virtins.com

Hi, I've got the RTA-168B speaker test kit with a calibrated mic. I've done sweeps out of my speaker and white noise and I've gotten a spectrum that shows it's frequency response. I've also gotten a .fcf file from that and I've applied it to my spectrum analyzer as the 2nd compensation after the RTA.fcf file (of the calibrated Dayton mic) that's the first compensation. Then when I do another sweep or white noise test I can see the much flatter response in the spectrum analyser from the same speaker. So while I use that calibrated mic I've somewhat normalised the frequency response of the speaker as the spectrum analyzer displays it.
But I now want to play white noise or a frequency sweep to a solid state sound recorder that records to wav files and has it's own mic whose frequency response I don't know. Is there a way to change the signal generator so that the sound level at each frequency is the same and it normalises out the frequency response of the speaker that way so I've got the same SPL at each frequency to test the SD card sound recorder. Can I normalise the sound level at each frequency that comes out of the speaker or postprocess the recorded WAV file to it's like the speaker had a flat frequency response and I can then see the mic response on the SD card recorder.

Thanks, Mark

Very good question!

It is possible to use the digital filtering function of the oscilloscope of MI to do it.

Here is an example. Suppose we have a speaker that has an inverted pink noise response (just for illustration purpose), that is, it has a slope of +3dB/octave. Now we need to filter the test signal such that, when the speaker is stimulated by the filtered test signal, the measured frequency response become flat.

1. Let's use the following trick to generate an inverted pink noise response:

1) From the Signal Generator panel (see figure below), generate a 10s white noise and save it to hard disk.


2) Use [File]>[Open] to open the saved white noise file
3) Set the X Scale of the Spectrum Analyzer to 1/3 octave
4) Set the Y Scale of the Spectrum Analyzer to dBr
5) Now we get a frequency response of an inverted pink noise as follows.


2. Right click the Spectrum Analyzer window and select "Spectrum Analyzer Generate Frequency Compensation File (flat)" to generate a Frequency Compensate File (FCF) which compensates the inverted pink noise response to a flat response.


3. Right click the Oscilloscope window and select [Oscilloscope Processing]>"Digital Filtering">"Arbitrary">"Frequency Response", and load the FCF file saved previously


Note: it is in the FCF subdirectory instead of the FRF subdirectory. Also, you will need to change the "Files of Type" in the File Open dialog from "Frequency Response File (*.FRF)" to "All Files (*.*) in order to see it. Then select "Persist" as you want to persist the change in the data. Now, you will see the frequency response displayed in the Spectrum Analyzer becomes flat (see screenshot below).


4. Use [File]>[Save As] to save the filter data into a WAV file.

5. To play the filtered test signal, choose "Wave File" in the Waveform selection box in the Signal Generator panel and select the above saved WAV file for output.


That is it.

By the way, the above procedure effectively converts a white noise signal to a pink noise signal. Similar methods can be used to generate other signals such as a band limited pink noise, etc..

Thanks very much. Sorry that I forgot to say thanks when you answered it.

It is really useful question. Thanks for the answer. I will definitely implement.

In a stepped sweep, one variable input parameter (frequency or amplitude) is incremented or decremented in discrete steps. After each change, the analyzer waits until a stable reading is detected before switching to the next step. The scaling of the steps is linear or logarithmic. The stability condition (settling) can be defined by the user.

Since the settling time of different test objects cannot be predicted, the duration of a stepped sweep cannot be determined exactly in advance.

For the determination of amplitude or frequency response, the stepped sweep has been largely replaced by the glide sweep. The main application for the stepped sweep is to measure the linearity of systems. Here, the frequency of the test signal is kept constant while the amplitude is varied. Typically the amplitude and distortion of the device under test are measured. This is also referred to as an “amplitude sweep”.