DiyMobileAudio.com Car Stereo Forum banner

421 - 440 of 479 Posts

·
Registered
Joined
·
393 Posts
Thanks very much - I was doing something different before.



OK - so I used an offset on the full range curve to try and get it about matched to my full range measurement. Then I uploaded the 3 individual house curves and applied the same offset. Slightly confused by the db figures that REW adds to the names of each overlay, but seems to look OK.



Does this look about right then as a starting point to figure out what changes I need to make to my DSP crossover settings?



BTW - My measurements were using the REW pink noise PN generator averaging around my head. May try other more time-consuming approaches later, but want to just do a first pass to see if I'm understanding things.



Also - there is TA applied already based on just entering measurements to drivers into my Bit Ten.



If I'm reading this right it looks like my sub crossover is about right. Looks like I need to move the mid-range low pass higher and tweeter high pass lower?



Should I do this with no TA also?



Thanks

Are you using 12 and 24 dB slopes on your midbass?

Spatial averaging measurements are fine. Keep doing that.

Measure post time alignment.
 

·
Wave Shepherd
Joined
·
2,398 Posts
Discussion Starter #422
Looks like a good start. All measurements with my tool can be done before or after time alignment ... it will make no difference since you are listening to each speaker one at a time. When you do time alignment, use a tape measure and the other sheet in my tool to help you set the delay times. The combination of filters you use to bend the acoustic response toward the target response does not matter either ... use whatever combination you have the most success with. Most systems will need a combination of non-standard crossover filters in addition to parametric EQ filters. It is, however, best to avoid boosting with EQ filters unless you are really familiar with the trade-offs.
 

·
Registered
Joined
·
393 Posts
Looks like a good start. All measurements with my tool can be done before or after time alignment ... it will make no difference since you are listening to each speaker one at a time. When you do time alignment, use a tape measure and the other sheet in my tool to help you set the delay times. The combination of filters you use to bend the acoustic response toward the target response does not matter either ... use whatever combination you have the most success with. Most systems will need a combination of non-standard crossover filters in addition to parametric EQ filters. It is, however, best to avoid boosting with EQ filters unless you are really familiar with the trade-offs.


I didn’t know you could use 12 and 24 dB crossovers. That’s good to know! I’ve never seen that but I guess it makes sense in certain situations. Thanks Jazzi!
 

·
Registered
Joined
·
78 Posts
Are you using 12 and 24 dB slopes on your midbass?
I have 24db Linkwitz-Riley on everything for now apart from the sub which has a Butterworth 24db (that's an error). I don't know enough to know if that's what the response looks like - presumably you're seeing something?

The combination of filters you use to bend the acoustic response toward the target response does not matter either ... use whatever combination you have the most success with. Most systems will need a combination of non-standard crossover filters in addition to parametric EQ filters.
This is the part I'm struggling to understand - think I'm getting there. In my case I can apply High, Low and Bandpass filters (12 or 24 LR and 6, 12, 18 and 24 Butterwoth) to each driver and I have 30 band graphic equalisation (no parametric).

So for example, on the measurements I have so far, would I be right in thinking that I need to raise the DSP crossover Low Pass on the mids from the current 2.5kHz up a bit because the acoustic crossover is slightly lower than the house curve crossover currently?

Do I do as much as I can to shape the response like that and then move on to using the graphic EQ I have to tune further? E.g. cutting near the desired crossover frequencies if needed?

And then I'd move on to EQing the parts of the response inbetween the crossover points (e.g. peaks on the mid bass at ~250Hz and 1.25kHz)?

I was planning to try using REW auto-EQ with Q of 4.31 to see what it suggests and then use whichever frequencies I have closest to those as a start.

And of course, later on looking at side to side differences (which seem quite large currently).

I understand I need to be careful with adding any gain during EQ. I have brought down DSP output gains to -7dB (-10dB on the sub) currently to give some headroom for using gain in EQ if needed. The system is stupid loud anyway (it's for use not any kind of competition and SQ is everything for me). I seem to get a noise floor issue anyway if the DSP output gains are all the way up.

Greatly appreciate all the work and advice you've put in on the tool and in replies throughout this thread and others.

Thanks to you both for your help with this.

Stuart
 

·
Wave Shepherd
Joined
·
2,398 Posts
Discussion Starter #425
So for example, on the measurements I have so far, would I be right in thinking that I need to raise the DSP crossover Low Pass on the mids from the current 2.5kHz up a bit because the acoustic crossover is slightly lower than the house curve crossover currently?

Do I do as much as I can to shape the response like that and then move on to using the graphic EQ I have to tune further? E.g. cutting near the desired crossover frequencies if needed?

