Strange response from Dayton AMT Mini-8

[zx82net]

I'm driving a AMT Mini-8 with a 4.5kHz HP filter and seeing a 8dB dip in the response around 5.9kHz, and I can't work out where it is coming from. (response curve attached)

So far I have tried:

• 2 different units, one in baffle, one unmounted: same dip in both cases, slightly different frequencies
• a different channel in the same amp: no change to the dip
• modifying baffle size, no change in frequency of the dip
• a different silk dome tweeter, same crossover: doesn't have the problem

Can anyone help?


The tweeter:

https://www.parts-express.com/Dayton-Audio-AMT-Mini-8-Air-Motion-Transformer-Tweeter-275-095

The HP filter:

https://www.parts-express.com/Dayton-Audio-4.5k-HPF-8-High-Pass-Speaker-Crossover-4-500-Hz-12-dB-Octave-260-1918

[bassman_soundking]

Looking at the response graph from mfg there is a dip, but it looks centered closer to 7k.
Your looks like an eq cut with narrow Q

[DeZZar]

Have you measured the tweeter without the HP filter? Easiest way to narrow down if its the tweeter response itself or introduced somewhere else.

And by the way - given you obviously have the capacity to take your own measurements I'd encourage you to head down the path of designing your own crossover - this HP filter isn't worth the materials its made from....crossovers and filters are application specific. Its probably a 2.2uf cap and a 0.56 coil? That's the mathematically derived 4.5khz xo and it fails to account for the tweeters own frequency response, impedance response, diffraction, offset, time-alignment and so on....

[DeZZar]

What distance have these measurements been taken at and what gating has been applied?

[zx82net]

What distance have these measurements been taken at and what gating has been applied?

~30cm, no gating.

I was slightly nervous about connecting the AMT to the amp without any crossover, but I can rustle up a capacitor from somewhere.

Agreed those crossovers are far from optimal, but I'm not actually using the crossover as a crossover, it is just a signal splitter between HF and LF channels in an active 3 way system. The roll-off for the driver is actually controlled by the DSP. There's a diagram here:

https://techtalk.parts-express.com/forum/tech-talk-forum/1486938-build-thread-whacky-weekend-not-linkwitz-3-way

[Wolf]

Is this measured with or without mid in play? Mid could have a breakup peak causing the cancellation. If not connected, then it's an artifact of the tweeter, which last i knew, was not good below 7kHz anyway.
Wolf

[danmarx]

Is the measurement dead on-axis with the tweeter? I've seen something similar but only when the mic is directly in the front of the tweeter. Trying moving off axis left or right 10° or so and then up or down and see what you get. Also try moving the mic further away. This looks maybe specific to mic placement. And while it's likely real for that exact location, not necessarily indicative of how the tweeter behaves or sounds as a whole. Also agree with the no crossover measurement, get that out of the loop to make sure it's not interacting in a weird way. Also, you can set the start frequency of your sweep in REW to ~2 kHz.

[zx82net]

OK, I've now run the tweeter in for about 8 hours and done a load of measurements:

• with and without crossover
• horizontal axis scan
• vertical axis scan
• distance scan
• baffle modification
• gating on/off

All of these are attached.

Note: only the tweeter was driven during these tests

Anything else to try? I'm beginning to think this is a manufacturing problem.

[zx82net]

More images:

[zx82net]

No smoothing:

[Billet]

I know you may be trying to clone an existing design, but can you simply cross the tweeter over at 8k or so. I'm pretty sure that 3" Tang Band will play as high as you need.

[DeZZar]

Agree, continue that 7K roll off or try a different tweeter.

[danmarx]

Hard to dispute that kind of data, that's definitely these tweeter's response characteristic. It doesn't match the datasheet very well, but then performance can vary in production I suppose. This might be a bit too far off to be acceptable though. At least for a drop-in for an existing design that was presumably based on some other performance characteristic of that tweeter. If you're up for a redesign of the crossover, you could make them work. I think I would just shelve this AMT in this case and find another tweeter. If you're set on the AMT, maybe try the bigger brother, the AMT2-4. Slightly larger face, but plays lower based on specs. Not sure about the AMTs though, never done a build myself with them, done a couple ribbons/planars, they definitely like a higher crossover point compared to domes. Generally speaking. Cool little speaker too, hopefully you can figure something out.

[Chris Roemer]

Oooh, I'd say there's nothing wrong w/your measurements.
If you overlay the mfr. curve, they look pretty darn close.
The mfr. curve shows a -7dB dip @ 6.5k, rolling off below 5.5.
Yours have a -10dB dip @ 6kHz, w/the rolloff near 4.5k.

Your shoulder and dip are about 10-15% LOWer than advertised.
You've nearly got an extra 1k of pass-band, but that's not uncommon.
It's equivalent to a dome w/a 1k advertised Fs that measures in closer to 850-900.
I'd say that's often typical.

[zx82net]

Oooh, I'd say there's nothing wrong w/your measurements.
If you overlay the mfr. curve, they look pretty darn close.
The mfr. curve shows a -7dB dip @ 6.5k, rolling off below 5.5.
Yours have a -10dB dip @ 6kHz, w/the rolloff near 4.5k.

Your shoulder and dip are about 10-15% LOWer than advertised.
You've nearly got an extra 1k of pass-band, but that's not uncommon.
It's equivalent to a dome w/a 1k advertised Fs that measures in closer to 850-900.
I'd say that's often typical.

Hmm, ok I guess that matches. Their datasheet graph is a little more favorable, there is a 5dB dip, but it looks more like a ripple there. It seems like their 3.5kHz to 40kHz spec is way off the mark. Fortunately the full range I have as a mid doesn't start becoming too directional until a little above 7kHz. I guess datasheets are datasheets, but I kind see this messing up a few designs.