SpeakerTalk

[mattcambs]

Hi all,
Thought I would introduce myself since I have recently joined after buying a nice condition pair of Ref 103.2. My name's Matt, I'm 33 and this is my first pair of KEF, although I was brought up on Chorales which my dad still owns.
I bought them after advising someone that the best speakers available for £200 were definitely not new ones! Having something of 70s/80s vintage from the likes of KEF, etc I came across this pair on ebay and just couldn't resist.
I have already made some custom stands and I'm very impressed by the sound. Very clean, accurate, musical and just sweet enough to be able to say they have some character. They're being driven by a Nait 3r which I think is a cracking amp. The source is usually a Pioneer DV-575A SACD player (dont laugh it's pretty good actually) but you may have noticed the Wadia 301 in the photo that I am borrowing from a colleague. Excellent CD player!

I have removed the grilles as the midrange sounds cleaner, but I suspect they were voiced for the grilles on as the treble may have a mild peak somewhere. This wouldn't surprise me as the tweeters are equi-distant from 3 cabinet sides. I'm going to confirm this by making some frequency response measurements at work ( I work for invisible loudspeaker manufacturer Amina Technologies). I would also like to establish that the original crossover circuits are in good order and that the pair are still well matched.

I'm really pleased with my purchase and I think they would respond to even better amplification. Doubt you could get better new for £2000 let alone £200!

Thanks for viewing. Apologies for the blurry photos, my phone's camera has some dirt in the lens.

[mattcambs]

Apologies for the size of the photos. I've re sized them in photobucket, but for some reason they are still showing very large

Anyway, I've been measuring my 103.2s to find out a bit more about them before I start thinking about replacing the caps on their crossovers.
I used the CLIO measurement system and made the following:

1. Impulse response of s/n 6079A with the grille on and mid-way between tweeter and woofer (the recommended listening axis in the owner's manual). CLIO inverts phase, but you can see that the tweeter arriving fractionally before the woofer and connected out of phase





2. SPL vs frequency at 1m, 2.83Vrms of 6079A mid-way between tweeter and woofer and with the grille on. Sensitivity is surprisingly high at around 87dB with my test signal. KEF quote 86dB with pink noise.





3. Now the same test with the grille removed (red trace) compared to with the grille (black trace). You can see that the grille makes the HF less smooth between 2KHz - 6KHz and also reduces the SPL





4. Next up was to test the frequency response with the mic on the tweeter axis compared (red trace) to the mid-way position (black trace). Again mic at 1m, 2.83Vrms and grille removed. You can see that with the listening axis level with the tweeter there is a broader dish in the response. I'm not sure whether this will make much difference once the room's contribution is considered, but the following off-axis measurements were made using the on the tweeter axis as the starting point





5. First though, here is a comparison of 6079A and 6079B both at 1m, 2.83Vrms and on the tweeter axis. You can see that they are both very nicely pair matched up to about 16KHz where 6079B's treble rolls off quicker. Not sure why this is because the impedance measurements show no increase in inductance. Probably a mechanical factor.





6. Terminal impedance comparison. You can the large capacitance in line with the woofers.





7. Now a look at the horizontal off-axis frequency response at 0, 10, 20 and 30 degrees. No nasty dips and suggests the speakers might sound better balanced without much toe-in. I'll experiment with this later, but my listening so far with them toed in at the listener does tally with the on-axis measurements in that the treble above 6KHz is a bit pronounced.





8. Now the vertical off axis above the tweeter and you can see that it wouldn't be advised to position the speakers low down unless you could angle them back (which may upset the intended time alignment)





9. Vertical off-axis below the tweeter shows that response changes are less severe compared to above the tweeter listening.





