Home Theater Forum and Systems banner

1 - 20 of 34 Posts

·
Premium Member
Joined
·
210 Posts
Discussion Starter #1
I have recently changed drivers from stock in my speakers. I would like to ensure the tweeters, mid and woofers are set at a proper level relative to each other. How do I use REW to achieve this kind of calibration?

I do have the ability to attenuate the mid and tweeter.
 

·
Registered
Joined
·
678 Posts
Rudy, I've been thinking ahead to this problem, as I have ordered a set of ALK ES filters for my two LaScalas.

So far, I've come up with two approaches.

(1) Use REW just as a signal generator. One could use the pink noise generator to cover the heart of each driver's response while avoiding the crossover, in my case, for example, 60Hz-450Hz for the woofer, 550-3900 for the midrange, and 4100-16000 for the tweeter, and measure each with an SPL meter. Then adjust the midrange and the tweeter so they give the same SPL measure as the woofer.

(2) Do full range measures of each speaker individually, with Audyssey Off of course, in the middle of the room as far away as possible from the walls, with the mic placed near field facing the speaker. By looking at the graphs, one could try to adjust the midrange to match the woofer in level, with no weird anomalies around the crossover, then do the same to match the tweeter level to the midrange.

I have no idea how well either of these techniques will work. Hopefully someone has already tackled this problem and can share his or her insight.

Cheers,
Bill
 

·
Premium Member
Joined
·
210 Posts
Discussion Starter #3
Bill: Very interesting. I have just finished working on my initial tests and think I found a good way to do this. BTW, I am using the ALK Universal with stock woofers, JBL 2470 mids on ALK Trachorns and Eminence APT-150 tweeters (same as the Crites CT-125).

I too wanted to take as much room effect out of the equation, since Audyssey can correct for that. So, I placed the mic fairly close to my Khorn.

What I decided to do was to unhook all but one driver from the crossover, then run a full sweep. Then, test each driver in succession with the other drivers unhooked.

I then could look at al three traces together and see what kind of response and dB level I was getting. It turns out my mid and tweeter are very well matched and my woofer is a little hot, which is a trait of the Khorn bass bin.

I had purchased three ALK tweeter attenuators anticipating I would need them for this balancing test, but not so. I'll be putting those up on the Klipsch forum next.

I am no expert on how to take readings etc, but beleive this is a reasonable way to measure what each driver is doing and at least balance the mid range on the crossover with the tap settings. The tweeter can be attenuated if required with the ALK tweet attenuator, but the woofer is what it is.

What do think? Needless to say, I am using REW to take the measurements. I am using a Behringer mic.
 

·
Registered
Joined
·
678 Posts
That sounds like a plan.

I was planning on leaving all the drivers connected, at taking measures over each range, which makes sense with the extreme slope filters. I can see why, with the Universal filter and more normal rolloffs in the crossover, you are thinking of measuring each driver separately.

The problem I see with the graphs is that, in room, there are a lot of reflection effects causing dips at various frequencies. With the smoothing recommended here for REW graphs, 1/3 octave, these are largely flattened out so the trends should be visible. I can clearly see a midrange dip in my stock LaScalas. I am hoping to avoid lugging the speakers outside just to reduce the measurement variation when comparing the levels.

The other idea I had was to try both my techniques and see if they give identical recommendations. If the SPL meter on range limited pink noise and the visual inspection of the graphs show the same difference in level between drivers, I will be more confident in making the corresponding change to the crossover attenuators.

As you are ahead of me on this path, I'll be interested in hearing how your technique works, whether the levels sound right to your ears when you are done.

Good luck,
Bill
 

·
Premium Member
Joined
·
210 Posts
Discussion Starter #5
My room is acoustically treated quite a bit, so I don't have too many room induced problems. I had worked on balancing the drivers before, and by ear had ended up with the X-4 tap setting in mid range drivers. I was visiting the issue again since I just changed tweeters and wanted to make sure they were properly set, hence my purchase of the ALK attenuators.

After my measurements, I found the mid and tweeter were dead on. It was really interesting to see each drivers total response ability.

