #51 · (Edited)
I've been experimenting with multi channel for a while. And applying the xover filters after Dirac correction is something I've been planning to try.

The only problem I see with this is grouping the stereo pairs in separate groups in the filter design screen. In my experience, no matter the relative level difference between the groups, Dirac will level out to the default curve (that first shows) killing the low freq rise. It even shows that way when switching between the measurement and filter design screens.

Here is the Dirac measurements from a pretune done in the plug in. In this case I did not defeat the sub LP as directed in the manual. Notice the relative levels between the sub/midbass and a xover point of 70hz:

View attachment 312429


Then when continuing to the filter design screen - notice the relative levels have now "normalized" to follow what is the default target in Dirac and xover has now moved to 50hz:

View attachment 312430


Here is a shot of the measured, optimized, and curtains used for each stereo group. Notice the grouping on the right side. Target files were created with Jazzi's spreadsheet, modified with extra data required and loaded into Dirac. Also notice the optimized targets relative to each other accounting of for typical LF rise & xover at the planned 70hz.

View attachment 312432

So one would think that this would measure somewhat like the targets were set at with their levels relative with each other intact but it does not. I do not have a screen shot proving it on my work computer but the measured corrected response basically follows the default curve that initially loads on the filter design screen.

I have plans to experiment with keeping the Sub & midbass in the same group. Hopefully that will prove a remedy to this issue.
so Truthunter ,
Each Dirac EQ is it’s own separate Dirac calibration and doesn’t take any of the other EQs (Dirac channels) into consideration unless there linked to the same group...

So in a multi-way like yours, it’s not going to match your targets across groups, it’s only going to match levels in the normalized view.

so what I noticed was more of a look at where each magnitude shape is the lowest and average of them all.

For getting your response to match each group , starting with the 0db line draw your targets the way you want not to exceed 10db window of correction per group. Get the shape right and maximize your gain in the target, lest you’ll be stuck with a lowered output across all channels.

you build your rise out of your groups after Dirac using only levels. You’ve at this point maximized your gains during calibration. So now go to the 8x12 app and start dropping levels on HF, mids , or whatever you want the shape to be....

the problem you’ll never be able to overcome doing this is you will have crossovers that won’t sum properly, and more then likely have crossover cancellations killing your reinforcement , you’ll be rolling the dice.

it’s apparent, to me , to do a multi-way in Dirac using groups to control the multi-way like this, you have to use a ruler flat target so the crossovers will sum properly, and turn off your crossovers before calibration.... it just is the way this need to be...

so start over and do it this way....with flat targets and no crossovers and draw targets as far out as the natural response allows.
then after your done do this

go to the output PEQ on every channel and make a shelving filter that draws your desired target and make the same exact shelving filter on ALL of the outputs so they all inherit the same phase shift that the shelving filter imposes...

that is the only way I can see it working properly... you’ll have good summation, crossovers will work properly, and you’ll achieve your target with no real losses to gain
 

·
Registered
Joined
·
1,820 Posts
Yes
You can now delete Dirac filters and have Dirac truly disabled for a real pre Dirac tune
I need to try this out... This is the one thing that everyone talks about needing a solution for!


so Truthunter ,
Each Dirac EQ is it’s own separate Dirac calibration and doesn’t take any of the other EQs (Dirac channels) into consideration unless there linked to the same group...

So in a multi-way like yours, it’s not going to match your targets across groups, it’s only going to match levels in the normalized view.

so what I noticed was more of a look at where each magnitude shape is the lowest and average of them all.

For getting your response to match each group , starting with the 0db line draw your targets the way you want not to exceed 10db window of correction per group. Get the shape right and maximize your gain in the target, lest you’ll be stuck with a lowered output across all channels.

you build your rise out of your groups after Dirac using only levels. You’ve at this point maximized your gains during calibration. So now go to the 8x12 app and start dropping levels on HF, mids , or whatever you want the shape to be....

the problem you’ll never be able to overcome doing this is you will have crossovers that won’t sum properly, and more then likely have crossover cancellations killing your reinforcement , you’ll be rolling the dice.

it’s apparent, to me , to do a multi-way in Dirac using groups to control the multi-way like this, you have to use a ruler flat target so the crossovers will sum properly, and turn off your crossovers before calibration.... it just is the way this need to be...

so start over and do it this way....with flat targets and no crossovers and draw targets as far out as the natural response allows.
then after your done do this

go to the output PEQ on every channel and make a shelving filter that draws your desired target and make the same exact shelving filter on ALL of the outputs so they all inherit the same phase shift that the shelving filter imposes...

that is the only way I can see it working properly... you’ll have good summation, crossovers will work properly, and you’ll achieve your target with no real losses to gain
I'll be experimenting with all of this soon and some other ideas I have. I found getting repeatable good results has been achieved with a pretune to the "default" Dirac curve (which is not ruler flat by the way) followed by a 2ch Optimization on top of it to follow a final target curve. But I think there is something to applying xover filters after that may be beneficial which you have confirmed. And the fact that xover filters can be cascaded or virtually cascaded with this unit makes for even more fun experimentation. Thanks for sharing on here Andy!
 

