AVS Forum banner
1 - 20 of 29 Posts

·
Registered
Joined
·
133 Posts
Discussion Starter · #1 ·
I have created a pair of stereo speakers to be used in my office. This is a 2.1 system using a powered sub.




using these Tang band drivers.

http://www.parts-express.com/pe/psho...50&ctab=2#Tabs

http://www.parts-express.com/pe/psho...65&ctab=2#Tabs


The SPL charts for these speakers look pretty good and appear to be a good match for these speakers.


Tang band Woofer:



Tang Band Tweeter:



It is driven by my Xonar sound card in my computer. I have listened to them for about a week and they are not bad but they could be better. My sound card comes with an ASIO driver that allows for bypassing the Windows processing but could not figure out how to use this with WinAmp so tried Foobar2000. This configured pretty well and I got the software to recognize and use the ASIO driver. The difference in quality of sound coming from the speakers was immediately apparent. All I can say is WOW. The sound stage, placement and clarity of instruments was far beyond anything I have heard from a computer driven sound system. Frankly, I was not prepared for the quality of sound that this computer driven system produces.


So, now that I have worked out the optimal sound card, driver and software to use I want to go back and tweak the cross over. I am not copying anything with this speaker design and it is my first attempt at doing it all myself. Previously, I have built speakers using others (expert) design and have been quite happy with the results. But I want to learn this too which is why I am pursuing an understanding of crossovers.


I used this program to design the cross over.

http://www.diyaudioandvideo.com/Calculator/XOver/


Inserting the impedance of the speakers and the cross over frequency of 2500 with a 2nd order LR (suggested cross over design from a forum member) into the tool I built the following cross over.




So all this is pretty straight forward. A forum member suggested I use Room EQ Wizard to test the speakers and I have done this. The SPL chart below is from my first test.




I am new to this software so not sure if I did this correctly. I set it to test from 20 to 10,000 Htz and you can see my subwoofer response starting at about 20. This is as expected. However, there are two issues I need help with.


1. There is a big dip at around 2000 Htz (the cross over design was for 2500). Is this valley caused by the cross over and what do I need to try out to tweak the cross over design to level this valley?


2. Are all the sharp peaks on the Room EQ SPL chart and dips normal or does this indicate that either I am not using Room EQ the correct way or there are other issues I need to deal with.


For now, I have used the Foobar equalizer to compensate for this dip and as I have indicated, they sound pretty good but I would like to take care of this correctly in the cross over design.


If the experts could make a few suggestions I will experiment and see if I can come up with a better cross over response. I understand from reading in the forum that it can take quite a lot of tweaking and testing of parts to create a good cross over design and I am willing to do this but I am not sure where to go from here.


If you make suggestions on parts or changes if will help me if you please note exact parts and what to change in the circuit.


All help appreciated.


Usil
 

·
Banned
Joined
·
15,491 Posts
Three problems. First, calculating the crossover using the driver's nominal impedance rather than its actual impedance at the crossover frequency will give less than optimal results. Second, measure outdoors, and reduce the resolution to 1/6 octave. Measuring indoors with full resolution you end up measuring the room too, not just the speaker, and it's very difficult to see what might be a response problem and what's just a boundary reflection. Finally, that dip is probably a phase inversion null. Reverse the leads on the tweeter.
 

·
Registered
Joined
·
133 Posts
Discussion Starter · #4 ·
I have reversed the phase of the tweeter and set smoothing to 1/6. This is taken at the same location as first test. I won't be able to get this outside right away. Just a note, this is a 2.1 system and there is a seperate powered sub woofer with amp crossover is set at 100 Htz.


Looks like dip is still there.




Usil
 

·
Registered
Joined
·
1,694 Posts
I'd add an impedance compensation circuit to the woofer. My WAG would be a ten microfarad cap in series with an 8 ohm resitor. Connect this across the leads of the woofer. Then try both polarities, and see which sounds the best. Then measure again. In case you didn't know, the parts you use can be about plus or minus 10% of your calculations, so don't be concerned about using a five microfarad cap for C2 for example, if you come up short on fours. When you get close to perfection, small changes will matter more than when roughing out an initial x-over.
 

·
Registered
Joined
·
133 Posts
Discussion Starter · #6 ·
Is this change to address the dip in the 2000 range or fix something with the woofer? I made a note on the last chart that I also have in the circuit a powered subwoofer. So that is most of what you see in the 20 to 100 range.


I have marked the crossover locations on the chart.


Usil
 

·
Registered
Joined
·
1,694 Posts

Quote:
Originally Posted by usil /forum/post/19648720


Is this change to address the dip in the 2000 range or fix something with the woofer? I made a note on the last chart that I also have in the circuit a powered subwoofer. So that is most of what you see in the 20 to 100 range.

It will change the woofer rolloff on the top end, and change the phase. The rolloff slope will be steeper. Nothing changes below about 500 hz. The tweeter and woofer phase may get closer to matching either with one tweeter polarity, or the other.


Have you heard of HOLMS? It's free software for measuring.
 

·
Banned
Joined
·
15,491 Posts

Quote:
Originally Posted by usil /forum/post/19648649


I have reversed the phase of the tweeter and set smoothing to 1/6. This is taken at the same location as first test. I won't be able to get this outside right away. Just a note, tthis is a 2.1 system and there is a seperate sub woofer amp crossover is set at 100 Htz.


