Speaker (Kef 104) tweeter polarity

[pjn]

I may be overthinking this -

I have been lucky enough to own a pair of Kef 104/2 speakers for the last 30 years. About 10 years ago, I noticed that they didn't really seem to be as good as they had been and on inspecting what Audyssey was trying to do to correct them, realized that the tweeters were toast - dried up ferrofluid.
So, I duly replaced the tweeters with mdt29 tweeters, paying careful attention to the wiring. Absolutely astonishing improvement! It is amazing what you get used to in your own little audio bubble.

Every so often, I run Audyssey calibration to see what is going on and correct for changes in room acoustics. It tended to complain about a speaker being out of phase and I tended to ignore it.

I ran these tests - https://www.audiocheck.net/audiotests_polaritycheck.php and it seemed that the Audyssey might be right.
I used Speaker Pop to test polarity more precisely on each driver (well, at the bass port in the case of the woofers) - https://studiosixdigital.com/audiotools-modules-2/speaker-test-modules/speaker_pop/
Indeed, the tweeters were out of phase with the other drivers. So I reversed their wires and they are now in-phase with everything. And sound better? But that is subjective and not blind or dispassionate testing.

Now I read that more complex passive crossovers may well reverse driver polarity and so those drivers should be connected to the crossover in reverse polarity.
But does this not correct the driver so that it will perform in-phase with the other drivers? Isn't the bottom line what happens to the air that the speaker units are moving and nothing else?

So I have a pair of Kefs which, to my ears, sound pretty good with excellent imaging. Am I doing something wrong by having the tweeters in-phase even though they are connected (apparently) incorrectly?

I'll welcome any corrections to my thinking.

 

[fineMen]

The reversed polarity, as needed for the better sound only tells, that the replacement doesn't snug fit. This most probably might be due to a differing resonance frequency or damping of the tweeters. Nearly all OEMs use the same schema for polarity, hence this conclusion.

But if the sound on the reference axis* can be equalized to flat and at the same time besides the reference axis there is not too much sound power, You are good to go, congrats!

*reference axis, in the vertical I mean, moving the microphone in up-down direction

 

[pjn]

The reversed polarity, as needed for the better sound only tells, that the replacement doesn't snug fit. This most probably might be due to a differing resonance frequency or damping of the tweeters. Nearly all OEMs use the same schema for polarity, hence this conclusion.

But if the sound on the reference axis* can be equalized to flat and at the same time besides the reference axis there is not too much sound power, You are good to go, congrats!

*reference axis, in the vertical I mean, moving the microphone in up-down direction

Thanks for the encouragement! My feeling is that what I hear is the most important thing, but as the majority on here will say, you can't trust your ears - I would add that you can't trust your brain either (a machine with a vast capability of lying or just making stuff up when lying won't do)

 

[fineMen]

I see, was under the impression, that the tests You mentioned involve a qualified microphone. That could be a valuable accessory for Your stereo anyway. It gives access to other fields of optimization also.

But it needs to familiarize with the science of the field. It is clearly not for everyone.

Maybe You better ask in the DIY department for personal assistance. Fortunately for someone living close by. Sorry.

 

[bcurtin]

@pjn would you mind sharing the measured frequency response of your KEFs assuming Audyssey outputs this? I have a pair of 104/2s that I've refurbed and have been measuring for DRC and would be interested to see what the mid range and tweeter frequency response of your speakers are, with and without the mdt29s if you have the data

 

[pjn]

@pjn would you mind sharing the measured frequency response of your KEFs assuming Audyssey outputs this? I have a pair of 104/2s that I've refurbed and have been measuring for DRC and would be interested to see what the mid range and tweeter frequency response of your speakers are, with and without the mdt29s if you have the data

I don't have the "before" results, except that the room correction was throwing an awful lot at the high end to compensate for the basically dead tweeters. Here is what happens now. I had vague intentions of doing a capacitor and doughnut refurb over xmas but life got in the way! As I can't hear a thing over 9-10k, I'm pretty happy with this profile.