·
Registered
Joined
·
4,980 Posts
Discussion Starter · #54 ·
I need to try this out... This is the one thing that everyone talks about needing a solution for!




I'll be experimenting with all of this soon and some other ideas I have. I found getting repeatable good results has been achieved with a pretune to the "default" Dirac curve (which is not ruler flat by the way) followed by a 2ch Optimization on top of it to follow a final target curve. But I think there is something to applying xover filters after that may be beneficial which you have confirmed. And the fact that xover filters can be cascaded or virtually cascaded with this unit makes for even more fun experimentation. Thanks for sharing on here Andy!
no problem...
Yes fun indeed.

the thing with crossovers, and Dirac , if theres not setup to sum properly before a tilt is applied, the tilt and the phase changes made by Dirac can (often times does) cause crossover cancellations. Where the two speakers being crossed over are fighting and canceling.

yeah the default target has a tilt, but the default target was not meant to tune crossovers in a multi-way.

if it’s to be used as such, the target needs to be flat with Crossovers off. the crossovers are meant to sum to the textbook using flat magnitude. Not a tilt. Any other way would be your own alignment and not a LR or BW or whatever your doing.... when the alignment is proper it’s magic

a tilt can be made after with shelving filters successfully....it will be exact and have the exact shape desired. Applied to every output so it’s a global change.
 

·
Registered
Joined
·
33 Posts
Gave this a go this evening after swapping subs. Whoa! Very nice 😊 Staging and overall sound are so much better. Only minor gripe is that the system volume is noticeably lower compared to my previous tune. Will play around with this some more when I have time.
 

·
Registered
Joined
·
206 Posts
Noob question, but can you share screen shots of how many Direc channels you are using and the matrix.

Just got DL and trying to figure it all out.

I’m going to run digital on a 2 way front end and single sub channel. So do I use Dirac 1 for the left side mid and tweeter and sub and Dirac 2 for the right side mid and tweeter and sub? Or I Dirac channel for each driver?

If you do a base Dirac tune, is that another Dirac channel and the multi-way are additional channels?

Sorry I work best with visuals :)

Also, is there a screen in the 8x12 software to show all channels and their crossovers like the Helix main screen?

For the Dirac tune is it correct that no crossovers should be set? If so, guess a tweeter protection cap should be used?

I’ll be moving from a Helix DSP Mini to 8x12DL once I sort out the install to make space for the 8x12


Sent from my iPhone using Tapatalk
 

·
Registered
Joined
·
4,980 Posts
Discussion Starter · #57 ·
Gave this a go this evening after swapping subs. Whoa! Very nice 😊 Staging and overall sound are so much better. Only minor gripe is that the system volume is noticeably lower compared to my previous tune. Will play around with this some more when I have time.
yeah I about shyt myself when I first heard that tundra.... I was like , why in the F*** haven’t I done this before ...
 

·
Registered
Joined
·
33 Posts
yeah I about shyt myself when I first heard that tundra.... I was like , why in the F*** haven’t I done this before ...
Did a little more tweaking this afternoon to correct a couple of mistakes I made. I didn’t read your post carefully enough probably because I had a super long day. I didn’t curtain off the target line at the roll off plus I set them too low. Sounding even better now. I cannot believe how enveloping the sound is.

Now, I have to either repair or replace my sub amp. eBay fail (smh).
 
  • Like
Reactions: Truthunter

·
Registered
Joined
·
279 Posts
I need to go back and re-read your original post because I still feel like I'm missing something. It's not clear to me how you get to the house curve you want without either setting the xovers first or having Dirac eq to the curve for each driver....

Sent from my Pixel 4 using Tapatalk
 

·
Premium Member
2006 Hyundai Sonata V6
Joined
·
317 Posts
I'm also rather confused by all this. I realize it's a big ask, but would anybody who has given this technique a try be willing to take some screenshots to illustrate exactly what you're doing?
 
41 - 60 of 236 Posts
Top