BG NEO 3 FR optimization

I always found that NEO 3 pdr is a nearly perfect tweeter excepted the frequency response.
When I studied the construction, I found that this tweeter sufering of difraction because of the magnets in front of the diaphragm.

So I've design some flexible resin shape to bond on the tweeter surface (on both side).

This shape cancel the diffraction effect of the tweeter housing

It's 3D printing with formlabs flexible resin:
attachment.php


It's far more effective than I though, nearly perfect response (green curve after modification, red before modification)

attachment.php


I will give up the waveguide! :green:
 

Attachments

  • plan1.jpg
    plan1.jpg
    150.2 KB · Views: 3,695
  • plan2.jpg
    plan2.jpg
    299.4 KB · Views: 3,559
wow

I wonder if any further improvement if those bars were "bullet" shaped ?

I wouldnt be suprised if your mod with a small rounded wave guide would smooth the sharp roll off at the top end a bit. However what you got looks so effective you are likely at the point of diminished return
 
Last edited:
Excellent result!
Did you measure the off-axis response? Given this result it makes sense to do a set of measurements for a directivity plot. I would also be interested in the cumulative spectral decay. To round it off you could measure the harmonic distortion, just to be sure that there isn't a trade off somewhere.

regards,

Gerrit
 
Ive been doing a ton of testing recently on slot shapes for tru ribbons. I see very similar FR shaping results as altoluma is getting here and have only seen improvments on off axis as well with no big changes in waterfall and distortion.

Basically sharp changes in geometry make wiggles. Nothing new there but that respose altoluma is getting from simple half round bars is as mentioned better than expected. Gota love that.

I suspect if you flush mount that neo3 so that tops of half rounds are flush with baffle it would be even better.
 
Account Closed
Joined 2001
I've seen some (and I've made some myself) Neo3PDR measurements and they look much better than your red curve.

Here's one example from John Krutke a few years ago. Very flat through the 3-5khz range where your measurement shows a huge 13db variation.
Zaph|Audio

There's something else going on here. :)

Dave.
 
Account Closed
Joined 2001
I'm not talking about back cup issues. :) I'm well aware of those.

The back cup was in place for both of your modification before/after measurements above, yes???
I think as Gerritt mentioned, you need to make some off-axis measurements as well.

Dave.
 
Account Closed
Joined 2001
no back up, free air!

Okay, thanks for the clarification.

On my Neo3's, removing the back cup caused a severely under-damped response. I think because of your microphone distance you're probably not seeing it illustrated.

I'm not a fan of removing the back cup on those tweeters. They already have a significant resonant behavior and removing the back cup only accentuates the situation. Plus, distortion increases as well.
I consider the Neo3PDR a supertweeter only.

Nice experiment though.

Cheers,

Dave.
 
How about the baffle ? what are you using?
I could be wrong, but looking at the measurement in the first post, I'd wager they were near field measurements with the mike quite close to the driver.

A bit of deja vu here, as just last week I stumbled on the Audeze patent for what they call "fazor-elements" used in there planar headphones.
FAZOR ELEMENTS | Audeze
 

Attachments

  • fazor.png
    fazor.png
    490.5 KB · Views: 882
  • US9258638B2.pdf
    140.4 KB · Views: 131
I'm not a fan of removing the back cup on those tweeters. They already have a significant resonant behavior and removing the back cup only accentuates the situation. Plus, distortion increases as well.

Dave i'm very suprised about your distortion claim. I found the opposite and Troels too:

WITH BACK cup

Neo3PDR-closed_0.25mtfb_-22.3dB.jpg


WITHOUT
Neo3PDR-dipole_0.25mtfb_-22.3dB.jpg



futhermore , remove the back cup lower the resonance frequency so I found it usufull to use this tweeter from 2khz. but maybe we don't talk about the same thing?
 
Last edited: