• The move to the new server is done. There are some software and database maintenance updates in process. This has us passing the hat around to help out. We appreciate any donations. Seriously, even a dollar helps. The payment page may be found here - https://www.audiokarma.org/support.html

KEF 107 frequency response Measurements

J

judgeschreber

Well-Known Member
I'm looking for a new tweeter for my newly acquired KEF 107 speakers, since one of them is blown.

I just received a package from madisound with 3 good-looking contenders, and I will also be testing out a couple I have around the house.

I just did measurements on the one fully-working speaker, and thought I'd post them now, in preparation for testing with the possible substitute tweeters. If anyone wants to follow along and give advice as I go, I'd be grateful. . . .

OK, so here are the measurements with a stock 107.

First graph is nearfield bass with the Kube equalizer set to 50 hz
Next is same thing but set to 18 hz (the "Q" and "contour" controls are set at 12 o'clock)
Third graph is nearfield midrange
fourth graph is nearfield tweeter
fifth graph is microphone pointed to the bottom of the tweeter enclosure, about 2.5 feet away.

my first question for all of you speaker gurus: why does the non-nearfield measurement look so ragged? I expected the bass response measurement to be unusable because of room resonances, but I thought that midrange and treble wasn't supposed to be affected so much. . . .

Anyways, I'll be testing possible substitute tweeters soon, so I'll update the thread as I go. For reference on the OEM tweeter's specs, and the substitutes I was thinking about, check out this thread: http://www.audiokarma.org/forums/showthread.php?t=175162
 

Attachments

Register to hide this ad
I don't do mids or highs nearfield. Reason: wavelengths get short enough for the driver to interfere with itself.

Looks like 2 kHz is your crossover frequency, and the mid is cancelling the tweeter in that region. Gotta shut off the mid or tweet to measure that way. One driver at a time, in isolation.

Unless you're doing quasi-anechoic time windowing, the further back you get, the more room reflections with interfere.

On tweeter axis is standard. Pick an axis and distance and replicate that condition for all comparative measurements.

Work to reduce reflections and "clean up" the response; the trends will become more clear.

"Testing Loudspeakers" is your instruction manual. :yes:
 
Thanks for the response Zilch.

This software is called "Fuzzmeasure," which supposedly uses a quasi-anechoic measuring technique. I'm not sure how it does it, but it seems to be the best software available for mac (at least according to several sources).

Clearly I don't have a lot of experience here, but I have a microphone, some software, and a few tweeters to try out, and I'm hoping it will help me out in selecting the right driver.

I've attached curves from the three vifa tweeters that I'm trying out, and a curve for the midrange and woofer without the tweeter. I will try to do the midrange and tweeters in isolation next. I also tried out three other tweeters (the Allison tweeter and two Focal tweeters), but the Allison was too ragged, and the focals were too hot.

Unfortunately, all three of these Vifa tweeters would require baffle modification, and seeing as how the baffle is some sort of hi-tech plastic, I'm not looking forward to that. . .
 

Attachments

I don't know any of these tweets from holes in the ground.

DX looks best, but needs a bit of attenuation to blend optimally with the mid.
 
Ok, I decided that I'm not going to open up the bass bin tonight, so the tweets are measured here without the mid, but with the woofers.

I also realized that since the "head pod" on this speaker swivels about 30 degrees, that I can do off-axis responses easily, and without compromising my microphone position. So the following charts are without the midrange, with on- and off-axis readings.

Zilch, I was thinking the XT rather than DX. The XT is a little less lively, but seems to blend with the midrange pretty seemlessly, no?
 

Attachments

OK, I'm noticing there are several resonances in common with all of the tweeters, namely everything from about 5.5k to 12k. All the tweeters have humps at about 5.5k, 7k, 9k and 12k.

So, I will try to change the microphone position, do one more series of full range readings on and off axis, and then call it a night (its starting to get late here on the east coast).

I'm thinking I won't get very accurate in terms of true frequency response, but I'll at least get a good ballpark idea of how the levels match, and what the tweeters' comparative levels are.
 
judgeschreber said:
I'm thinking I won't get very accurate in terms of true frequency response, but I'll at least get a good ballpark idea of how the levels match, and what the tweeters' comparative levels are.
Click to expand...
Yes, all that.

I don't have the benefit of actually hearing them, but it looks like DX is holding up better off-axis, as well.

Repeat until you're satisfied that what you see is indicative of what you hear, then choose accordingly....
 
judgeschreber said:
OK, I'm noticing there are several resonances in common with all of the tweeters, namely everything from about 5.5k to 12k. All the tweeters have humps at about 5.5k, 7k, 9k and 12k.

So, I will try to change the microphone position, do one more series of full range readings on and off axis, and then call it a night (its starting to get late here on the east coast).

I'm thinking I won't get very accurate in terms of true frequency response, but I'll at least get a good ballpark idea of how the levels match, and what the tweeters' comparative levels are.
Click to expand...

