Poor man's frequency response analysis

  • Thread starter Thread starter crazydoc
  • Start date Start date
darrin_h2000 said:
Most surround sound recievers have Pink noise output for testing the levels of the speakers. If you can put that on a cd.
Click to expand...

So that is what that is for!
-Kirstin
 
There is indeed a new very useful tool, and it's freeware, too: The Rightmark Audio Analyzer

http://audio.rightmark.org/

The main problems, however, remain: the room (reflections, resonances) and the speakers (uneven frequency response, distortion, etc). Still, the results can be useful if used with care.

Also you can use the analyzer to measure your soundcard, line equipment etc. It's pretty accurate for stuff like that, provided you have a decent 24bit card capable of 96k or more.
 
Thank you Gentlemen for bringing back this thread.

I was using DC Five, but since then I moved and reinstalled my computer system, and don't seem to find the disk. Thank you for leads for software.

<In fact, pink noise might be even worse with its high freq rolloff, since with the single cone speaker I'm using I already have to compensate so much at high freqs.>

Crazydoc,

Pink noise is a standart for measuring and doesn't have HF roll off--it is flat. Rather, white noise has 3 db raise with each octave up, so pink noise uses opposite filter to compensate for it. With high levels white noise can even damage your tweeters.
On ebay they sell a lot of CDs with recorded noise for 10 bucks. Another option would be to get Velleman Pink Noise Generator kit. It goes at ~$15.

For mic measurements put the speaker in the middle of the room to avoid early reflections. You can use booth made out of tent--you know, they sell those in Sport Authority--made out of aluminum tubes, and hang around it woolen blankets for good dampening.
For measuring cardiodid mics the distance between mic and speaker should be at least 2' in order to avoid proximity effect. Also, this distance will help if you use two or three way speakers as a sound source.
 
Originally posted by Marik


<In fact, pink noise might be even worse with its high freq rolloff, since with the single cone speaker I'm using I already have to compensate so much at high freqs.>

Crazydoc,

Pink noise is a standart for measuring and doesn't have HF roll off--it is flat. Rather, white noise has 3 db raise with each octave up, so pink noise uses opposite filter to compensate for it. With high levels white noise can even damage your tweeters.

Marik - I think what I meant was the relative rolloff compared to white
noise (you'll notice the post is going on two years old), treating white noise as the reference. The speaker I was using was so high frequency deficient that I needed all the boost I could get in the highs.

For mic measurements put the speaker in the middle of the room to avoid early reflections. You can use booth made out of tent--you know, they sell those in Sport Authority--made out of aluminum tubes, and hang around it woolen blankets for good dampening.
For measuring cardiodid mics the distance between mic and speaker should be at least 2' in order to avoid proximity effect. Also, this distance will help if you use two or three way speakers as a sound source.

Thanks for the suggestions - if I ever get back to messing with this. I hadn't thought of the proximity factor when I was doing this.

Thanks to c7sus and Rossi for your links - I've downloaded them but haven't installed yet.
 
Crazydoc,

Sorry, I was thinking not of 2', but 2 meters, which is about 6'. Sorry for confusion.
BTW, I am driving to Tempe in a couple weeks to perform at Katzin at ASU and to make some recordings. How about some beer and shooting some pool?
 
Marik said:
BTW, I am driving to Tempe in a couple weeks to perform at Katzin at ASU and to make some recordings. How about some beer and shooting some pool?
Click to expand...
Marik-

:mad: :mad: Crap, I missed this post, and just saw it now when this thread was referenced in another one. I'd have liked to have met you.

Let me know if you're going to be in town again. :)
 
<Let me know if you're going to be in town again.>

I drive down there often. Next time in May. PM me with your phone# and I will give you a call when I am there.
 
I guess that it is quite important for comparative testing that the measuring criteria and situation is consistent, so it might be good to use a recording of noise each time, rather than use the raw output of a generator? Also obviously the relative positioning of the source, the microphone that is subject to the testing, the stand etc all be the same each time? Another thing to consider of course would be other environmental influences such as boundaries etc that can cause reflections?

I find AB testing like this to be very productive, as it immediately gives you a direct means of seeing the immediate results of modifications.

Great stuff!
 
audioboffin said:
I guess that it is quite important for comparative testing that the measuring criteria and situation is consistent, so it might be good to use a recording of noise each time, rather than use the raw output of a generator?
Click to expand...

Not really, it depends on sample length. Anything longer than a few seconds should be sufficiently random.

Also obviously the relative positioning of the source, the microphone that is subject to the testing, the stand etc all be the same each time?
Click to expand...

Yes.

Another thing to consider of course would be other environmental influences such as boundaries etc that can cause reflections?
Click to expand...

Yes, although you can alleviate that by decreasing distance to source. That will change the low frequency response of a directional microphone however, due to proximity effect. But generally when people are concerned with a microphone's response, they are more interested in presence range behavior as that is where directional microphones tend to vary the most.

There are also more sophisticated techniques to control for room response when measuring low frequencies. Or there is the simple method, which is to go outside.
 
I've just tested the HF response of my mic AT4022. I don't have a spectrum analyser so I used a signal generator and oscilliscope, and wound the frequency slowly up noting the peaks and troughs, which are often very close together. These peaks and troughs are wildly different in level, sometimes 20 times = 26dB. What causes these peaks and troughs? The mic was sufficiently close to the tweeter of the speaker that it couldn't be room echoes, and slightly varying the distance completely changes them so where there was a peak there is now a trough. The figures obtained are so far apart that a graph is meaningless, so I added each adjacent peak and trough and divided by 2 to give a usable figure, and I did three sets of measurements at 6" 12" and 18" and summed the results. The final graph produced looks very similar to the response curve that came with the mic except the peak has moved from 8.7kHz to 8kHz. My question is what causes these peaks and troughs? It cant be the mic or the tweeter because they change as you alter the distance between the mic and tweeter. Also it cant be interference between the tweeter and the midrange speaker because it is the same using a tweeter on its own.
 
timtimtim said:
I've just tested the HF response of my mic AT4022. I don't have a spectrum analyser so I used a signal generator and oscilliscope, and wound the frequency slowly up noting the peaks and troughs, which are often very close together. These peaks and troughs are wildly different in level, sometimes 20 times = 26dB. What causes these peaks and troughs? The mic was sufficiently close to the tweeter of the speaker that it couldn't be room echoes, and slightly varying the distance completely changes them so where there was a peak there is now a trough. The figures obtained are so far apart that a graph is meaningless, so I added each adjacent peak and trough and divided by 2 to give a usable figure, and I did three sets of measurements at 6" 12" and 18" and summed the results. The final graph produced looks very similar to the response curve that came with the mic except the peak has moved from 8.7kHz to 8kHz. My question is what causes these peaks and troughs? It cant be the mic or the tweeter because they change as you alter the distance between the mic and tweeter. Also it cant be interference between the tweeter and the midrange speaker because it is the same using a tweeter on its own.
Click to expand...

That's a bit challenging to answer exactly; variations are normal as a result of tweeter and room response, but the magnitude is larger than I have typically seen for high frequencies (it's not unusual for bass frequencies though). The usual method is subtraction, where you compare to a mic of known response, usually a flat-response small-diaphragm omni. That will mostly eliminate variation in high frequency response *if* you are careful to colocate the capsules.
 
mshilarious said:
That's a bit challenging to answer exactly; variations are normal as a result of tweeter and room response, but the magnitude is larger than I have typically seen for high frequencies (it's not unusual for bass frequencies though). The usual method is subtraction, where you compare to a mic of known response, usually a flat-response small-diaphragm omni. That will mostly eliminate variation in high frequency response *if* you are careful to colocate the capsules.
Click to expand...

Yep, without reference measuring mic it is impossible to get any meaningful results.

Those huge HF peaks and dips can be result of speaker baffle diffractions, esp. on the frequencies when the size of the baffle gets close to the wavelength.

Best, M
 
I havn't got a measurement mic of course.
I was only measuring response from 3kHz to 17kHz so it was in the range of the tweeter only. The "troughs" are so low that it would seem they are caused by cancellation of some sort, rather than just bad response of the mic or tweeter, and anyway they move in frequency as you change the distance. What is surprising though is that by adding adjacent peaks and troughs, I came up with a very non-spikey graph from 4kHz to 17kHz which very closely resembled the graph which came with the mic. The reason I'm doing all this is because I want to design a little filter to remove the 4dB peak at 8.7kHz. Obsessive I know - I doubt if I'll even hear the difference, but it would be nice to know I've got a perfect mic!
 
You must log in or register to reply here.
Back
Top