10. Last I measured both woofers in the nearfield to get an indication of their bass response. I placed the mic at 45mm so as to get an accurate measurement down to 20Hz. You can see A and B are very closely matched and that these speakers are actually quite bass light (obvious when you listen). The -3dB point relative to the level at 100Hz is approximately 72Hz (exactly where the driver/box resonance is on the impedance trace). The -6dB point is roughly 55Hz which actually works quite nicely in my listening room which has a room mode at 50Hz. You can also see that the crossover appears to start compensating for the rise in SP@L due to the baffle step at around 200Hz.




So I'm really happy that both speakers are still very closely matched as this suggests that both crossovers are probably still working close to how they left the factory. I actually expected the on-axis frequency response to be flatter, but there is evidence that the design axis is actually 10-30 degrees horizontally off-axis as the tweeter seems to integrate better.

I hope you've found this interesting. I'd love to hear from other 103.2 owners on their experiences.

[SaSi]

Interesting measurements.

I would recommend that you experiment and measure off axis at 5, 10, 15 degrees vertically, above the axis. As the speakers are small, even mounted on a small base, it may be that the response is optimized for the listener position - higher than the tweeter position.

The broad dip in the mid frequencies *may* suggest that some capacitors are off in values and the crossover frequency has shifted causing a gap.

I've had such an experience with a pair of Calindas where high frequency response was smeared on one and bad on the other. Turned out that several capacitors were off by more than 50%. Replacing them changed the sound radically. I didn't do any actual measurements by like what I hear.

[mattcambs]

The 8th graph is the vertical off-axis showing on-axis, 10 degrees above and 20 degrees above tweeter level. If the crossover hasn't shifted from factory norm then I wouldn't recommend listening above the tweeter axis on 103.2s. Somewhere between the woofer and tweeter axis seems ideal. In the KEF manual they state that the speakers should be listened to mid-way between the woofer and tweeter which suggests to me that they are definitely not intended to be placed on low stands.

I'll repeat the measurements with re-built crossovers as I too suspect that some of the caps are out of tolerance. Interesting that both speakers are still very closely matched, though.

I also experimented with speaker positioning in light of these first measurements. A little wider apart and less toe in has really improved things in my room.

[mattcambs]

Not convinced that the Bass alignment is right with these I decided to simulate the woofer and crossover (woofer leg) in LSPCAD. If you look back at graph no.10 you'll see (and I noted) that the bass is rolling off quite early. Also KEF's specs state -2dB at 60Hz which I don't believe my pair are currently achieving.

So using the T/S parameters from the KEF museum for the SP1075 woofer and the crossover schematic from the technical section of this site I built the following in LSPCAD



You can see I have put some switches in place which just allowed me to analayse what the various elements in the crossover were doing. The main thing here was to adjust the series capacitance which should be 600uF.

The graph below shows the simulated frequency response (into half space, hence the high SPL and baffle step compensation in evidence) of the correct 600uF in series (blue line) and then with 1500uF in series (black line). You can see that the 600uF line does not look like my measured nearfield woofer response. I then tried different capacitance values until the bass roll off looked liked my measured response, which it did at 1500uF.



I'm now less confident in the tolerance of the electrolytics in the crossovers. A new set are already on order from Falcon Acoustics, so I'll be rebuilding them and re-measuring the speakers very soon.

[proffski]

Some interesting results, thanks for sharing.
Seeking perfection in audio is like seeking the holly grail.
I've been at it since 1968 and still searching, once in a blue moon I say enough, stop.

Sadly I never do... keeps me off the streets I guess.
_________________
I contend that for a nation to try to tax itself into prosperity is like a
man standing in a bucket and trying to lift himself up by the handle.
-Winston Churchill

[speakerguru]

[mattcambs]

[speakerguru]

http://www.klippel.de/fileadmin/klippel/Files/Know_How/Application_Notes/AN_39_Merging_Near_and_Farfield_Measurements.pdf

and

http://diy-audio.narod.ru/litr/Keele_1974-04_AES_Published_-_Nearfield_Paper.pdf

A simple closed box should be OK.