A Great Day With Lots of Data

[toddalin]

Wow, as useless as those images are (not your fault - there are no scales on either axis... ), the "after" clearly shows (to me... ) that the autoEQ is trying to compensate for the loss of LF response related to your DIY in-line resistor.

Check it out...

L (after) - LF has a spike where there was a gentle peak, and also a flat-line (200Hz...?) correction (boost) where there was a swale. The MF and HF are mostly unchanged.

R (after) - LF, ibid. A pronounced peak is now introduced where before there was a mostly flat correction; and, the swale (around 200-400Hz...? ) is replaced by a flat-line correction (boost).

What is the non-attributed bottom window below the two channels - is that the overall reading, before-correction? If so, you can sure see the compromised LF response (probably...) relating to that DIY in-line resistor...

No resistor in place in any of the pics. This is simply the result of minor crossover adjustments and placement and toe-in relative to the rear wall.

Actually, if you look close, the second set are better in the bass area. The "spike" you note represents an increase in just one band whereas the "gentle slope" shows that the correction is actually over a wider range of bands.

The center one below the pair represents the eq for the center channel, below that are the four surrounds and two presence channels. All channels get the eq treatment. I was focusing on the L/R as that's where we're working.

[boputnam]

No resistor in place in any of the pics.

OK.

Actually, if you look close, the second set are better in the bass area.

Yes, I thought so, too.

But are these the corrections, or the response?

The center one below the pair represents the eq for the center channel,

Again, are these the corrections, or the measured response?

[toddalin]

OK.

Yes, I thought so, too.

But are these the corrections, or the response?

Again, are these the corrections, or the measured response?

The graphics show the applied corrections, so it stands to reason that the actual frequency response is the conjugate of what you see.

The resistor on the woofer came about as a matter of experimentation and I don't actually run them this way.

Initially I put a 16-ohm L-pad on the woofer to dial in its volume and noticed that as I increased the resistance, the mid-range volume came up smoothing out the dip. I reported this and most said I was crazy.

When I sent Zilch one of the crossovers I asked that he also perform this test. He said that he did and could not confirm my findings. He reported that it only made the woofer quietier.

This was merely a test to show that I am in fact sane and the speakers behaved just as I had reported.

[grumpy]

This was merely a test to show that I am in fact sane and the speakers behaved just as I had reported.

This is just too weird. We're missing something. The impedance run showed that there
was a bad assumption somewhere. I ran a few sim plots that we should be able to
roughly replicate looking at WT2 output again and measuring voltage drives at each
driver.

Note that the last plot below, has the impedance scale moved up 20 ohms...
which is exactly what we saw w/WT2 when we thought the hook-up was ok (nominal
shifted up 20 ohms across the band). What I expected was a rise in the impedance of
the low-end only (4th plot), and a drop in low-end level (3rd plot).

Also note that no cabinet tuning or LF speaker resonance is modelled here. The effects
should still be grossly similar.

-grumpy

[toddalin]

This is just too weird. We're missing something. The impedance run showed that there
was a bad assumption somewhere. I ran a few sim plots that we should be able to
roughly replicate looking at WT2 output again and measuring voltage drives at each
driver.

Note that the last plot below, has the impedance scale moved up 20 ohms...
which is exactly what we saw w/WT2 when we thought the hook-up was ok (nominal
shifted up 20 ohms across the band). What I expected was a rise in the impedance of
the low-end only (4th plot), and a drop in low-end level (3rd plot).

Also note that no cabinet tuning or LF speaker resonance is modelled here. The effects
should still be grossly similar.

-grumpy

"The impedance run showed that there was a bad assumption somewhere."

I think that the "bad assumption" is that the mid and tweeter determine the overall speaker impedience for signals in their range. I think that because the woofer is directly coupled to the amp (albiet through a coil of wire), it's impedience is the primary speaker impedience regardless of the frequency.

The mid and tweeter are only indirectly coupled to the amp through a capacitor, so while their impedience affect their crossover frequencies, it doesn't do so much for the overall speaker impedience.

Recognize that even though we can't hear the woofer over a couple thousand Hz, your impedience plots clearly show woofer resistance (impedience) at the upper frequencies.

Of course I could be totally wrong.

[grumpy]

Heh... well, I'm cool with you disagreeing , but that's part of what a crossover
normally does. LPF section with resistive/simulated 2235 is about 50 ohms at
3KHz (almost all due to the 2.2mH coil). From about 1KHz up, its all high-passed
components/drivers.
-grumpy

[toddalin]

Heh... well, I'm cool with you disagreeing , but that's part of what a crossover
normally does. LPF section with resistive/simulated 2235 is about 50 ohms at
3KHz (almost all due to the 2.2mH coil). From about 1KHz up, its all high-passed
components/drivers.
-grumpy

Yes..., until you add the Zobel network, then as WT2 demonstated, the impedience of the woofer stays pretty flat throughout the audio range even though the frequency is rolled off.

I imagine that without the Zobal, we would see something like the plot that you expected (i.e., the resistor primarily adds to the woofer load at low frequencies).

But I could certainly be wrong.

[grumpy]

Plot above -is- with the zobel (which is where the shallow dip below 1KHz comes from).
The zobel can't reduce the viewed impedance of the 2.2mH inductor... it's in series.
The impedance equalization (zobel) occurs -after- the 2.2mH inductor, so that the inductor
will see a mostly resistive load and the crossover will behave a bit more predictably.

If you take out the 2.2mH, you're right, the impedance "curve" should more
or less look flat, if the speaker/zobel combination is implemented well.

If I scaled the WT2 network impedance curve more usefully, I think this would be
easier to see... perhaps. It would take a re-run, as the resolution was set low.
-grumpy

[WTPRO]

Hi Grumpy

I see you are using the new WT2 software. If you look in the options pull down, you will find that you can now select the pulse, MLS, noise and chirp real time modes for generating fast impedance plots with lots of resolution. The TS tests themselves will revert back to sine mode for maximum precision and noise rejection. We enabled these modes for our customers that needed ZMA data for the ICD crossover designer. If you look at the WT2 simulator you will also find that you can specify a series cable resistance and see the modified response. However, I just found the software clips the cable resistance value at 5 ohms, and you wanted 20!

Best regards,
WTPro

[grumpy]

Yep... Just downloaded the update a few days ago Cool features!
Just got a brief taste at home. Looking forward to exercising 'em further.

Didn't bring a copy of the new s/w with me to Todd's,
only the original CD (so what you're seeing was old .woo files, collected
at Todd's place, then imported into the newer version for display). -grumpy

[subwoof]

The gray magnet is the older one. It was only used for about a year until the engineering changes and new coil ( to become the 2225 ) were introduced. The magnets ARE identical from the outside but the internal configuration was alluded to be different in one ( or more ) of the MANY "debates" ( happy searching ) posted on this site.

A sure way to check is to remove the foilcal and look at the recessed casting number that is on the back plate. As far as I know, NO listing of these numbers is available anywhere from any source other than a call to JBL and even then it would be a head-scratcher to anyone NOT employed in magnet design 25+ years ago.

sub