Ok, this is my first project and I am moving allong VERY slowly. I am confused by my tweeter responce. I think that the baffle diffraction looks very exagerated (ie +/- 3db) and it is making the tweeter crossover very difficult. I have the tweeter recessed, but the rebate ended up a little too shallow, so there is a little lip. Also I am not flush mounting the driver because I dont know how. I guess my question is, is this normal? Do I just have to design around this, or am I doing something wrong?
Front baffle
Close-up of tweeter rebate.
Tweeter responce in question.
So, do I just rebuild with tighter tolerances on my woodworking (again this is my first time at this) or is there a bigger problem here that I am missing?
Thanks in advance.
Chad
An externally hosted image should be here but it was not working when we last tested it.
Front baffle
An externally hosted image should be here but it was not working when we last tested it.
Close-up of tweeter rebate.
An externally hosted image should be here but it was not working when we last tested it.
Tweeter responce in question.
So, do I just rebuild with tighter tolerances on my woodworking (again this is my first time at this) or is there a bigger problem here that I am missing?
Thanks in advance.
Chad
Well you could try experimenting with felt strips around the tweeter like this site. And see if that smooths out the response. Personally I've had my own disaster at flush mounting since I only have the Dremel router. So from now on i just surface mount then put on a block of felt around the drivers.
http://www.speakerdesign.net/home.html
http://www.speakerdesign.net/home.html
I do not think it could be the mic, as when I do a nearfield, it looks a lot like the seas spec sheet... also the same pattern does not show up on the woofer responce. I guess that also confuses me, why doesn't it show up on the woofer... they are on the same baffle. Is it because of their diameters relative to the baffle size?
An externally hosted image should be here but it was not working when we last tested it.
I am stuck in a closet sized room and I can not get more than 1.5m from the side wall. I just thought that heavily windowing the impulse would take care of that (2.4ms). It looks like I have windowed out the first major reflection, but is it possible that it is still a boundry reinforcement issue? The mic is on a grip arm also about 1.5m from the nearest boundry (in its case a corner).
Tweeter Impulse responce
Perhapse I am just expecting too much from my measurement closet? I realize it is less than ideal, but it will really suck to have to move my big computer to another location, plus the only larger room is the living room... that might not go over very well with the wife! Is that the direction I should be moving though? Faulty measurements due to poor room?
An externally hosted image should be here but it was not working when we last tested it.
Tweeter Impulse responce
Perhapse I am just expecting too much from my measurement closet? I realize it is less than ideal, but it will really suck to have to move my big computer to another location, plus the only larger room is the living room... that might not go over very well with the wife! Is that the direction I should be moving though? Faulty measurements due to poor room?
A quick simulation in Edge with approximate measures from your pictures predicts this baffle step for your baffle design. It appears clear to me that the dip at 3 kHz and the ripple around that frequency originates in the baffle step. I don't see the dip at 8 kHz though in the simulation, and possibly it could be explained with that the flush mounting of the driver isn't perfect.
The peak at 25 kHz is IMO either a measurement error or a membrane resonance (break-up).
Edit: the fact that the peaks and dips disappear when you measure close-field further supports the idea that they are caused by the baffle diffraction. When you put the microphone closer to the driver than the edge, you increase the relative strength of the direct sound, compared to the sound re-radiated at the baffle edges, so the baffle diffraction then becomes supressed.
An externally hosted image should be here but it was not working when we last tested it.
The peak at 25 kHz is IMO either a measurement error or a membrane resonance (break-up).
Edit: the fact that the peaks and dips disappear when you measure close-field further supports the idea that they are caused by the baffle diffraction. When you put the microphone closer to the driver than the edge, you increase the relative strength of the direct sound, compared to the sound re-radiated at the baffle edges, so the baffle diffraction then becomes supressed.
Wow! That simulation looks a hell of a lot like my plot! Perhapse that just is my speaker, and measureing it differently or recessing the tweeter perfectly would only make minor differences? Maybe I should just move on with the crossover design (this whole process sounds a lot easier than it really is!).
