Qu re: Tweeter LCR resonance compensation

[Dissi]

Hello Australia,

I think the statement in the vifa sheet simply means:

If we assume a Vifa D25AG-05-06 (Fs=1.5kHz) with a simple first order crossover (C=6.8uF), the resulting driver voltage is:

Tweeter withC.jpg

We get a 4dB hump at the resonance frequency. And this means 4dB more excursion and 4dB more power in comparison to a filter with LCR added.

Of course John K is right, the hump is also visible in the acoustic response. In practice a notch is often not necessary, because ferrofluid, L-pad and 2nd or 3rd order filters serve the same purpose.

[Jonathan Bright]

Oooooooooooooooookaaaaaaaaaaaay!
Let's try just once more.
The quote in the second paragraph of the first post is lifted straight from the Vifa application notes. I know typos can creep in so I read it several times so that I reproduced exactly what they said. Now as far as I can tell (i) no one has disputed that the LCR network flattens the impedance. (ii) I think we all agree that this allows for a more or less theoretically accurate termination for the preceding cross over.
It is THEIR claim that it also reduces the dome excursion and increases power handling that I am curious about.
Now my first thoughts were the same as john k's; i.e. I didn't see how that would occur but Vifa make the unit (and many others) and they do it for a living and so I did give them the benefit of the doubt that they might know something that I didn't.........and I do that with most people as it is almost always true, ha ha...

So anyone want to address the original question pleeeeeeeeeeeeeeease?
Either there (i) is an answer based in physics (ii) there is a typo in their own notes or (iii) Vifa don't know what they are talking about......

Cheers Jonathan

[Moondog55]

Rereading that piece; and thinking about crossovers I have built; perhaps it is Vifas roundabout way of saying the resistance in the return path should not be too high. That is use a parallel coil with a reasonably low DCR,

[Moondog55]

http://www.d-s-t.com.au/data/Vifa/D26NC-15-06.pdf

Using a .22mH coil from Jaycar 0.8mm wire( 20AWG) whose nominal DCR is 0.3 Ohms seems reasonable as the DCR of the tweeter is nominally 4.6 so about 6.5%
What happens when you use a coil with bigger wire/lower DCR such as the Jantzen 16AWG ( 1.29mm) at 0.16ohms resistance??

[Jonathan Bright]

Yup. That was one my thoughts. I have a lot of data sheets from 20 yrs ago and they seem to like relatively big coils across the tweeters. (I mean keeping the same time constant but reducing the capacitor in proportion. If that is policy there must be a theoretical basis for it as it would be more expensive.) I don't think we can completely discount a linguistic issue. If you make their first sentence run from the present beginning up to the word "excursion" I wouldn't have thought much more about it. But then that would still leave their assertion that a compensation network is essential for high power use.

And the whole flattening of the impedance curve is not THAT crucial with second order series networks and that is what they illustrate, give values for and recommend as a minimum....

Yours, Jonathan

[DrBoar]

If you have a high Q tweeter and a low order crossover you van get far more than a 4 dB hump. I have seen were the crossover has been at 6-8 kHz 6 dB/octave and were the 1 kHz resonance of the dome has resulted in an output not far from the level of the output above the crossover not supressed 12-18 dB as it should be.

[wintermute]

Originally Posted by Jonathan Bright

So anyone want to address the original question pleeeeeeeeeeeeeeease?
Either there (i) is an answer based in physics (ii) there is a typo in their own notes or (iii) Vifa don't know what they are talking about......

Cheers Jonathan

Hi Johnathan, I think that Dissi did just that what he is saying is that it is not the LCR circuit that gives less excursion itself, it is the fact that the crossover works as it should (because there isn't a big impeadance bump stuffing it up) that reduces the excursion!!

ie without the LCR the crossover is ineffective around the resonance frequency causing higher excursion than you want. With the LCR the crossover works as intended and you get more attenuation at resonance and hence less excursion.

Tony.

[Jonathan Bright]

Thanks DrBoar, I can see how the conjugate net work aids the response with low order networks but I am still not sure we have addressed why such an LCR network would increase power handling and reduces excursion and that seems to be what Vifa are claiming.

What we really want is someone au fait with network analysis.....and as yet they haven't surfaced......

[Jonathan Bright]

Hi Tony, sorry our posts crossed. I saw Dissi's post and I understand it. I didn't include all the material from the Vifa App Note. They (Vifa) won't even countenance a first order network for this unit. And I think most posters here are aware that second order networks (at least in series configuration) are largely immune from interactions with the res' peak. I should have posted the whole section so as to avoid the distraction of 1st order cross over// res' peak effects.
You may be right. I'll have a longer think. Just never did enough work on resonance networks..... I sort of feel that the LCR components must be shunting power away from the driver at that frequency for Vifa to insist they be included for high power usage. I mean if they are in parallel and if they end up producing a flat response then at resonance the network must offer a substantially lower impedance and so reducing the power to the terminals.....
The accompanying graphs show an 8 ohm peak with a DC resistance of less than 6. Rough calculations mean that the LCR series resistance must be about 3 ohms to get that sort of result.......if that is the case then it would seriously increase power handling and reduce excursion......and their comments would then make sense. Capeche?

Jonathan

[wintermute]

I personally find the whole "it only flattens the impedance, but does nothing for the power handling" thing counter intuitive...

I really haven't gone into it at an electro mechanical level, but just in laymans terms I envision it (most probably incorrectly) as the LCR network effectively bypassing the tweeter at the resonance frequency (or at the very least shunting off a certain proportion of the signal to earth at the resonant frequency), which I would have thought would reduce the output of the tweeter at that frequency.

I'd be very interested to see the spl graph of a tweeter with only an LCR filter compared to just the raw driver.

Certainly it should be easy enough to simulate...

Tony.