Very strange tweeter behaviour

[ToTo Man]

I was testing a recent tweeter purchase yesterday (Celestion HF2000) and am flummoxed by some very odd behaviour.

I ran an initial periodic pink noise signal through it and the readings appeared ok, it's slightly less sensitive and exhibits earlier HF rolloff than some of my other HF2000 units but such variance is par for the course with these tweeters.

Running sine wave sweeps, however, I noticed the measurements didn't match the initial pink noise measurement, and not only that, each repeated sine wave sweep produced a different HF response! It wasn't the just overall SPL that was different but also the shape of the curve:

I checked and re-connected the wiring in case I had a poor connection on a crocodile clip etc but the problem persisted. I then swapped this HF2000 out for some of my other HF2000s and these measure consistently, which rules out my DAC, amp, crossover and wiring as potential sources of fault.

Exploring things further today, I've been playing about with periodic pink noise and notice that as I gradually increase the SPL of the pink noise by 1dB at a time, the tweeter's output sounds dull, but then suddenly springs into life when I reach a certain SPL. As I decrease the SPL, the tweeter's output remains strong, but when I reach a level that's about -7dB or -8dB down from the level at which it suddenly 'came to life', it abruptly goes dull again.

It's as if it has a lazy or seized voice coil that needs an electrical kick up the arse to get it working properly! Has anyone else encountered this sort of behaviour from a tweeter?

 

[Fatmarley]

It's as if it has a lazy or seized voice coil that needs an electrical kick up the arse to get it working properly!

I haven't a clue, but that was my first thought.

Is it possible to take it apart and have a look?

 

[ToTo Man]

I haven't a clue, but that was my first thought.

Is it possible to take it apart and have a look?

Not unless I want to sacrifice it! Even Dave of VintageGale won't attempt to repair an HF2000 as apparently centring the voice coil is nigh on impossible. I'm probably going to return it for a refund, I just want to try and learn as much as possible from it before I do that. Dave sent me some photos of HF2000s with corroded voice coils, apparently this happens when the speakers are stored in damp conditions. I've had my share of HF2000 with wonky frequency responses and unacceptable levels of harmonic distortion but this is the first I've had that does not react linearly to a change in SPL!

 

[davidsrsb]

Ferrofluid?
Try the single AA battery test and try both ways to see if the dome moves in and out evenly with no jerking or rubbing noises

 

[ToTo Man]

Ferrofluid?
Try the single AA battery test and try both ways to see if the dome moves in and out evenly with no jerking or rubbing noises

No ferrofluid in the HF2000. To do the battery test I assume I'd need to remove the mesh grille which is glued on and, if I'm returning the tweeter to the seller for a refund then I probably shouldn't do that.

 

[a.palfreyman]

Distorted v/c former rubbing the pole piece? You'd expect 3rd order harmonic distortion but it could be that giving it the beans frees it until the output drops again and then it starts to rub again. Not sitting in the gap ie too far back or forward? You'd expect 2nd order harmonic distortion.

 

[cooky1257]

Reflow /resolder the solder on terminals/vc wire.....?

 

[Fatmarley]

The Dayton Audio DATS has a Rub & Buzz test. Also, an impedance sweep with the DATS may help to diagnose the problem.

Have you checked the resistance with a multimeter?

 

[davidsrsb]

Reflow /resolder the solder on terminals/vc wire.....?

Not if he wants to send it back

 

[ToTo Man]

DCR is ok at 4.1 ohms, but from previous experience the DCR of an HF2000 doesn't give much insight into its operational condition.

I removed the tweeter from the loudspeaker and tested it on its own, and I cannot replicate the behaviour it exhibits when installed in my Ditton 66. Running tests from 45dB SPL to 80dB SPL in 1dB increments, the SPL and frequency response remains consistent and doesn't suddenly change at a certain SPL (well, it does change a little once you get down below 55dB but I suspect this is more to do with the mic's own frequency response and noise floor at such low SPL!). Harmonic distortion does increase as SPL increases, this HF2000 isn't the cleanest I've measured but it certainly isn't the worst. Remember this tweeter is designed to be crossed over no lower than 5kHz.

My dad suspects the tweeter is behaving normally now because I'm measuring it on it's back, pointing up to the ceiling, as the voice coil is now being evenly supported. I don't think gravity would have any effect on such a low mass unit, but I'll rotate it 180 degrees when I reinstall it in my Ditton 66 and see what effect it has. If the strange behaviour re-emerges then the finger of blame will be pointing at the Ditton 66 and not the tweeter.

5kHz-20kHz periodic pink noise 1dB increments:

2.5kHz-24kHz sine sweeps 5dB increments:

THD from 2.5kHz-24kHz sine sweeps 5dB increments:

 

[Fatmarley]

DCR is ok at 4.1 ohms, but from previous experience the DCR of an HF2000 doesn't give much insight into its operational condition.

I was just thinking about what Cooky said, and if there was a fracture, but still a connection it's possible it could have had a very low measured resistance.

 

[a.palfreyman]

Distorted baffle / seating distorting the frame? Or possibly some 5hit in the magnet gap and removing the tweeter has dislodged it? You have to put it back as before and see if the problem returns!

 

[a.palfreyman]

Interestingly going back to what Cooky said about the soldering of terminals, I bought a pair of mid-drivers from an HB3 some years ago. One measured fine (about 7ohm dc) whereas the other was all over the place. 12-16 ohms, but appeared to work and was similar volume (by ear). The following day, it measured O/C, but it still worked and then measured about 12 ohms after use. A bit of head-scratching lead me to believe that a tinsel to coil-end solder joint was dry so applied a hot soldering iron to the join on the back of the cone and hey-presto it measured fine and has done ever since.

 

[ToTo Man]

Reinstalled the offending HF2000 into my Ditton 66, but this time rotated 180 degrees. The problem re-emerged, - once I reach a certain SPL on periodic pink noise the HF response suddenly jumps up by +5dB, and with sine wave sweeps there is variation in the FR depending on what SPL I run the sweep at. At higher SPLs there is also more harmonic distortion than there should be between 3kHz-10kHz. I'm puzzled as to why this tweeter performs better when out of circuit than in, - perhaps there's a fault with it that's putting stress on the Ditton's crossover? I told the seller on Saturday that I'd let him know if I was returning it for a refund so I best cut my losses, pack it up and get it off to him in case he thinks I'm at it. Thanks for the input guys, - every day on pfm is a school day!