Driver variability and crossover consistency issues

[fjhuerta]

Mods - this is a reprint from a previous thread of mine. If you feel it should be deleted, please go ahead.

I decided to repost this because it's a bit off topic from a previous post.

------

(I measured a pair of woofers and a tweeter from each side of a Natalie / Modula tower, because I believed I had issues with the crossover. Here's what I found about the impedance curves).

The tweeters' curve is vastly different, but only on at or below resonance.

The woofers' curve is different by about 0.3 Ohm. I can't understand it. (It shouldn't be an issue, or so I thought. Is this usual?)

So, I took each measurement and simulated it on my crossover design.

This is the FR plot.



Comparing the left vs. right speakers shows that one of them has a midbass hump, and some extra energy on the upper end.

The final impedance curve does vary a bit from the left speaker to the right speaker, too.



I'd guess my question is... is it that normal for drivers to be this different? Should I just live with the difference? (I see no other alternative). Would other drivers be more consistent? I was actually pretty surprised to see how different the tweeters' impedance is (although tweeter response doesn't vary that much).

I'd expect different drivers to behave a bit differently, but not this much. I'd assume this could be a problem.

[digital desire]

Javier,
I can see a slight difference in the graphs.
Can you hear it?
I do not have any experiance in testing, but what is the error range? If you took the same speaker and tested it on three different days, would the graphs lie on top of each other perfectly?

[dlr]

Drivers frequently are not perfect matches

The tweeters' curve is vastly different, but only on at or below resonance..

Not surprising. You don't say what make/model it is. Some are more prone to variation than others.

The woofers' curve is different by about 0.3 Ohm. I can't understand it. (It shouldn't be an issue, or so I thought. Is this usual?)

Different how? 0.3 higher across the band or at the Fs or somewere else?

Comparing the left vs. right speakers shows that one of them has a midbass hump, and some extra energy on the upper end.

The upper end may easily be an SPL variation in tweeters or a woofer. What is your XO Fc and slope? MId-woofers will often have variance at their upper end.

The final impedance curve does vary a bit from the left speaker to the right speaker, too.

This looks more to be from the tweeter low end Z variance.

I'd guess my question is... is it that normal for drivers to be this different? Should I just live with the difference? (I see no other alternative). Would other drivers be more consistent? I was actually pretty surprised to see how different the tweeters' impedance is (although tweeter response doesn't vary that much).

Depends on the make/model. It is possible in some cases to do a DIY mod and improve consistency.

[JonMarsh]

10% variations in a driver's impedance curve is a bit on the high side, but you'd have to plot the driver Z and specify where it varies. In the resonance area of a tweeter, that might be expected, especially if it is a driver using ferrfluid, for example, because sometimes the application is not so precise.

OTOH, with the tolerance of components most people are willing to use, and not using pre-selected and matched drivers, a pair of commercial speakers that track as closely in level as whay you plot would be considered rather good. Many high end manufacturers actually measure and select the drivers in pairs, then adjust the crossover network as needed so that the finished system matches the reference "gold standard". Of course, you don't get that kind of QA process in a $1,000 a pair set of speakers.

~Jon

[dlr]

Looking at the curves more closely...

it looks like the 400-800 SPL offset is due to the impedance variation interacting with the crossover.

It would be helpful to see the overlaid raw driver impedance and SPL measurements of each driver pair if you can.

[Evil Twin]

It would be utterly miraculous if you found two positions in your room for your speakers which tracked as closely in net response as these plotted SPL curves do.

I suggest putting the whole matter in a more realistic perspective...

[Amphiprion]

That looks perfectly acceptable to me. I would measure the drivers individually and attribute the rest to xo tolerances. If you can't tolerate it, use all 1% parts you've hand matched or go active. But really, that is quite good.

[fjhuerta]

Thanks a lot for your help!

I went ahead and built the crossovers with new pieces, and didn't mind too much about the impedance issues. Happily, everything turned out all right - the speakers (both) measure pretty much alike, and they sound excellent.

I'll try not to be so anal about such stuff in the future.

(As an aside, I can't imagine how people who can hear the differences between cables can live with driver & crossover component tolerance issues...).