How to fix tweeter distortion problem

[paulspencer]

When I recently purchased a Celine Dion CD for my mum, I played it for her on my speakers, cranked it up a little but was alarmed at the distortion I was hearing through the tweeter!

It was a tinny ugly sound in the vocals coming through the tweeter. My initial thought was to blame the recording, as I thought the level was in the range they could handle. However, it is in the lower end of the range handled by the tweeter. On headphones I don't hear the same sound, so I suspect it might be in the crossover.

This surprises me as the xo point is 3.5k with a 2nd order filter. Vifa recommends a 2nd order xo at 3k >>> 100 watts power handling. So I would think it should be fine. The Tweeter is D25AG. The amp I'm using is an old Akai with only about 50 watts rms at most and I tend to have the treble at -4 or -6db (I should really pad the tweeter). I would have expected that set up this way the tweeters should not be distorting at all!

The crossover design comes from a Jaycar kit. It is described and has a schematic on my site (see link in signature).

Any and all suggestions welcome as to what the problem might be and how to fix!!!

[4real]

Wasn't the amp clipping? That might very well be the source of the distortion. Looking at the specs, the tweet should do just fine.

[phase_accurate]

The amp I'm using is an old Akai with only about 50 watts rms at most and I tend to have the treble at -4 or -6db (I should really pad the tweeter). I would have expected that set up this way the tweeters should not be distorting at all!

I am convinced that things will be better as soon as you have padded down your tweeter.

1.) This will not only reduce the overly loud upper end that you currently reduce with your treble control, but it will also reduce the low-frequency content that makes it through your crossover.

2.) Since the treble control is more effective at 10 kHz than at 3.5 kHz you will end up with a hump above the crossover frequency, making the sound harsh.

Regards

Charles

[Zaph]

Originally posted by paulspencer
Any and all suggestions welcome as to what the problem might be and how to fix!!!

I know you said it's the tweeter, but it's probably the woofer. Newer versions of the P17 are not nearly as smooth as they used to be. The chances are the crossover is not taking care of peaks in the woofer.

I've attached raw in-box response curves of these drivers. Note how the P17 has a peak at 8khz.

I can model it and see if there are any problems.

I do have a question though. You mentioned you use the tweeter without padding. Did you build the tweeter crossover without the two padding resistors? Also in that schematic, what are the values for P1 and P2?

Building woofer two's crossover to be like woofer one will remove all baffle step and sound like midrange glare. Bad idea.

This is probably a better design:
http://www.kaiaudio.com/diy2001/kaiopen.html

[rabbitz]

Celine Dion........ nuf said.

Pad your tweeter. I've heard a set with 3.9ohm for the series and 8.2ohm for the parallel and works well with the D25AG / P17.

[Zaph]

Originally posted by Zaph

Building woofer two's crossover to be like woofer one will remove all baffle step and sound like midrange glare. Bad idea.

On the other hand, Celine Dion sounds like midrange glare on any system.

[Mark25]

sometimes this type of sound is on the CD itself, there is a classic example on Mellisa Etheridge's CD "never enough". A speaker with an over accentuated mid/lower trebble just brings it to your attention more.

The trebble control only rolls off the trebble top end, the L-pad would cover the whole tweeter range, which would remove the afforementioned over accentuated mid/lower trebble.

[paulspencer]

Thanks for all the suggestions everyone. Since posting I have:

1. Done some more careful listening to different CDs, listening up close to the tweeter as well as further away.

2. Changed the xo back to the orignal 2.5 way design, which means effectively BSC is now included - a subtle improvement although I think it needs more compensation given the distance to the rear wall, but possibly also affected by a midbass dip in room response (will measure later).

3. One tweeter is padded by 3db and the improvement on vocals is noticeable - more natural

The ugly sounds coming through the tweeter with Celine Dion persist. With some careful listening I concluded that it wasn't the amp clipping as it happens at only 1 watt. The sound also occurs with sounds that aren't loud at all ie. it's not distortion originating from the tweeter or it would consistently follow the loudest parts covered by the tweeter at the low end of its range. I'm pretty sure that it is in fact a poor recording. It's sad that so many recordings have unnatural treble sounds added. I think this tweeter is just unforgiving. With nicely recorded CDs it sounds magical. I have taken back my original thoughts about this tweeter, that it is too bright and harsh.

