Fixing my end game speakers

[muad]

Hey everyone! I have some technical questions and really could use some help figuring out my issue. Along the way during my audio journey I decided I wanted some expensive speakers for a change (because that means they're good right?), but because I couldn't afford them I decided I would go the DIY route. I had someone design the speaker for me, and build and test the crossover. I spent a large chunk of last year building them, put my heart and soul into the build, and they look gorgeous but out the gate didn't sound right.

Either way I played with the eq at the approximate crossover points and I really think it's just a matter of the tweeter and midrange are just too high. After shelving the mid 4db down, and the tweeter 2db down, they sound pretty damn good. I did some in room measurements, and compared them to Elac DBR, which is properly balanced as per Klippel, and there is definitely too much energy above 400-500hz. I wan't to measure the speaker as objectively as possible and fix it at the crossover level. I don't know what the actual crossover points are, but I am waiting for a response from the designer. Also, I don't care about the added bass from what I want to do, as I can use room correction to fix that.

I just have a couple of questions.

1. Due to the driver layout I am having a difficult time doing quasi anechoic measurements. The bottom woofers proximity to the floor is making gating an issue. I don't really care about getting a detailed measurement, as I am just interested in getting an accurate relation between the mid and woofer levels. Is there an accurate way to measure this? Would close micing the woofer (+adding 3db for second woofer), give me an accurate spl in the 200-1000hz range? Any other suggestions?

2. The horizontal on and off axis are really good on this speaker. The dispersion mismatch is minimal below the tweeters omni directional point. Would dropping the relative levels between drivers while keeping the same crossover points have a massive adverse effect to my on off axis uniformity?

I really want to keep these speakers. I put way to much effort, money and time into them, and I love how they look. Any advice would be much appreciated.

Thanks!

 

[levimax]

There are many ways to fine tune the response of a DIY speaker. I made some home made 3 way active speaker (using analog crossovers) and got "close" to the response I wanted but as I learned to use REW I realized they were not quite right. Not sure your playback chain but if you can use DSP convolution filters you might want to try this.

1. Take measurements with REW at listening position using RTA and MMM method. I find I get same result as averaging multiple gated sweep measurements in about 5% of the time.
https://www.avnirvana.com/threads/mmm-mouving-measurements-method-which-one-is-best-with-rew.6360/

2. Use Rephase to generate EQ adjustment filters. There is a learning curve but not as hard as building those nice enclosures!

3. Remeasure and listen.

4. Repeat until you are happy.

Good luck

 

[muad]

There are many ways to fine tune the response of a DIY speaker. I made some home made 3 way active speaker (using analog crossovers) and got "close" to the response I wanted but as I learned to use REW I realized they were not quite right. Not sure your playback chain but if you can use DSP convolution filters you might want to try this.

1. Take measurements with REW at listening position using RTA and MMM method. I find I get same result as averaging multiple gated sweep measurements in about 5% of the time.
https://www.avnirvana.com/threads/mmm-mouving-measurements-method-which-one-is-best-with-rew.6360/

2. Use Rephase to generate EQ adjustment filters. There is a learning curve but not as hard as building those nice enclosures!

3. Remeasure and listen.

4. Repeat until you are happy.

Good luck

Thanks,

I'm going to try this. I'm just a little miffed that the respective driver levels are so far off.

 

[levimax]

Thanks,

I'm going to try this. I'm just a little miffed that the respective driver levels are so far off.

Yea crossovers tuning is certainly a PITA especially for DIY. When a commercial passive speaker is designed they spend a lot of time on the crossovers to fine tune response / house sound after the enclosures and drivers are finalized. The good thing about FIR filters is that you don't have to solder in different value components and you don't waste amp power like you do with pad down resistors.

Good luck.

 

[muad]

Can you recommend a guide for using REW with Rephase. A few sources say not to boost the nulls, which makes complete sense. However since I feel like the woofer level is set too low, this will not help fix my relative driver levels.

I'm not sure how to go about keeping relatively linear between drivers without boosting the nulls. Any suggestions?

This is what I came up with, not sure if this is acceptable.

 

[Blumlein 88]

http://audio.claub.net/software/FRD...curate In-Room Frequency Response to 10Hz.pdf

This paper should be helpful. How to blend various measurements to get proper measures in room down to 10 hz. How to splice the woofer at the correct level onto the rest of the response.

I forget which user first posted links to this paper, but it is useful for your purposes.

 

[levimax]

Can you recommend a guide for using REW with Rephase. A few sources say not to boost the nulls, which makes complete sense. However since I feel like the woofer level is set too low, this will not help fix my relative driver levels.

I'm not sure how to go about keeping relatively linear between drivers without boosting the nulls. Any suggestions?

This is what I came up with, not sure if this is acceptable.

This is a big subject not without some controversy. The REW generated EQ filters imported into Rephase to generate the convolution wave file is a great place to start provided the initial measurement is accurate which the paper referenced above goes through with a lot of detail about the issues and solutions. Here are a few other points that might help get you started.

1. Frequencies below "Schroeder" frequency (300 Hz or so) are going to have a lot of crazy peaks and valleys which are caused by the room not the speakers. Knocking down the peaks is the best way to start, you can try some boost to the valleys but keep it 8 db or less (controversial subject). The valleys look bad but don't really sound bad where as are the peaks are very noticeable. I use MMM RTA at listening position to measure this area and make EQ adjustments this measurement which is considered OK for these frequencies.

2. For 300 to 900 Hz I still use MMM RTA and EQ against that (controversial subject) as some say not to. In your case where you are trying to correct large FR imbalances between the drivers I would not worry about the controversy and just use the MMM RTA up to 900 Hz.

3. Above 900 Hz I use "near field" ~ 1 meter gated sweep to EQ against because if you take listening position measurements it is nearly impossible to separate the room from the speakers and since at these frequencies you "hear" the initial response from the speakers you don't want to EQ the room response.

4. I would NOT worry about phase at this point. Maybe after you get FR you like but most on these boards, myself included, don't notice much of any difference from phase EQ's.

After you get what you feel is an accurate measurement in REW I would let REW do it's EQ corrections and export to Rephase. You can look at them and see what it does (pretty good usually) and take a listen and remeasure. You can then make manual adjustments and see if you like them better.

 

[muad]

http://audio.claub.net/software/FRD_Blender/White Paper - Accurate In-Room Frequency Response to 10Hz.pdf

This paper should be helpful. How to blend various measurements to get proper measures in room down to 10 hz. How to splice the woofer at the correct level onto the rest of the response.

I forget which user first posted links to this paper, but it is useful for your purposes.

 

[muad]

Thanks for the reply guys, I'm going to try all your suggestions. I plan on close micing but I need a mic stand for that.

I had another idea on how to boost the relative driver imbalances without correcting the nulls. Im thinking if I run the autoeq with smoothing at 1:3 that should average out the bottom end and lower mids just to get relative driver levels in line. I can then rerun at variable smoothing to just fix any peaks?