Wierd tweeter response?!

[sobazz]

I'm having some troubles with the measured response of the Peerless 811435 neodymium tweeter. The problem may very well be related to the microphone used - I donno - which is a WM60AY combined with Rod Elliot's mic preamp (not line level). The microphone is connected to the microphone input of my Audigy. This is problematic because of the bias voltage for a standard direct connection to the capsule. In the case of my Audigy the mic input seems to be configured quite differently from others. I have measured ground-3.2v-1.8 instead of the standard ground-3v/5v-signal.

Anyway here's the response graphs. I've put a 21uF pp cap in series with the tweeter. This is the gated response:


And the ungated farfield:



What is the problem here? And why is there such a big difference between the gated and the ungated response (especially below 2KHz)? Sorry for the difference in the scales.

[Svante]

You don't say what you think the problem is, but I suppose you mean the +6dB step above ~4kHz? In that case, congratulations!. You have just discovered the "baffle step" and in this case the baffle is small (I suppose the tweeter was unmounted?). Put it in a baffle and note the difference. Then figure out how to compensate your system for this effect!

PS Peerless measure all their speakers mounted in a wall, so they don't get any baffle step in their response curves.

HTH

[sobazz]

Thank you for your reply, Svante. I know about baffle diffraction step, but I have not seen such an extreme example before.

The tweeter is mounted in a baffle: approx. 26mm from the top, 31mm from the left and 61mm from the right baffle edge. The baffle width is 145mm.

The bass/midrange is mounted 9mm from each side edge. Calculating the f3 point of the BS using f3=4560/Wb where Wb is the width of the baffle, I get approx. 800hz.... but then again - the tweeter is not flush mounted... hmmm...

I using the width of the tweeter frame which is 53mm you get 2185hz.... are you sure that this is due to baffle diffraction step?

[AGGEMAM]

Remember it's half the width, so 4270 Hz. which corresponds nicely to the measurements you've got.

[Svante]

Originally posted by sobazz
Thank you for your reply, Svante. I know about baffle diffraction step, but I have not seen such an extreme example before.

The tweeter is mounted in a baffle: approx. 26mm from the top, 31mm from the left and 61mm from the right baffle edge. The baffle width is 145mm.

The bass/midrange is mounted 9mm from each side edge. Calculating the f3 point of the BS using f3=4560/Wb where Wb is the width of the baffle, I get approx. 800hz.... but then again - the tweeter is not flush mounted... hmmm...

I using the width of the tweeter frame which is 53mm you get 2185hz.... are you sure that this is due to baffle diffraction step?

No, I'm not sure. It just fitted so (well reasonably) well with an unmounted tweeter. Anyway, there is a *big* difference mounting a driver in the centre of a baffle and at the edge. You can try different baffle placements with this program:
http://www.tolvan.com/diffract.exe
I'm starting to feel like a marketer now (I have posted this link a few times the recent days in this forum, stop me guys if I am to persistent).

One way to test if it is the baffle step you see, is to take a piece of cardboard and tape it to the front of the baffle, with the tweeter sticking out of it in the middle. You should then see a change in your curves.

Was any of the measurements taken in the close field? I suppose the woofre (wow, brittish mistyping??? ) was disconnected?

[sobazz]

AGGAMAM: Is it half the width of the baffle? Never read that before.

Svante: The woofer was disconnected and the measurements were both done in the farfield at approx. 60cm distance. I will try the cardboard trick.

Does it matter if the driver is flush mounted?

[AGGEMAM]

Originally posted by sobazz
AGGEMAM: Is it half the width of the baffle? Never read that before.

Well, I could be wrong but I always thought it was the distance from the center to edge that determined the frequency where the baffle amplification started to take effect.

That's what I always figured when designing a speaker and it works beautifully.

And square-baffled speakers like the peerless one is a particular problem when not baffled properly.

[sobazz]

Well, it sure sounds reasonable.

I start to wonder why a placed the tweeter off centre....

[Svante]

Originally posted by sobazz

Does it matter if the driver is flush mounted?

I wouldn't think so. It goes out just a few millimetres, right? Then there should only be a minimal reflection there, that certainly could not make a 6dB+ difference at 10 kHz where the wavelength is 3.4 cm.

I ran it through my diffract.exe, and I see a peak at 2-3 kHz and one at 9 kHz. In fact I think see those in your measurement as well, but with a superimposed tilt of maybe +6dB/octave. Hmmm... You mentioned that you have a 21uF cap in series with the speaker, are you sure that it is OK? There seems to be a HP function way up high. Possibly you should check the equipment once more?

[sobazz]

Can I use a notch filter for correction? If yes, component values would be nice

Thank you very much - all of you - for the help so far.