And then I'd move on to EQing the parts of the response inbetween the crossover points (e.g. peaks on the mid bass at ~250Hz and 1.25kHz)?
Stuart,

Yep, that all sounds good. Thankfully The exact order you do adjust the filters and the exact kinds of filters you use is not important. If you find a method that works for you, that's great. I find the crossover filters all by themselves to be incredibly powerful and if you can get set them just right, they can reduce the number of graphic or parametric eq filters you need.

I was planning to try using REW auto-EQ with Q of 4.31 to see what it suggests and then use whichever frequencies I have closest to those as a start.
I once saw you can set both the center frequencies and the Q within the REW Auto-EQ feature. You might try to lock it down a bunch if you intend to use the auto feature.

Honestly since you have a 31 band graphic eq the only sliders you CAN adjust are gains. It might look overwhelming at first, but I promise if you experiment in the EQ module of REW that adjusting the gain sliders will become intuitive pretty quick. It's a really solid skill to develop if you have a little patience.

I understand I need to be careful with adding any gain during EQ. I have brought down DSP output gains to -7dB (-10dB on the sub) currently to give some headroom for using gain in EQ if needed. The system is stupid loud anyway (it's for use not any kind of competition and SQ is everything for me). I seem to get a noise floor issue anyway if the DSP output gains are all the way up.
That's a good way to think about it. Another method would be to make a measurement of each speaker with all filters set flat/disabled and then save those as your "base" that you start with. You can add any combination of filters, including boosting, and you'll be pretty safe so long as your "end" tune for each speaker is not much louder than the base that you started with. The boosting that folks are afraid of is boosting above and beyond what they *started with*. For example you can boost +3dB at 40hz and have a crossover filter at 80hz and you'll be just fine, which shows that not every boost is a bad boost.

I'm happy you find this thread helpful and I'm glad you are here asking good questions too. Keep them coming!
 

·
Registered
Joined
·
78 Posts
Thanks very much. Car in the shop today for a service, but I'm playing with REW and the Bit Ten software offline in the meantime

I realise I shouldn't have said auto-EQ because if you leave the filters as auto, REW will change freq, gain and Q.

But as you say, you can set up the filters as manual (20 at a time) for the graphic EQ frequencies with Q of 4.3. Then play with those and see REWs prediction on what it does to the curves. As you say, that looks pretty powerful.

For others new to this like me, it's possible to save a set of filters so you can set up your manual filters with frequencies and Q of 4.31 and then call them up for each tune. I'll have a set for each of the individual drivers.

Will focus on maximising use of crossovers first however. Thanks for the help again.

Excited to give this another try. I know I'll need to do a bunch of tuning by ear possibly as well as every tune I've done so far has sounded very harsh in the upper ranges (some female vocals, strings, piano notes, rimshots) and I I've usually ended up just dropping the gain on the tweeters to make it bearable - hope I can do a more subtle job with these methods.

FYI - the Bit Ten agrees with your calculation of delays from distances :)

Will post up pics of curves as I progress in case it's helpful for anyone else learning.
 

·
Wave Shepherd
Joined
·
2,398 Posts
Discussion Starter #427
Donanon, I suppose you found the download link afterall?
 

·
Wave Shepherd
Joined
·
2,398 Posts
Discussion Starter #429
You're right. However my (limited) understanding of high energy speakers is that they produce *less* acoustic output than expected, so if anything that would work as a margin of error in my favor since I'm trying to find out how to protect the driver from over-excursion. But again, I'm at the end of my current knowledge on the subject and can't say for certain.

I need to skip all this B.S. degree nonesense and get into the Masters degree of Acoustics at Penn State as soon as possible .... :mad:
It feels strange to quote myself, but I love this message from four years ago. I am now in the master's degree program at Penn State and learning a ton of interesting things such as this problem I struggled with earlier and my other post about real and complex number math. What a journey!
 

·
Registered
Joined
·
1,654 Posts
It feels strange to quote myself, but I love this message from four years ago. I am now in the master's degree program at Penn State and learning a ton of interesting things such as this problem I struggled with earlier and my other post about real and complex number math. What a journey!
I’ve been thinking of PMing you about the Masters degree program you’re taking. But if there is a lot of complex math, that might bring up too many bad memories of my bachelors and masters degree in Industrial Engineering (with a couple years in Electrical Engr).
 

·
Wave Shepherd
Joined
·
2,398 Posts
Discussion Starter #431
I’ve been thinking of PMing you about the Masters degree program you’re taking. But if there is a lot of complex math, that might bring up too many bad memories of my bachelors and masters degree in Industrial Engineering (with a couple years in Electrical Engr).
I have never solved an integral by hand outside of a math department class. I don't need to remember what Stoke's Therorem is or how to find a Curl or see if an infinite sequence is convergent, and I can't do any of that off the top of my head anyways. I have also never needed to cite my sources using MLA format or respond to a company email about which countries were at war in the 1700s in Europe.

There is certainly math in a technical degree like this however the teachers are more interested in testing your knowledge of the new subject so all math solving tools are encouraged from WolframAlpha to Matlab and Mathematica and whatever "cheating" graphing calculator you want to use. There is certainly complicated math in the classes but it's usually the part where the teacher derives stuff on the white board and then simplifies it with some assumptions so you can work with the simpler formulas a little easier.

I've needed to read partial differential equations but never had to solve one. I've been manipulating sums and complex numbers (imaginary) with basic trig functions and super basic integrals at the worst. Logarithms are really important and I'm using plenty from algebra, with a little bit of geometry and trigonometry. The most challenging for me is matrix operations since I've never had to use it for anything, however Matlab does all the hard work for me so long as I remember how it's supposed to work.

I would bet folks with a technical degree who once did math classes would not struggle too much with an acoustics degree since the math involved is very focused and very repetitive. I think the more challenging part for some people would be learning Matlab and using it for almost every homework assignment. It's like learning a programming language at the same time so it could be hard if you haven't used it or haven't done programming before.

tl;dr if you're even a tiny bit interested you should give it a go, and I'm happy to share more
 

·
Banned
Joined
·
1,554 Posts
I have never solved an integral by hand outside of a math department class. I don't need to remember what Stoke's Therorem is or how to find a Curl or see if an infinite sequence is convergent, and I can't do any of that off the top of my head anyways. I have also never needed to cite my sources using MLA format or respond to a company email about which countries were at war in the 1700s in Europe.

There is certainly math in a technical degree like this however the teachers are more interested in testing your knowledge of the new subject so all math solving tools are encouraged from WolframAlpha to Matlab and Mathematica and whatever "cheating" graphing calculator you want to use. There is certainly complicated math in the classes but it's usually the part where the teacher derives stuff on the white board and then simplifies it with some assumptions so you can work with the simpler formulas a little easier.

I've needed to read partial differential equations but never had to solve one. I've been manipulating sums and complex numbers (imaginary) with basic trig functions and super basic integrals at the worst. Logarithms are really important and I'm using plenty from algebra, with a little bit of geometry and trigonometry. The most challenging for me is matrix operations since I've never had to use it for anything, however Matlab does all the hard work for me so long as I remember how it's supposed to work.

I would bet folks with a technical degree who once did math classes would not struggle too much with an acoustics degree since the math involved is very focused and very repetitive. I think the more challenging part for some people would be learning Matlab and using it for almost every homework assignment. It's like learning a programming language at the same time so it could be hard if you haven't used it or haven't done programming before.

tl;dr if you're even a tiny bit interested you should give it a go, and I'm happy to share more
I have a PhD in Math and over 7 years of experience teaching mathematics to engineers and scientists. I will tell you do not be afraid of the intellectual challenges or technicalities you may face - while these may require your best efforts at times, most possess the abilities required to make the effort more than worthwhile. The key is to formulate your long-term goals, and keep them in mind always during the process.
 

·
Registered
Joined
·
1,654 Posts
Thanks so much for the thorough response Justin. I will give it some more thought and may reach out for us to talk. Thanks again!
 

·
Registered
Joined
·
252 Posts

·
Registered
Joined
·
1,548 Posts
Justin, would it be possible to add 48db/oct slope option to the spreadsheet?
 

·
Wave Shepherd
Joined
·
2,398 Posts
Discussion Starter #438
Justin, would it be possible to add 48db/oct slope option to the spreadsheet?
Thank you for asking and making me look at the math again. I finally made sense of it. School is good for something! I also figured out how to work with complex numbers in Excel so now I can do stuff with phase too yay! I have the low-pass filter working, and I'll tackle the high-pass filter later this week. We will have Butterworth and Linkwitz-Riley to choose from, up to 8th-order slopes.


 

·
Registered
Joined
·
1,548 Posts
Thank you for asking and making me look at the math again. I finally made sense of it. School is good for something! I also figured out how to work with complex numbers in Excel so now I can do stuff with phase too yay! I have the low-pass filter working, and I'll tackle the high-pass filter later this week. We will have Butterworth and Linkwitz-Riley to choose from, up to 8th-order slopes.


Wow, now that is customer service! Ask and you shall receive! :p

Seriously, Thank YOU for doing what you do Justin :thumbsup:
 

·
Wave Shepherd
Joined
·
2,398 Posts
Discussion Starter #440 (Edited)
Version 5.4 is now available.

---> download version 5.4 from my dropbox right here <---

Updates:
Rebuilt the 2-way and 3-way crossover sheets with more advanced formulas. You can now choose up to 8th-order Butterworth and Linkwitz-Riley crossovers.

Update update:
The midbass curve on the 3-way sheet was not cooperating and should export properly now.
 
421 - 440 of 479 Posts
Top