Tweeter Impulse and Step Response Has Two Peaks and a Massive Dip

[PhoenixFieryn]

Hello, I'm having trouble understanding and diagnosing my tweeter's impulse and step response. The negative peak seems to be greater than the positive, and the positive seems to be split. This was measured using a UMK-1 3 inches away from the driver.

I'm unsure what to make of it, whether it's in phase, and whether this is acceptable or if I need to make some changes. My other drivers' impulse and step response measurements look normal.
The amp's connection to the tweeter has a simple 800Hz HPF from Dayton for protection. The tweeter is a Scan-Speak D2904/710003. It has a big red sticker next to one of its terminals, which I assume means positive. I've used two amplifiers thus far, Topping PA3s and Hypex FA123, with the same results.

If this is fine, then do I look at the first peak for time alignment?

I have attached the REW measurement file as a zip. If anyone can help, I'd greatly appreciate it, thank you!

 

[Salt]

Your results means filter & tweater.
Leave the tweater alone
or
filter appropriate to Your speaker design to see what to get in the end.

 

[Keith_W]

Tweeter impulse response like this may be normal. The polarity may have been deliberately inverted by the designer. What matters is that the impulses of the left and right tweeter are the same, and there are no phase cancellations at the crossover point. If you see cancellation, you might want to consider inverting the polarity of both tweeters to see if it improves.

FYI this is my tweeter's impulse response:

You will see that it is an inverted version of yours. I use the largest positive deflection for time alignment. On mine, the time difference between the first negative peak and the positive peak is 0.041 milliseconds. I don't think I could hear the difference between using one for time alignment vs. the other.

 

[PhoenixFieryn]

Thank you both for the info and help! It seems like the HPF's 180-degree shift slipped my mind . I was too caught up by the DSP's 4th-order LR filters and thought that all the filters in the system were 360 degrees. I really appreciate the image of your tweeter's response!
After some preliminary tinkering and swapping the polarity on the tweeter, the entire system's response is looking much better now!