By the way the tweeter is a SEAS 27TBFC/G (H1212) with the hex-cover removed, so the ultrasonic peak is surely the membrane breakup and was not unexpected at all.
Thanks for the help!
By the way the tweeter is a SEAS 27TBFC/G (H1212) with the hex-cover removed, so the ultrasonic peak is surely the membrane breakup and was not unexpected at all.
Thanks for the help!
How close was the mic to the speaker? I don't really se why, even with those dips it should cause a problem with your Xover design? Most speakers have a few peaks and dips of 3dB or so.
You might be finding it hard to do the actual Xover if the mic is too close to the speaker for the sound from each driver to integrate properly over the Xover range.
You might be finding it hard to do the actual Xover if the mic is too close to the speaker for the sound from each driver to integrate properly over the Xover range.
ch83575 said:
I think that if you want to get rid of the FR oscillations, you should experiment a bit with The Edge (in my signature) to get a flatter response and (sorry) rebuild the baffle of your speaker.
And, just in case you did not note it, my simulation did not include any information on your driver other than that the diameter is 25 mm, all the effects comes from that and the geometry of the baffle. Edge is not a loudspeaker simulator, it just models the baffle step.
I do not think that it is protruding bolts
on 8kHz length of a wave ~43mm
most likely it is protruding frame LF driver
http://www.zaphaudio.com/mtg-surface.html
on 8kHz length of a wave ~43mm
most likely it is protruding frame LF driver
http://www.zaphaudio.com/mtg-surface.html
I would first replace the grill of the tweeter as this might be designed to help dispersion at high frequencies (hence the odd HF peak) and then measure again. If the 8KHz and 3KHz dips are still there then play with off-centre positioning of the tweeter on the baffle. Don't worry about variations as small as the 8KHz one as it will be completely changed by room influences anyway. The 3KHz one is worth trying to fix though.
Ok, I re-measured with the cover on, and got this:
The cover makes the tweeter responce look much more like Zaph's measurements of this tweeter above 14k. BUT, in initial listening sessions with soundeasy's crossover simulator I think I like the sound better without the cover (even though it does seem to smooth out the 8k problem). I will experiment with the cover later... right now I am worried about the baffle and the major dip at 3k and peak at 2k (which it seems to be the consencus are diffraction related). I will download the Edge, and maybe try to figure out the diffraction simulator in soundeasy as well (although it doesn't seem as comprehensive as the edge). I originaly wanted to keep the crossover point low enough to avoid the woofers rather harsh-sounding breakup at 5k, but with the major dip at 3k in the tweeter I can not seem to get good phase and amplitude matching unless I cross at 2900 or 3k to make use of the natural roll-off from the 5k peak. This unfortunatly requires a 3rd order electrical filter on the woofer and a trap for the breakup node. Not exactly what I had in mind.
Any great ideas?
An externally hosted image should be here but it was not working when we last tested it.
The cover makes the tweeter responce look much more like Zaph's measurements of this tweeter above 14k. BUT, in initial listening sessions with soundeasy's crossover simulator I think I like the sound better without the cover (even though it does seem to smooth out the 8k problem). I will experiment with the cover later... right now I am worried about the baffle and the major dip at 3k and peak at 2k (which it seems to be the consencus are diffraction related). I will download the Edge, and maybe try to figure out the diffraction simulator in soundeasy as well (although it doesn't seem as comprehensive as the edge). I originaly wanted to keep the crossover point low enough to avoid the woofers rather harsh-sounding breakup at 5k, but with the major dip at 3k in the tweeter I can not seem to get good phase and amplitude matching unless I cross at 2900 or 3k to make use of the natural roll-off from the 5k peak. This unfortunatly requires a 3rd order electrical filter on the woofer and a trap for the breakup node. Not exactly what I had in mind.
Any great ideas?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.