I'd say that those "resonances" are diffraction effects, they are there in T33 measurement, but not in the nearfield. If they were resonances under the dome or whatever, they'd show up in the nearfield measurement. I also do not suggest testing tweeters in the nearfield, but you sure did get a nice curve there. You could try a felt ring around the tweeter, however you want to have an irregular shape since any material used will reflect some sound.

Did you measure the DC resistance of all these tweeters?

The attenuation level at 1 kHz is interesting, the T33 seems to have much more, Perhaps there is not much of an input impedance rise at resonance with the T33?

It is important for the replacement to have about the same rolloff freq and rate so that the amplitude and phase response matches the original tweeter. Then the crossover transition will also match the original. The phase will follow as long as you match the amplitude response since the tweeter section alone is a minimum phase system.

Have you considered cutting down the tweeter faceplate rather than the enclosure to make them fit?
 
Thanks for the advice, Pete. I think I will definitely work on the faceplate rather than the enclosure.

For the rolloff slopes in the XO region, it looks like the DX25 is a bit better than the XT25, and in general seems closest to the original. It is slightly less smooth than the XT25 for overall frequency response (I did a few more measurements last night and this morning, and although the DX is very flat, the XT seems even flatter), but seems like a better choice for level matching and blending with the midrange. Off-axis response is definitely much better with the DX25.

I did some listening today, A/B'ing the DX and XT in mono, and the DX was the clear winner.

I was surprised at how noticeable the 1-2db difference was! The DX was a tad bright, but blended overall much better than the XT. The original tweeter was a tad bright to my ears as well, so at least this is more in keeping with the original design (maybe I'm just too used to the "east coast" tame highs sound). The XT was muffled sounding.

The off-axis response was a no-brainer. I swiveled the "pod heads" away from the listening position, and the XT sounded like it was in another room, while the DX was almost as clear and sparkling as on-axis.

So, looks like the DX is the clear winner (and it's about half the price of the XT, too!)
 
Yes, I was wondering how they sound and you answered that well, thanks.

I'd be curious to hear what you think of the AC25 just as a low cost replacement.

I do agree that the DX25 looks best based on your measurements, very close to the original - nice work!

I don't know how these fit the enclosure, but does the AC25 have more discontinuities from the flange to the enclosure? This would explain the increased ripple in the response.

I have a pair of AC25s here that I'm considering for another project but have only measured the T&S parameters so far. They're an excellent match to each other, and so close to the published specs that it's not even worth posting the measurements.
 
The AC25 were just a little too hot to my ears. The DX25 were slightly less bright, and that, combined with the smoother response, convinced me to go for the DX25.

The AC25 look like they should fit a little better, but none of them really fit into the cabinets during testing. I just had to do as best I could, and then fill in the gaps with a little foam.

Once I decided on the DX25, I modified the mounting plates with a rasp, which worked very well. I then had to very slightly modify the cabinet as well (the rasp worked well here, too). The new notches in the enclosure for the tweeter leads were very small, and could not be seen (and would have no practical effect), in case I end up finding a pair of the original tweeters and going back. I'm not so sure I'd want to do that, though, since these sound even better than the originals, particularly with respect to dispersion, and most of the original tweeters are in danger of dried-out ferro fluid anyways.

Regarding the AC25, I'd say that if you like slightly bright speakers (or if you are willing to add some attenuation to the crossover), they are worth a try. However, since you're going to have to modify the enclosure and/or the tweeters, I'd take some time to try a few different tweeters out.
 
Your measurement at 2.5ft looks about right. There are all kinds of things going on that contribute to the raggedness of the curve: intrinsic driver resonances, room and cabinet reflections, path length effects, diffraction, etc. This is especially true with complex, multi-enclosure/driver systems. As the frequencies get higher, the wavelengths get shorter, and so the peaks and dips tend to appear closer together.

Here are some other thoughts:

1- There are a few ways that a smoother curve can be obtained:

- averaging the curves taken with the mic a slightly different locations.
- a shorter windowing time, (at the expense of accuracy at lower freqs).
- smoothing the data in post processing.

I took a look at the Fuzzmeasure site to see what the software is capable of doing, but I couldn't find any technical documentation. If you have a manual I can look at, I can probably suggest some settings.

2- Keep reflecting surfaces away from the mic.

3- It is important to test your tweeters recessed in a baffle. Otherwise, the edge of the tweeter faceplate will cause significant response abberations.

4- I noticed that all your curves showed an unexpected rolloff at the very top end. Are you using a microphone with a known response curve?

5- Although it costs more, it would probably be a good idea to replace both tweeters with the same type.

In the end, because speakers are complicated physical systems with directional and temporal properties, it really isn't fully possible to reduce their actual performance to just one or two curves. Curves are a useful convenience, but not a complete characterization, unfortunately. I would suggest using the measurements to get you into the ballpark, then using your ears to make the final selection.

I think its great that you are making objective measurements of the tweeters you are considering. Stick with it!

