Damaged KEF 104/2 tweeter?

bcurtin

New Member
I've posted a few threads documenting my restoration of a pair of KEF 104/2s and decided to measure the room response using REQ Wizard. Granted my setup is fairly inaccurate (using my iPhone tethered to my laptop as the microphone), the main objective was to see how these speakers fared in my oddly shaped room.

During the restoration, I clipped the wire in the tweeter that extends from the voice coil to the terminals and soldered a thin wire to bridge break. I measured identical DC resistance between both tweeters and played some test tones and couldn't hear obvious distortion, so proceeded with the repair using the tweeter as is.

After finding high distortion when measuring both speakers, I ran measurements on each one individually and arrived at the following (brown trace at 100% is signal, black trace is relative THD):

Left speaker
Ok0u42Q.png


Right speaker
nOa4kQf.png


There's a dramatic increase in THD at ~2.5 KHz to 10%, which I believe is around the crossover frequency between the midrange drivers and tweeter, on the right speaker only. I didn't track which tweeter was repaired but believe it was the right one based on some visual clues.

Would this sort of behavior be expected for a repair like mine, where the thin metal membrane was worked on with a soldering iron/tweezers? I'd imagine so but curious what others thought before I order a pair of new tweeters. I've ran a few sweeps and this increased distortion is repeatable.

The distortion may be audible to me, but I'm not 100% convinced that I can hear it consistently. I've noticed some distortion while listening but suspected it was the recording.
 
Last edited:
I would re-visit your repair of the tweeter and ensure the wire you used is of ultra low mass and is secured to the lead-in/voicecoil and across the suspension with some thinned nail polish or similar. It is possible the wire is vibrating (there are resonances in that plot) and causing the THD spike. Also ensure the absolute minimum amount of solder is used as it is quite high mass.

Run the plots again afterwards, and if the issue is still there, swap tweeters to confirm it is not X/over related, and run the plots again.

If it's still there- get a new pair of matching tweeters.
 
I would re-visit your repair of the tweeter and ensure the wire you used is of ultra low mass and is secured to the lead-in/voicecoil and across the suspension with some thinned nail polish or similar. It is possible the wire is vibrating and causing the THD spike. Also ensure the absolute minimum amount of solder is used as it is quite high mass.

I used a strand of thin, stranded wire so mass should be low, but had a fairly blunt soldering iron tip so the beads are larger than I would have liked. I trimmed them back with wire cutter to keep them as small as possible. I basically had two solder beads across as short of a length of wire as I could manage.

I used a fast drying epoxy to secure the solder/wire assembly, which covered a portion of the voice coil as well as the suspension. I would be surprised if any piece was loose or rattling around and could have made the suspension more rigid where I applied epoxy.
 
I used a fast drying epoxy to secure the solder/wire assembly

Just a tiny amount I hope.

Unfortunately, the tweeter may be functional, but ruined sonically. Run some low* level test tones (2K and up) through it outboard and see if you can isolate the problem audibly to a resonance or general issue with THD.

* no more than a volt or two.
 
Just a tiny amount I hope.

Unfortunately, the tweeter may be functional, but ruined sonically. Run some low* level test tones (2K and up) through it outboard and see if you can isolate the problem audibly to a resonance or general issue with THD.

* no more than a volt or two.

Strange... playing a 4K test tone through REQ Wizard and using my phone to analyze the spectrum shows no obvious distortion at tolerable levels. Both tweeters show a clear peak at 4K and intentionally adding higher order harmonics (even at 10%) are clearly visible in either tweeter. The sweeps were at higher SPLs so perhaps there are sonic issues when the tweeters are being driven harder?

I don't recall how much epoxy I used, but it was certainly enough to encapsulate the wire and anchor the solder joints to the tweeter. I'd guess that this exceeds what you would consider to be "tiny" :)
 
Perhaps rule out the crossover in one speaker as contributing first? Swap the tweeters over and test- see whether the high freq, high level THD follows the driver?

Or were you doing the tests outboard (no crossover) as I already suggested?
 
