Tweeter distortion

AntM

Member
2005-05-27 11:12 am
Edinburgh
Does anyone know of any standard tests that can be applied to determine whether tweeters are damaged ?

The tweeters I'm using (Morel MDT32-s) sound a bit unpleasant (possibly a resonant peak making them very lispy), but I don't know whether they are damaged, or if it is a design problem.

Any help would be greatly appreciated.
 

AntM

Member
2005-05-27 11:12 am
Edinburgh
Around 3kHz, 2nd order, with a series attenuator.
The problem does not seem to particularly amplitude-dependent, so I don't believe it is due to using a Xover freq too low (also the resonant freq is 700Hz on these tweeters).

The problem is still there if I use a simpler 1st order crossover, which seems to rule out a peak due to an underdamped crossover.

When I disconnect the midbass unit, the problem is still there, which rules out driver interaction.

The drivers are not recessed, so I also wondered if the front-plate edge-effect could be introducing a resonance.
 

AntM

Member
2005-05-27 11:12 am
Edinburgh
Thanks for the replies. I'll try changing the source, in case it is an amp problem, I'll also look at changing the louspeaker cables in case this is the problem.

Andy, the attenuator is actually just a resistor anyway, but that reminds me... time for another post on tweeter attenuators
 

MBK

Member
2002-04-03 3:40 pm
Singapore
Hi,

I noticed that small differences in the crossover / the frequency response over a certain frequency range can have a large influence on perceived "harshness" , while more gross aberrations go by unnoticed elsewhere.

In the 1-3kHz region, even 0.5 - 1 dB difference in level, or a small peak (1-2 dB) in frequency response, can turn the system from natural to unpleasant sounding. Given that frequency responses of real life speakers are flat at best within a 3 dB band, an unfortunate little peak might cause your problem. You might try to play with level (L-pad) or purposely tweak the crossover (read, slightly misalign it) to fix that.

If you have measurement capability, this will help identify possible targets. As for level, when I set the level in my tweeters MLS/FFT measurements were practically useless. Unless heavily averaged the differences were hardly visible on screen, and looked insignificant at any rate. But 1 dB less on the tweeter made the difference in my system.
 
originally posted by carlosfm
Excellent post.
Yes, that's it.

I agree, that`s true and very well said.

But as AntM said "very lispy" I assume the problem being rather elsewhere.

The Morel MDT32 is a textile dome tweeter (I`m otherwise not familiar with this brand/model) and I`m sure it is coated (as almost all textile domes are coated) in one way or another to damp it`s inherent resonancies in the upper frequency range.

I have seen textil/fabric dome-tweeters where this coat failed after a few years just because of expose to sunlight (UV-rays) and thereby producing a sharp 10db (!) peak resonance at 13kHz(backed up by own measurements).

Such a thing could well produce what "AntM" described as "very lispy".

As I mentioned, I don`t know the Morel MDT32, I don`t know if it`s new or used (if it has ben exposed to UV or not) , so take this just as a wild guess.
 
It's a new one, so I'd be surprised if that's the problem, but it's an interesting observation all the same.

On MBKs comments, I've certainly noticed that reducing the tweeter level by a slight amount (reducing a 3.9 ohm series resistor to 2 ohms, around 1.5dB change) makes a huge difference to subjective quality, so on this basis I can well imagine that a small peak could be responsible. I suppose the question is whether this can be compensated using a notch filter without compromising the response at surrounding frequencies (or messing up the phase response)
 
AntM said:
On MBKs comments, I've certainly noticed that reducing the tweeter level by a slight amount (reducing a 3.9 ohm series resistor to 2 ohms, around 1.5dB change) makes a huge difference to subjective quality, so on this basis I can well imagine that a small peak could be responsible. I suppose the question is whether this can be compensated using a notch filter without compromising the response at surrounding frequencies (or messing up the phase response)

Is that series resistor before or after the crossover?
Btw, you are not reducing the tweeter level, it's the opposite.
And if the resistor is after the crossover, when you reduce it's value you are also crossing at a higher frequency, which is probably what you should do with that tweeter.
 
Sorry, yes, the resistor was increased to reduce the level.
I've tried both configurations for the series resistor. My instinct is to put it before the HF Xover (in series with the capacitor) - see my other post on this - L-pad

the only real difference I noticed was an increase in treble output around the Xover frequency with the resistor in series with the tweeter - obvious when you consider that the Xover frequency is being reduced due to a higher load impedance. I can understand the reason for the L-pad, to keep the load seen by the filter constant, but actually this only works if the tweeter impedance is constant. Anyway, I'm starting to ramble...
 
I've just completed a MTM TL with a Vifa D25AG metal dome tweeter, crossing over at 1800Hz. I'm not happy with the tweeters performance; it's very harsh sounding. The tweeter's Fs is 850Hz, so I thought using the 2x rule that 1800 would be okay. I'm hoping that part of the problem is Solen cap breakin, but we'll see. I'm using a "professionally designed" 2nd order crossover and the graphs look good. Any suggestions?


Bob