One problem I have if I leave it all connected is that there is some level of boost at the crossover points. With the ES networks, of course, you won't see nearly the crossover issue, but it would still be interesting to 'see' the response of each driver.

My graph has four simperimposed plots, each driver and the full speaker response. As a minimum, it is very interesting to see the contribution of each driver to the 'whole'.

I really can't move my Khorns outside very easily, so will have suffice with the runs I have made.

If anyone has figured out a better way to do this, I would love to hear about it.
 

·
Premium Member
Joined
·
210 Posts
Discussion Starter #6
Bill, what other upgrades have you made to your La Scalas? What made you go with the ES network?
 

·
Registered
Joined
·
678 Posts
I have not made any other changes. I thought about the Trachorns, but I'm sure someone in my family would go to the closet for a gun before they would let me take a saw near the LaScalas.

I opted for the extreme slope filters after reading Al's analysis in his article. When discussing phase issues he pointed out that the overlap in the crossover of a horn inherently creates a phase issue, as the drivers are at different distances from the listener. Thinking about this, and the description of how Audyssey tries to cluster data points in the time domain as well as the frequency domain, I figured that cleaner crossovers would make it easier for Audyssey to equalize the entire range.

Most who have tried them have written positive comments. So I'll give them a try; it's probably the only time I will ever spend this much time and effort improving the LaScalas.

My children are already making plans to fight over them when the time comes.
 

·
Premium Member
Joined
·
210 Posts
Discussion Starter #8
AL is definately correct on the overlap issues. In my graphs, you can actually see this! I too have heard really good things about those crossovers....they are just pricey. I opted to spend my exctra cash on the Trachorns and mid driver upgrades. If you have not heard the Trachorns, don't....they will cost you more money and you will have to have them.

My center La Scala just got its Trachorn installed. That required a jig saw, a steady hand and a lot of courage. I really hated cutting up the speaker, but it had to be done to fit the Trachorn. You can see the final product in my HT page. It's on my sig.

Let me know how your ES networks work out.
 

·
Premium Member
Joined
·
210 Posts
Discussion Starter #10
Not really necessary, just use an appropriate gate for the impulse response in REW to remove reflections.

brucek
Thank you very much. That goes to show you how little I know thus far about the capabilities of REW. I will have to look into how to do that. Hopefully the help file addresses this issue.
 

·
Premium Member
Joined
·
210 Posts
Discussion Starter #11
I just found a few threads addressing how to properly gate results. Thank you for the tip on how to take the readings we need.
 

·
Registered
Joined
·
7,514 Posts
properly gate results
Be sure you move the speaker at least the distance that the gate covers to remove any reflections, and be aware of the mic and its stand. Position the mic head pointing directly at the driver at about 12 inches or less and try and keep the stand out of the way. Be sure you're on-axis, as even small off-axis positions may offer poorer results. In fact an off-axis measurement is also useful.

Remember that gating puts a limit on the lowest frequency and the resolution of the response. For example, if you were trying to limit reflections from surfaces 1 meter away, you would use a minimum 6msec gate (d=(time*speed)/2)=(6msec*344m/sec)/2, but this would limit the lowest frequency of usable response information to ~167Hz (freq=1/period)=(1/gate time). So, you could select a left gate of 1msec and a right gate of 5 msec. You can go less for tweeters and midrange. I seem to remember using a 3 msec gate last time I was testing my tweeters for a test.

Open the IR Windows pop-up and select a gate that you want and watch the impulse response to ensure it's appropriate (with respect to reflections), then select Apply Windows and your response graph will adjust to that gate time response plot. The popup will show the gate window and frequency resolution for you, so you don't need to do the math on that. You're trying to exclude reflections from any surfaces nearby. It could be as simple as the mic stand, but predominantly hard surfaces such as the floor. Of course, shut off any smoothing.

brucek
 

·
Premium Member
Joined
·
210 Posts
Discussion Starter #13
brucek: Thank you very much for all your help. I greatly appreciate the tutoring.
 