-k

www.kenkantor.com
www.ztamplifiers.com
 
ken kantor said:
I think its great that you are making objective measurements of the tweeters you are considering. Stick with it!
Click to expand...
A big :thmbsp: to that, of course.

I'm certainly pleased to see Pete and Ken participating more here on AK, too.... :yes:
 
Ken, Thanks for all the suggestions!

Yes, having the tweeters not fully mounted in the baffle was a real annoyance, and I know that that must have had a detrimental effect on the curve due to diffraction. Unfortunately, I knew going into this that two out of the three pairs of tweeters would have to be returned, so I couldn't do any sort of modifications to make them fit correctly.

Fuzzmeasure does have the ability to adjust the time window, and it can also average several curves (thanks for these tips!), but I am not sure what "smoothing the data in postprocessing" means.

I'm also a bit perplexed about the rolloff on the top end, especially since all of the curves I've seen for these tweeters (both published and DIY-generated) have them going well past 20hz. The microphone is a Behringer ECM8000, which I guess is a pretty standard measurement microphone, although I have not had it calibrated. I haven't figured out yet whether fuzzmeasure might be limiting the high-end response yet or not.

I definitely replaced both tweeters, not just the one.

Finally, thanks for the encouragement! I only get to drag out the measuring equipment to the living room every once and a while, and I learn a lot every time I do. I just packed it all up earlier today, but next chance I get I will try to take more accurate measurements (with a shorter time window and averaged curves) of the completed system with the new tweeter.
 
Have you tried the "Room Equilizer Wizard"? It's a free download. It has many of the features of expensive speaker measurement systems.

http://www.hometheatershack.com/roomeq/

And, there also is a calibration correction curve for your ECM8000 you can input into REW. The file is called 'newecm.cal' and can be found here:

http://www.hometheatershack.com/forums/downloads-area/19-downloads-page.html

The Hometheatershack website has a forum devoted to mics and calibrations with tons of info on REW and ECM8000 mics which may help you with your measurements.
 
Just wondering, if you look head on at the dome with the AC25 does the front plate look to be a bit off center? I noticed by sighting the gap between the suspension and front plate. Probably less than 1/32, perhaps 1/64"? Both of mine look to be off, otherwise the construction looks very good. I often see minor, and sometimes not so minor construction issues with drivers from Peerless, SCAN, SEAS, Vifa, and more.
 
judgeschreber said:
Fuzzmeasure does have the ability to adjust the time window, and it can also average several curves (thanks for these tips!), but I am not sure what "smoothing the data in postprocessing" means.

I'm also a bit perplexed about the rolloff on the top end, especially since all of the curves I've seen for these tweeters (both published and DIY-generated) have them going well past 20hz. The microphone is a Behringer ECM8000, which I guess is a pretty standard measurement microphone, although I have not had it calibrated. I haven't figured out yet whether fuzzmeasure might be limiting the high-end response yet or not.
Click to expand...

Hi...

What I meant by "smoothing the data in postprocessing" is to smooth out the display curve after the measurement is over by using some kind of point-to-point mathematical averaging. (Terms like, "1/12th Octave Smoothing," refer to this kind of approach.) It looks as if Fuzzmeasure is already doing some of this already, but perhaps can do more. For example, "1/3 Octave Smoothing" would take some of the hash out of the curve, without losing too much detail.

That Behringer mic is generally very good. The rolloff might be coming from the sample rate of your computer's "sound card" or interface box. If you have the option to increase the sampling rate of your computer's audio, give that a try.

-k

www.kenkantor.com
www.ztamplifiers.com
 
Hey, Carl.

From what I understand, REW uses a "sine sweep and convolution" measurement process. This makes it generally impossible to do "windowed" measurements. It looks fine for determining a room's long-term impulse response, but I wouldn't try to get speaker measurements from it, unless one has an anechoic chamber....

-k

www.kenkantor.com
www.ztamplifiers.com
 
Carl, thanks for the tip - I'm always game for exploring some free software that works on Mac! I will play around with it a bit.

Pete, I didn't notice anything like that on the tweeters, but I didn't look too hard either. Unfortunately they have already been packed up and shipped back to the retailer, so I can't take a closer look.
 
ken kantor said:
Hey, Carl.

From what I understand, REW uses a "sine sweep and convolution" measurement process. This makes it generally impossible to do "windowed" measurements. It looks fine for determining a room's long-term impulse response, but I wouldn't try to get speaker measurements from it, unless one has an anechoic chamber....

-k

www.kenkantor.com
www.ztamplifiers.com
Click to expand...

Anybody tried fooling with ATB PC Pro speaker testing gear? I use it and found it to be quite good. It's not as feature filled as REW but for basic SPL response testing it's nice. I replaced their cheap mic with the ECM8000.

http://www.kirchner-elektronik.de/gb-atb-pc-pro.php
 
You must log in or register to reply here.
Back
Top Bottom