Tweeter group 5

Seas Excel Millennium, North Creek D28, Scan Speak D2904/6000, LPG26T

Frequency Response

Note that the tweeters were not flush mounted. Consequently, a fair amount of edge reflections are present. I know this makes it hard to appreciate subtle FR changes.

Off axis performance

Seas excel millenium

SS6000

D28

LPG26T

Seas millenium CSD

SS6000 CSD

North Creek D28

LPG26T

A note on the above. Because the drivers were not perfectly flush mounted, this causes reflections which show in the CSD, predominantly as the low level "grunge" between 1k and 10k. So this is artifactual. However, the LPG issue is a bit different. There is a metal grill and a plastic diffuser on the tweeter, which creates all sorts of reflections that are "splattered" throughout the CSD plot. So, let's look at the time domain tonebursts.

Linear distortion

At 2k

At 5k

At 8k

At 11k

At 15k

Nonlinear distortion

At 1k

Seas millenium

SS6000

D28

LPG not done at 1k.

At 1.3k

Millenium

SS6000

D28

LPG

At 1.5k

millenium

SS6000

D28

LPG

At 1.8k

millenium

SS6000

D28

LPG

At 2k

millenium

SS6000

D28

LPG

At 3k

millenium

SS6000

D28

LPG

At 5k

millenium

SS6000

D28

LPG

At 7k

millenium

SS6000

D28

LPG

Conclusion

Of these four, the millenium is able to cross the lowest with respect to 2nd and 3rd order distortion products. But higher order products from the excel are somewhat disappointing. If you don't need to cross low, say, in a three way, the D28 and SS6000 are better. And, except for some higher second order products from the SS6000, it to can be crossed low and does better up higher. The linear distortion numbers are similar for all except the LPG, which did slightly worse, likely because of the grill and diffuser. Unfortunately, the LPG also failed to distinguish itself on nonlinear testing, with lots of nonlinearity.

Lets do a quick comparison of the Seas 27TDC and the millenium

nonlinear distortion

At 1k

Now let's look at the reverse where the 27TDC is OVERLAID. Anywhere you see red is excess Millenium distortion.

I hope you can appreciate the significance of these two graphs.

Let's do the same thing for 7k.

Now, we'll flip the two graphs. Anywhere you see red is EXCESS MILLENIUM distortion.

Wow, the millenium is unable to pull away from it's low budget sibling down low and falls behind as you go up in frequency. I have checked, rechecked this with both 27TDC's, and the single 27TDFC at 1k, 1.3k, 1.5k, 1.8k, 2k, 3k, 5k, and 7k. Either the 27TDC/TDFC is really impressive, or the millenium is not. Or, some of both.

What about linear distortion?

At 2k

5k

15k

I don't have quite enough directly comparable linear distortion data to say. The excel is better at 2 and 5k, but not by much as all three drop quickly to -30. At 15k, the TDFC is the winner. So, the excel isn't the clear winner here either.

So what am I saying, explicitly? In these tests, the millenium cannot be said to be significantly better that it's cheaper siblings. These tests are by no means the exhaustive test and summary judgment of these tweeters, but on the whole, I was disappointed by the millenium's numbers. Now, had I never tested the 27 series, I'd be inclined to say the millenium is pretty darned good. And it is. But its siblings, at a much cheaper price, offer similar performance.

As an aside, let me comment on some aspects of the LPG's performance in light of the following observation. It's always puzzled me why so many builders noted problems with their speaker's sound, just by moving the xover point 200-400 hz down or up. If you look at the difference in the second and third order families of distortion products at say, 1.5k and 2k, you don't notice that much of a difference. So are all those builders wrong? Well, probably not. Here's my observation, for what it's worth. Consider the LPG nonlinear distortion plot above, at 2k. Well, at 2k, the offensive higher order products are -60dB and, maybe just as important, they are centered between, roughly, 15k-25k. So, not that audible, except maybe as that added detail, or perhaps, metal dome sound. Now, push the xover to 1.5k. Nothing significant has happened to the second and third order product levels, but now the higher order distortion products have increased in level to -50, they've increased in number, and the center of the rub/buzz distortion products is now at 15k and the spread is roughly between 7.5k and 22.5k. So the spectrum of the higher order products has drifted down significantly, and is likely much more audible.

Just food for thought. Also, by implication, looking at second and third order products may not be that helpful.