Originally posted by 4real
Wasn't the amp clipping? That might very well be the source of the distortion. Looking at the specs, the tweet should do just fine.

Nope, I have eliminated the amp.

Originally posted by phase_accurate

I am convinced that things will be better as soon as you have padded down your tweeter.

I'm also convinced now having tried it. It sounds better than using the treble control.

Originally posted by Zaph


I know you said it's the tweeter, but it's probably the woofer. Newer versions of the P17 are not nearly as smooth as they used to be. The chances are the crossover is not taking care of peaks in the woofer.

I've attached raw in-box response curves of these drivers. Note how the P17 has a peak at 8khz.

I can model it and see if there are any problems.

I do have a question though. You mentioned you use the tweeter without padding. Did you build the tweeter crossover without the two padding resistors? Also in that schematic, what are the values for P1 and P2?

Building woofer two's crossover to be like woofer one will remove all baffle step and sound like midrange glare. Bad idea.

This is probably a better design:
http://www.kaiaudio.com/diy2001/kaiopen.html

John,

I'm convinced it's not the woofer as the sounds can be easily identified as coming from the tweeter.

I can't see any significant peaks at 8k in that plot! But yes, I know the P17 is not as smooth as it could be. I wonder if it would benefit from some midband eq. The xo came from a Jaycar kit which was designed with Leap. I haven't designed a xo, I have only modified that one as described. I didn't build that xo, it came pre-assembled. There were no padding resistors for the tweeter.

P1&2 are polyswitch protectors, which I'm sure you are familiar with. I considered them unnecessary, and if anything a detriment to the sound, thought not sure if it's audible. I have taken them out - I don't driver them so hard that they should be damaged - they lose fidelity before they blow up. I like to operate everything comfortably within limits.

I can't say the 3db difference means "midrange glare" vs sounding just right - the difference is subtle.

Why do you say the Kai audio design is probably better? Any particular reason? I have seen this one before, from memory it is a standard MTM with BSC in the xo. Perhaps it would be better if the xo I'm using was a hack job, but it was designed reasonably well, although I think they should have padded the tweeter - this tweeter needs it as it is revealing. Note: it's ok without padding on the best recordings I have, it's on pop recordings where they overdo the treble and it sounds tizzy.

I plan to get my speakers measured at some stage. I have also thought of putting in the Vifa XT tweeter. I think it should be ok with the high xo point of 3.5k, and it might require experimentation. But I suspect I might like this tweeter a little more than the D25AG. Although I have heard it often calls for a 3rd order xo. In the meantime, any suggestions that might be worth trying with the xo?

Originally posted by rabbitz
Celine Dion........ nuf said.

Pad your tweeter. I've heard a set with 3.9ohm for the series and 8.2ohm for the parallel and works well with the D25AG / P17.

I used 1ohm for the series and 10ohm for the parallel - the closest I could find to the values given for 3db padding. 10 W resistors.

Originally posted by Mark25
sometimes this type of sound is on the CD itself, there is a classic example on Mellisa Etheridge's CD "never enough".

Mark, I think you hit the nail on the head. On my better recordings, the Lpad made a significant improvement on the vocals. On crappy pop recordings, there is nothing you can do cos they overdo the treble way too much.

[ekaerin]

Hi all,
I have experienced this with Celine several times. For a while I could
not stand her voice at all. I even thought my tweeters were gone.
However, in my case it boiled down to the amplifier. It sounded crap
even at low levels. I have a BIS recording of a female vocalist that kills
90% of the amps I have listened to.
One of the few setups that produced the BIS record flawless was Naim
electronics with Totem Acoustics small two way.

But also I think tweeters should be padded. In my ongoing 2way it is
5dB down but I'm still in progress with that guy.

So even if the amp is OK with phones (mine is too) don't count it out.

And yes, most celine recordings are kinda overcompressed etc and does
sound thereafter.

/ Regards / Mattias