Impedance EQ - what's it good for?

[mr_push_pull]

Please explain me when and why is impedance EQ necessary/recommended. I'm referring to speaker impedance equalizing.

[theChris]

passive crossovers have a response based upon impedance.

[Mark25]

stopping tweeters and midrange drivers from producing increased out of band output due to the resonant impeadance peak in their response curves. It's more beneficial with lower order roll-off X-overs. ESP give a good explanation with some diagrams too, here.

[jomor]

Apart of the easier filter design and the tweeter's protection from getting frequencies close to Fs, there is another very important purpose, than many few designers take care of.

Speaker impedance varies with frequency. This is a behaviour ( non resistive ) creates phase shift between voltage and current at the amplifier's output.
And this phase shift varies with frequency too. By using impedance compensation networks (zobel on drivers - RLC on speakers tuned closed to crossover frequency) we
manage to take a resistive behaviour of the speaker, so the amplifier sees an easy load, a load which does not create phase shifts between voltage and current

This makes the amplifier's life easy, driving the speakers up to maximum with no effects of compression and strain.

Look at the impedance and phase plots of a 2-way vifa project of mine seen below. Above 300Hz the total impedance is nearly flat, making the phase being also flat. The woofer's impedance peaks cannot be compensated since they demand huge and very expensive parts, so this idea has been dropped years ago by all manufacturers ( and dyi-ers).





Thats what a speaker being an easy load is all about (sensitivity counts on this too). Impedance equalization doesnt really make the speaker work better. It actually makes the AMPLIFIER work better and perform the maximum.

[mr_push_pull]

That crossover tutorial seems very good. Sorry for being such a lamer and asking this without searching too much first, I just wanted a quick answer, but now I'll get down to studying xovers properly, I used to be among the many that uderestimate them.

[mr_push_pull]

Everyone's attention seems to be captured by the active xover thread, but maybe you'll find a minute to answer one of my questions.
I studied those crossover tutorials you pointed me to, and I can say that now I'm beginning to get an idea what a xover is really all about. With the emphasis put on "beginning to..."
As expected, I have more questions, just more precise ones.

1. I tried to design a resonance compensator for my tweeter (I will link to the specs below). I don't have the tweeter right now, so I decided to design the compensator and xover just as a didactical exercise. OK, I used that Excel spreadsheet, Crossover Designer or something like that, they're all named the same anyway, to compute the values from the published specs. I checked the values with the formulas from some xover tutorial I found on the net and they were ok. But when looking at the compensated impedance response I started scratching my head. The plot had a big null at a frequency that wasn't nowhere near the resonance of the tweeter. Now... the published T/S specs may not be 100% accurate, but I expected some correlation between their published impedance curve and the published T/S params, after all the T/S params are derived from the impedance characteristic. The only potential source of the problem seems to be completely wrong published specs. Btw, I forgot to mention this earlier, the ZDA data was taken form www.thielesmall.com, and visually the plot generated from those values is close to the graph from the Visaton website.
The published impedance plot shows the resonance at slightly below 2KHz, the Fs is 1900 Hz. The impedance at Fs is about 10 Ohms, but Crossover Simulator tells me that the impedance at resonance is around 80Ohms. Wow, a 1000% error!
This leads to another question...

2. ... How does one derive the T/S params from the impedance curve? I mean, I have an impedance plot and I want to get the T/S params. What are the fomulas?

3. I've heard that the impedance compensators can be integrated in the filter itself. Is this correct? And if yes, how?

The tweeter params: http://www.visaton.com/english/artikel/art_170_1_8.html
The frequency response and the impedance plot: http://www.visaton.com/english/bildg...quenzgang.html

[jomor]

Formulas you find on the web dont work as easy and accurately as they claim. Classic example is the Zobel formula. It rarely gives accurate results. The easiest and safest way,
is to use the formula to calculate the initial values of your compensating network, and then measure the result and alter the values accordingly. This always worked for me untill now.

[mr_push_pull]

Actually the Zobel seems to work ok. The problem is at the tweeter's resonance correction circuit.
Here's what I'll try to do. I found somewhere on the web all the stuff needed to simulate a driver in a generic circuit simulation program (like Electronics Workbench or Orcad, whatever), equivalent electrical circuit, the formulas for computing the values of the parts in the equivalent circuit from the T/S params, all that stuff.
I'll simulate the circuit in EWB to see if the resonance corrector works at all, or if the calculated impedance curve resembles the published one. If the later is false, it means that the published specs are completely worthless and I wasted my time for a problem that didn't even exist. See, I'm a big boy, I can work it out for myself

[Mark25]

don't forget the published T/S parameters are for a driver that's been run-in. From what i read here, i gather the published figures get more acurate, as driver cost increases.

[mr_push_pull]

It seems that I've been misunderstood. Now that I reread my last post, that figures.

OK, ignore what I said earlier. Let me rephrase.

Let's say that I take a hypothetical driver, with random T/S params. Forget about correlation between published and real T/S params, it's a hypothetical driver, and I'm not even discussing these correlations. Everything clear so far? Good.

Then I calculate the values for the parts in the speaker's equivalent circuit, and I simulate it in some generic circuit simulation program. I get the typical curve for a driver's impedance.
Then I compute the resonance corection circuit (series RLC) using the long-proven formulas, for instance the ones in this paper http://users.ece.gatech.edu/~mleach/...obel/zobel.pdf
Then I include the corrector to see the effect. Well, instead of seeing the resonance bump flattened, I see a null above the resonance frequency.
That null seems to be caused by the inductance of the voice coil, because if I remove it (L2), the impedance is flat.

To make things clearer, I will send some pictures of what I've done and the results.

Now, this driver is not that hypothetical, it actually is that DT94 tweeter that I want to use, but that's of no importance in this context because now I'm only discussing the corrector.
Now I see that I'm exaggerating a bit, because the characteristic doesn't look as bad as in Crossover Designer, but since I took the trouble to go through all this I thought I should explain what I meant. I expected it to be perfectly flat until the rise caused by the inductance of the coil, and it is not. But it is obviously better than in the uncorrected circuit, not to mention the phase response. That being said, I answered my own question
I had to use a bode ploter for impedance analysis because there's no other way to do that in the program I used (It's not Spice of course). Anyway only the overall shape of the impedance plot counts, not the exact values.


The equivalent circuit: R2, L2, R3, C2 and L3 is the equivalent circuit for the driver, and R1, C1, L1 forms the corrector. Rs is used to measure impedance dirrectly using the bode plotter.


The magnitude impedance characteristic of the uncorrected driver (corrector disconnected):


The phase impedance characteristic of the uncorrected driver:


The magnitude impedance characteristic of the corrected driver:


The phase impedance characteristic of the corrected driver: