What's going on here?

[wintermute]

Hi,

I haven't got a proper mic yet but I couldn't resist playing around, but when I did I got a nasty surprise (to the tune of a 15db dip at about 1.2Khz!!! as well as a very nasty sawtooth shaped freq response curve).

I had initially measured some freq response curves at about 20-30cm away from the tweeter and midrange (all drivers running) and got a somewhat rough freq response. Yesterday I decided to measure at 1M. The result was awfull! I've attached an image with measurement at .5M 1M and 1M with tweeter polarity reversed. The crossover freq is at 3K but the mid I think starts rolling off at about 2.7K. (the crossover was purchased pre built before I had any knowledge about suitability of drivers etc etc etc).

Is what I am seeing some sort of cancelation between the mid and tweeter, or perhaps a standing wave which has it's null point at 1M? The fact that the dip moves from just over 1K to 1.2K when the mic moves back makes me think it's interference, obviously reversing the tweeter polarity makes it worse, to the tune of about 20db down, is it possible that the phase in the xovers is really screwed up (like 90 deg or 270 deg different? I figure I should try and find the reason for these sorts of things before I start on my next project, so I can avoid them if at all possible.

The curves are unsmoothed, and I tried two different mics, which were both reasonably consistent up to about 3K but wildly different above that.

Regards,

Tony.

[Volenti]

hmm, disconnect the x-over off the mid and run it full range and test it again, both with and without the tweeter going. If it's still there with just the mid by it's self with no x-over then you know that the x-over at least isn't the problem.

After that take the mid out and mount it in a flat baffle, open baffle style, if the dip is still there there's not much you can do about it.

[wintermute]

Thanks Volenti, I was thinking about testing the divers individually but that's a better Idea

At this point I'm wishing I had used connectors on the drivers rather than soldering onto the terminals!

The mid actually has it's own sealed back (its a dome one) so the OB isn't possible......

Regards,

Tony.

[Volenti]

Originally posted by wintermute

The mid actually has it's own sealed back (its a dome one) so the OB isn't possible......

Regards,

Tony.

oh.., well then what is the x-over point between the mid and the woofer? the problem may lie there, but the full range test will be a good thing to do anyway.

[wintermute]

woofer xover is at 500Hz. The mid is actually running way past its range at both ends so could be part of the problem.

I think I'll desolder off the xover bit less risky ..... About to pack up the computer and cart it downstairs to do some measuring. It's easier to drag it down, than unplug everything from the amp and bring it and a speaker up to the computer!

Regards,

Tony.

[wintermute]

Well It's either the mid or it's interaction with it's surroundings.

The first test I did was with the mid only at 1M and the big dip was present. Interestingly the graph in the attached image was the second test I did after moving the mic slightly which changed the curve from having a sharp valley to having a flat bottomed one.

the first graph is with the mic 1cm from the dome, I had to turn the rec level and volume down as it was (I think) maxing out the mic.

I then took at 10cm 20cm 50cm (I already had the 1M one). I haven't included the 10cm one because of space, but it was flatter than the 20cm and more peaky than the 1cm.

The 1cm graph is a bit deceptive since its been flattened somewhat to fit on the screen the difference is around 3db, the last one (at 1M) has a difference of around 15db. I adjusted the mic and volume levels again at 50cm as there was quite a bit of ambient noise.

So I guess it's conclusive in that its not cancelation between the tweeter and mid, and not the crossover. As to whether it's the mid itself or the baffle, or room/furniture I don't know. I think I'll wait till I have a proper setup and go to the warehouse at work and test in a really large space before doing any more measurements, but any opinions on possible causes would be appreciated

edit: oh and I changed the range to only show 300 to 10Khz. All measurements in speaker workshop using the on-axis measure menu option.

Tony.

[claudio]

Hello Tony,
how many ms. of useful data did you have in the gating?
I don't understand why you did an on-axis measure (that is gated) at 1 cm.: shouldn't it be a near field (that is not gated) at that distance?

The distance mic-baffle, in the far field on axis measure, that varies from 0.5 to 1 meter depending on the driver; then you compare the curves taken at, for example, 60-80-90-100 cm., to see which ones have the same shape, indicating that you are in the far field .

Regards

Claudio

[qwad]

Hi, couldn't help but interject after having read the whole sad saga; a question= what dome mid are you using? also please take not that any dome driver is in break up mode over the majority of its freq response so those last screen shots explain the sawtooth phenomena, also have you done an imp plot over the same range this is handy to verify how the mid is behaving, break-up etc, etc, you can tell if it is break-up by little wiggles in the imp curve which apart from the hump at fs should be smooth...... also mic quality comes into this should be caliberated, the best and cheapest I know of is the mitey mike from old colony if they still have it as ordinary mikes off the shelf are fairly rough in their response [ just my oz 2 cents worth] cheers and best regards Tomcat

[MarkMcK]

Tony,

Sorry, but I did not come online until you were already off. I hope you have already figured out your problem.

It is hard to troubleshoot just looking at your response graphs, but I think you may be looking at decay responses in the graphs with the microphone away from the speaker.

Do a pulse response as I described in the "software" thread the other night. Look at your time response to see if onset response is within the window (gating) and if there are any boundary reflections within the window.

I have seen some poor performing drivers before, but this is worse than I would expect. Almost has to be something in the test set-up.

Good luck and keep us posted with the solution.

Mark

[janneman]

Tony,

I must say that the very first curve you posted (the 0.5m distance) is really a quite good speaker curve! If you are not happy with it, I think you will have a VERY hard time to improve on that! Don't expect curves like an amp from a speaker. Again, with all the difficult variables in measuring and acoustic interferences from the environment, I am sure there are not many that have so good a curve in a practicle measurement.

Jan Didden