do my frequency response measurements make any sense?

[JLC7]

Hi guys, this is my first time taking real driver frequency response measurements. Could you guys tell me whether or not they make sense and if they're usable for crossover design? Thanks a bunch.

My setup looks like this:



ECM8000 mic --> xenyx 802 preamp --> line-in soundcard
+
soundcard output --> preamp --> power amp --> driver

I'm using Speaker Workshop for measuring the frequency response of the drivers (sample rate 44100, sample size 16384)

This is the pulse measurement I made to set the gating times. Did I mark off the pulse from the reflections right? I put the marker before the most significant reflection I could see.


After marking off the pulse and setting the gate times, I made a gated measurement of the tweeter from the position of the mic in the picture.



and the woofer after that:



What I don't understand is why the tweeter rolls off so quickly before 20 kHz. The mic is supposed to be flat until 20 kHz and the tweeter (vifa XT25TG30) should be flat until 30-40 kHz.

I thought I might not have aligned the mic exactly on axis with the tweeter so I made a near field measurement of both drivers and port.



I didn't adjust the levels yet, but the tweeter measurement still rolls off way before 20 kHz even 1cm in front of it. Is this normal? I checked both tweeters I have and they both do the same thing.

The last measurement I made was a far-field non-gated of both drivers.



Could guys let me know if these results are typical and any improvements that could be made?

[Conrad Hoffman]

I don't know much about the specific hardware, but can you connect the sound card line out to line in, and see what the response is using the exact same method? Divide and conquer- see where you're loosing the response.

[augerpro]

Good suggestion by Conrad. When you do this loopback measurement you should have amplitude and phase dead flat. If not there may be a difference in latency between the soundcards channels, which you should be able to adjust for in SW.

Your guess of the marker placement looks reasonable, but you really should measure the actual distances to the mic and first reflection points to calculate exactly where these markers should be.

For your mic distance (1m?) and the gate length you get too many phase wraps. Strange. I'm not familiar with SW though, so I'm not sure what would cause this under the circumstances. Typically you see this when you include the time of flight in the gate, but it doesn't look like you did that.

The ECM8000 is not flat to 20k. Have you had your's modified or do you have a calibration file for it? I would say that was the problem but most ECM8000s I've seen show a rise in the top octave, not a drop.

[JLC7]

You guys are great.

I happen to have two sound cards in my computer. One being the built in on-board sound and a dedicated Audigy 2 ZS (same one Zaph uses minus the front I/O panel).

I used the onboard for the driver measurements because I have feedback problems with the audigy because it doesn't record from line-in unless you un-mute the volume for line-in playback. There's an option to turn off monitoring while recording but it only goes into effect when something is actively recording from line-in. The rest of the time it plays back whatever comes in and I get feedback problems that threaten to blow my tweeter.

Here's the onboard sound card hooked back into itself.


Pulse measurement


Un-gated frequency response

Here are the same measurements with the Audigy 2 looped back into itself.


Pulse measurement.


Ungated frequency response.

It looks to me like they both start rolling off before 20k and supposedly the Audigy 2 is a high end sound card.

Here's a thread that mentions that the ECM8000 is flattest at 70-80degrees off axis. But like augerpro said the measurements seems to roll off alot when the mic itself shouldn't.

[augerpro]

Yeah that looks like garbage. It may not be the soundcards though. Like I said check for a way to change interchannel latency on the soundcard. There are some good tips on making stable computer for measurements on the Praxis site. Turning off non essential services and network controllers, etc. I'm sure it's just something to do with your setup.

One thing that stands out is the impulse is at 20+ms, which seems too far for your mic distance. Are you sure you calibrated SW correctly?

[Salas]

I would simply check with a second program, Arta demo for instance, and see if its the card or the software giving the most trouble. So to start tweaking rightly.

[JLC7]

There's not much to calibrate in SW. I did the channel difference which is just a loop back. Didn't help. I'm not measuring impedance with SW so that calibration wouldn't make a difference either.

I already disabled the Creative sound application startups before in MSCONFIG. Not sure if crippling my installation of windows is required to get this working right though.

I'll try arta. Any other freeware programs out there I can try that will get the job done?

[Salas]

Use SynRTA from Liberty to check with RTA too. Its free.

[JLC7]

I'll try that too.

Here's an interesting site that does a comparison between the frequency response of different consumer and professional sound cards. Suffice to say, they don't look like what I'm measuring.

http://jimmyauw.com/2006/08/09/sound-card-comparison/

[Salas]

Here is my loop correction for Audigy 2 in my stationary PC. It looks consistent with your link too.