Perhaps rule out the crossover in one speaker as contributing first? Swap the tweeters over and test- see whether the high freq, high level THD follows the driver?

Or were you doing the tests outboard (no crossover) as I already suggested?

Thanks John - I'll run some tests this week and report back. I misunderstood your earlier comment and was testing the speakers by passing test tones through a DAC and not bypassing the crossover.

I'll try:
  • Frequency sweeps at various levels to see if it's SPL-dependent
  • Swap L/R channels to rule out the amplifier
  • If distortion is SPL-dependent, directly drive tweeter at similar SPL without crossover
 
If distortion is SPL-dependent, directly drive tweeter at similar SPL

Keep your powers under a few watts and only at X/over freq or above. Be careful.

The good thing is, those tweeters are readily available on the secondary market and maybe NOS as well.
 
Keep your powers under a few watts and only at X/over freq or above. Be careful.

The good thing is, those tweeters are readily available on the secondary market and maybe NOS as well.
Did the volume check tonight and can clearly hear second and third order harmonics kick in at higher volumes. They aren't nearly as audible and measurements show much lower THD at lower volume.

I also tried lightly pressing on the tweeter dome when outputting a 4K test tone and it seemed to suppress harmonics, so I bet something is rubbing around the voice coil.

I've been looking for a spare tweeter on eBay and they seem to show up every once in a while, but KEF T33 tweeters come in a variety of models and the ones I have (SP-1191) are few and far between. I had looked into the Midwest Speaker Repair variant when I thought the one I repaired was toast and this might be the best replacement unless another solo tweeter shows up soon.

Was the old ferrofluid replaced?
Yep, replaced ferrofluid in both tweeters.
 
Last edited:
I wouldnt think its the end of the world if you need to replace them as the Morel replacement is very well reviewed by some people I saw who replaced their tweeters with these.
 
Is there felt/foam behind the dome attached to the pole piece on the T33? That could be causing it.

And you cleaned out the old ferrofluid with paper/card etc?
 
Is there felt/foam behind the dome attached to the pole piece on the T33? That could be causing it.

And you cleaned out the old ferrofluid with paper/card etc?
Yep, there's felt attached to the pole piece. I blew this off before reassembling the tweeters and inspected again after pulling them apart.

Seems like the root cause of the THD is rubbing on or around the voice coil. I pulled out the tweeter and ran some test tones, with measurable THD always present but the peaks varied as I moved the tweeter/coil assembly around. Obviously too far and it rubbed against the base, but it seems like the coil is either slightly bent, some of the pieces from my path job are rubbing, or some slight indentions around the coil where I was working might be making contact and rubbing against the magnet.

I used an exacto knife and tried to clean up the thin metal ring as much as possible, trimming back some of the epoxy, flattening some protrusions/burrs from when I removed the tweeter, and used a test tone to place the voice coil in place where distortion was minimal. I didn't realize how poor of shape this tweeter was in, and took a few photos to show the patch job and knifed ring/burrs.

Soldered wire bridging contact lead and voice coil wire:
5chjXLl.jpg


Metal burr before flattening:
kXBpK98.jpg


Did another THD measurement and it seems like I was able to tame the right tweeter down to ~2 - 3% max instead of 10% peaks as I saw before. Still worse than the left tweeter but I'll take a 2 - 3x improvement.
 
Yep, doesn't look pretty. The inside of the VC former has protrusions and the VC itself looks like it is delaminating from the former here:

tempsnip.jpg

It appears rather 'fat' and may be rubbing the inner or outer gap at that point.
 
Wanted to follow up on this thread - I found the right tweeter to be a bit quiet and knowing that it was distorting bothered me, so I purchased a pair of MT-1191s from Midwest Speaker Repair and installed them today. Listened to some tracks that I was familiar with and noticed that the speakers seemed brighter but in a good way, and measured distortion to be < 0.5% across all frequencies now so the previous tweeters weren't doing these speakers justice.

Very happy with this upgrade and would recommend these tweeters to anyone looking for a replacement.
 
Back
Top Bottom