·
Premium Member
Joined
·
210 Posts
Discussion Starter #14
Here are some initial results of my very unscientific test. This graph is a set of three individual runs with only one driver hooked up at a time. The mid range has a very low point which is some sort of aberration, no way the mid can go that low. I think I have the mid and tweeter pretty well set up. The woofer shows the characteristic Khorn 'hump' and a large 50 Hz dip due to room modes.

I figure Audyssey can take it from here.
 

Attachments

·
Premium Member
Joined
·
210 Posts
Discussion Starter #15
This last one, shows the full range spectrum superimposed on all three plots. Very cool! I really like REW!

If you look closely at the crossover point between the bass and mid, you will see the problem Al was talking about.
 

Attachments

·
Registered
Joined
·
2,604 Posts
Are you talking about the 60Hz WRT that low aberration on the mid?
I'm sure that's noise not worth worrying about, since it's line frequency (AC) I wonder what's picking it up...
 

·
Premium Member
Joined
·
210 Posts
Discussion Starter #17
Are you talking about the 60Hz WRT that low aberration on the mid?
I'm sure that's noise not worth worrying about, since it's line frequency (AC) I wonder what's picking it up...
When I get a chance I'll run it again and see what the deal is. I have never seen that before.
 

·
Registered
Joined
·
678 Posts
This last one, shows the full range spectrum superimposed on all three plots. Very cool! I really like REW!

If you look closely at the crossover point between the bass and mid, you will see the problem Al was talking about.
Rudy, I find your graphs fascinating. I do wonder about the dip and peak right above the 400Hz crossover. How did you verify the phase of the midrange and woofer when you installed the new crossover? Does the behavior of the freq resp curve change if the phase is changed on the midrange?

By the way, the null at 52Hz suggests a 1/4 wave reflection off a wall 5'5" away or perhaps 16'3.6" away. I have found the freq-dist program on the RealTraps site very helpful.

Curious,
Bill
 

·
Registered
Joined
·
7,514 Posts
Here are some initial results of my very unscientific test.
You haven't told us the gate times you used for the test. Can you tell more about your methods for the testing (i.e. gates, mic orientation and position, speaker position in room with reference to near boundaries, etc).

brucek
 

·
Premium Member
Joined
·
210 Posts
Discussion Starter #20
Rudy, I find your graphs fascinating. I do wonder about the dip and peak right above the 400Hz crossover. How did you verify the phase of the midrange and woofer when you installed the new crossover? Does the behavior of the freq resp curve change if the phase is changed on the midrange?

By the way, the null at 52Hz suggests a 1/4 wave reflection off a wall 5'5" away or perhaps 16'3.6" away. I have found the freq-dist program on the RealTraps site very helpful.

Curious,
Bill
Bill: Very perceptive on your part. The room dimensions are 20' wide and 27' long. I'm sure one of those measurements is causing the dip at about 52Hz.

As far as the mid driver phase, that has been an ongoing issue. It is a JBL 2470, which according to the literature and to a direct contact with the factory technician, has a reverse polarity. I also used a battery to verify that a positive charge to the 'black' terminal of the driver causes the diaphragm to move forward. I had also run some plots right at the crossover region to confirm this, but can't find the plots right now. So, I will run them again. IIRC, the plot very clearly showed that the driver has a reverse polarity just like JBL said, when compared to the woofer.

You haven't told us the gate times you used for the test. Can you tell more about your methods for the testing (i.e. gates, mic orientation and position, speaker position in room with reference to near boundaries, etc).

brucek
Bruce: I am still learning about setting the gate etc, so these plots were taken before I knew I could change it. They were taken with default settings. The mic as placed about 3' away, at driver height, pointed directly at the driver, on axis. The speakers are Klipschorns, which were designed to be placed in room corners to 'finish' the bass bin horn. I do not have suitable corners, so I made false corners, commonly accepted alternative. However, they are still places just a few feet from the very corners of the room. The very corners now have large floor to ceiling 4" thick bass traps.

A picture is worth a thousand words,and the link in my sig will show you plenty of pictures of the setup.
I really appreciate any input and ideas you guys may have. I am going to take some more readings today and see if I can play with the gate just for fun.
 
1 - 20 of 34 Posts
Top