Looks like dip is still there.




Usil

Since the dip is still there it's not a phase issue. That being the case it looks like the actual low pass corner on the woofer is around 1kHz, and the high pass corner on the tweeter is around 3kHz. The way to be sure is to test each driver/filter independently.
 

·
Registered
Joined
·
133 Posts
Discussion Starter · #9 ·
I will do some measurements with each speaker seperately and post results.


I thought that this would be relatively easy since if you look at the speaker SPL charts seperately, each is relatively flat at 90 db level. You would think that all I need is to filter the lows away from the tweeter and filter the highs away from the woofer. No other compensation appears necessary.


Usil
 

·
Registered
Joined
·
712 Posts
They are flat in an anechoic room and probably on an open baffle... The box will change that, especially true for the woofer.
 

·
Registered
Joined
·
133 Posts
Discussion Starter · #11 ·
I have measured the tweeter and woofer alone without a cross over in the speaker box - (1st two SPL charts). This looks nothing like the SPL charts supplied in the pdf for each speaker. The third chart is the tweeter and woofer with the cross over. I have turned the powered sub-woofer off for this test.


Woofer - no cross over:



Tweeter - no cross over:



Woofer and Tweeter with 2 order LR cross over at 2500 (powered sub is off):



I am not sure how to read this to determine the cause of the big dip between 1000 and 4000. Is appears the first half of the dip is from the woofer fall off and the second half is from the tweeter (less efficient at that frequency) and the cross over has not addressed it properly. Looking at the tweeter SPL chart it looks like 2500 was a good point for the cross over. As for the woofer - the pdf for that speaker looks flat to 5000 so why the measured sharp fall off at 1000 I am seeing? If it had been flat to 5000 I don't think I would be having this problem.


So, the questions are:


1. Is the cross over correct for this match and just needs adjustment?

2. Is it the wrong cross over?


Another question is why is the tweeter flat from 5000 on in the tweeter only chart and it raises in the chart where the cross over is active?


This is like a puzzel and find it interesting but I would be happy to fix the big dip.


Usil
 

·
Registered
Joined
·
26,476 Posts
Can he gate the meaurement in REW? I have not look at gating in REW. Its time for him to run HOLM. I think its a far better full range speaker measurement tool.


How about putting felt around the tweeter? I thought I read something about baffle diffraction and tweeters causing dips. Not sure if its that low in the frequency though.
 

·
Banned
Joined
·
15,491 Posts
The charts of both the woofer and tweeter are so bad that I think your measuring protocol is at fault. Measure outdoors, the speaker on the ground facing up, the mic suspended above it by at least six feet. That will give an accurate half-space result above the baffle step frequency. Below the baffle step measure ground plane, the speaker stood normally, the mic an inch above the ground, also at least six feet away. Combine the two for a complete half-space result.
 

·
Registered
Joined
·
133 Posts
Discussion Starter · #14 ·
I just realized that the testing was done with the computer software sound equalizer was not set to the flat default setting so results were colored by equalizer settings. Also, since this is a computer system, the computer fans are audible so I think this may be clouding the results too.


Some preliminary placement locations of the speaker and quick tests show that postion in the room have an effect on the spl results. I need to work out an optimal location in my office or determine the final location and retest. I am planning to mounting these on a back wall with a bracket that places the front of the speaker about 18 inches from the wall. I am not going to be able to get the computer system outside so will have to look at optimizing the final location for my office space. I will get back to this once the speakers are mounted but I am sure that the cross over is going to need some work.


Usil
 

·
Registered
Joined
·
133 Posts
Discussion Starter · #15 ·
After calibrating the Room EQ wizard, insuring that the equalizer is not used and placing the speaker in my test location on my desk here is the default spl chart from which I can work. This seems workable for correction in the crossover but not sure what to modify.




Usil
 

·
Registered
Joined
·
712 Posts
Redo the measurements for the woofer and tweeter separated. Make sure the microphone is on-axis with the tweeter. Under normal listening conditions it is recommended that the ear be on level to the tweeters.
 

·
Registered
Joined
·
133 Posts
Discussion Starter · #17 ·
Do you mean each speaker without going though it's crossover or keep in the crossover and simply do one speaker then the next (disconnectiong one speaker for each test)


Usil
 

·
Registered
Joined
·
26,476 Posts
Are you saying you are measuring your speakers in stereo?


You should only measure one speaker period. Measure one speaker, each driver separate (disconnect the other drvier). Measure 1 meter from the speaker with the mic positioned centered between the woofer and the tweeter.
 

·
Registered
Joined
·
133 Posts
Discussion Starter · #19 ·
No, I am using only the left speaker. The right speaker is disconnected.


I am not sure what he wants but thought he wanted me to disconnect the tweeter and test the woofer then reconnect the tweeter and disconnect the woofer in the one speaker box to test the tweeter. But not sure if he meant speakers only testing or speaker with crossover testing.


Usil
 

·
Registered
Joined
·
1,694 Posts
Save yourself some grief, and get setup with HOLMS. Like Bill said, your measurements look too far off to be valid. Another option would be pink noise and a RTA, or warble tones with an SPL meter. Do you have another speaker you can measure, and compare? It will probably look just as screwy.
 
1 - 20 of 29 Posts
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top