Measurement of frequency characteristics using periodic sweep

Shows how to measure frequency characteristics by combining WaveGene (hereinafter WG) and sister software WaveSpectra (hereinafter WS).

There are the following methods to measure the frequency characteristics with WG + WS.
(1) A method of using periodic sweep in the user waveform of WG.
(2) A method using the sweep signal of the WG directly. (Method using the peak hold function of WS)

Normally, accurate frequency characteristics can be easily obtained in a short time (at most several seconds), so method (1) shown below is recommended.

This method can be used in most cases, such as the characteristics of the sound device itself when connected to a loop, an external circuit such as an amplifier connected to the sound device, or a simple measurement of a speaker via an amplifier / microphone. (When a direct response can be obtained in real time) If

(1) cannot be used, for example, if the response does not return in real time, such as the recording / playback characteristics of a recorder, the general sweep signal + WS peak in (2) The method using the hold function will be used.

For notes on (2), refer to the explanation "Measurement of frequency characteristics using WaveSpectra" on the WS side .

1. Occurrence of periodic sweep by WG

• Please download the periodic sweep or periodic noise signal to be read as the WG user waveform separately.
  After downloading UserWaveSample1.ZIP from The WaveGene page, unzip it and put it in the same folder as WG.EXE.
  
  The number in the file name of each signal indicates the length of one cycle (number of samples). (2048 --131072)
  
  
• Select the user waveform in the Wave1 waveform combo box and register the appropriate periodic sweep or periodic noise signal.
  For how to register, refer to How to use user waveform
  
  * Both periodic sweep and periodic noise can be used in the same way, but periodic sweep has a better S / N and periodic noise has a very slight (+ -0.01 dB) disturbance in flatness, so it is usually periodic. It's best to use a sweep.
  
  As an example, create a periodic sweep FLATSWEEP_004096.WAV for 4096 samples.
  
  
  
• Set the format of the playback device and sampling frequency used for measurement as appropriate.
  
  * The time of one cycle of the periodic sweep signal is the number of samples / sampling frequency.
  
  In the case of the example, if the sampling frequency is 48000Hz, it will be 0.085 seconds because it is 4096 samples.
  
  
• Set the output level in the amplitude combo box. (Approximately 0 to -3 dB, see the note below)
  
  
• Make sure that the output button to the sound device can generate it continuously.
  (The ability to create a Wave file for a specified time with the Output to Wave file button is the same as before.)
  

2. WS side settings

• Set as follows in the setting dialog of WS (V1.40).
  
  On the "Play / Record" tab, match the recording device and recording format with the WG side.
  
  Set the recording format to 48000Hz 16bit Stereo, which is the same as the WG side.
  
  * The recording device and driver settings are MME and Wave Mapper here, but specify the same device and driver that were actually set on the WG side.
  
  b. On the "FFT" tab, match the number of sample data with the length of the periodic sweep (number of samples) registered in the WG.
  
  For the example, set it to 4096.
  
  c. Similarly, make sure that the window function is "none (rectangle)".
  
  

3. Measurement of frequency characteristics

• After making the above settings, operate both WG and WS. (The order is not specified.)
  (Press the "Output to Sound Device" button and the "Input / Record from Sound Device" button, respectively.)
  
  A graph of frequency characteristics is drawn for each cycle of the cycle sweep.
  
  If the number of samples in one cycle is long, the graph will not be accurate until the end of one cycle, so please wait for a while.
  
  
  * First, connect the input / output of the sound device (loopback) and measure the frequency characteristics of the sound device itself.
  Input / output connections can be made with a cord or specified by the mixer of each device.
  After confirming that it operates stably, it is recommended to measure via the device under test (amplifier, speaker, etc.).
  
  
  
  
• You can make the graph easier to see by changing the vertical axis "range" and "shift" on the "Spectrum" tab of WS.
  
  If you set "Range" to + -1dB etc. and "Shift" to Norm1k etc., you can draw a graph based on the general 0dB.
  
  
  
  
  The "shift" Norm100, Norm1k, and Norm10k are options that always fix the 100Hz, 1kHz, and 10kHz spectra to 0dB, respectively.
  It is useful for graphs of frequency characteristics such as filters and equalizers.
  

4. Precautions, etc.

• As the sound device used in both WG and WS, be sure to use the same device that can record and play back at the same time.
  You cannot use different devices on both sides.
  
  Also, the recording / playback clocks must be synchronized.
  (It seems that some sound devices cannot synchronize the clock during recording / playback, but such devices cannot be used.)
  
  Also, the frequency response display is not possible even if an internal sampling rate converter is included. It seems that it may not be usable due to stable shaking.
  Even though the specification states that it is supported, it was actually unstable and useless at high sampling rates of 96kHz and above.
  
  Both can be judged to be usable if the display graph of the frequency characteristics of WS is displayed stably.
  
  
• The input / output level should be as close to the maximum value (0 dB) as possible, but depending on the sound device, input / output at 0 dB may not be possible (distorted), so in that case, try lowering it by a few dB. please give me.
  (First, check the loopback as above)
  
  
• The length of the periodic sweep is the FFT length as it is, so the longer it is, the higher the frequency resolution.
  (Frequency resolution = sampling frequency / number of FFT sample data)
  However, select so that the frequency characteristics of the measurement target do not change within that length of time.
  
  Also, in most cases, it is not so relevant for normal measurement, but if the impulse response time of the measurement target is long, set it to a longer time.
  (For example, if the impulse response time is 1 second, 65536 samples are required for 48000s / s)
  
  
• Since it is necessary to operate both WG and WS stably, it is recommended to reduce the load on the CPU as much as possible.
  
  Set the "Drawing method" on the WS settings dialog, "Wave" tab, and "Spectrum" tab appropriately to reduce the CPU load.
  See WS Help for more information.
  
  
• If the WG and WS cannot operate at the same time for some reason, use the WG's Wave file creation function to output the periodic sweep to a Wave file of the length required for measurement, and then stand the second WS. Please raise it and play it.
  (At this time, if the second WS is set in the same way as the first WS, the characteristics of the periodic sweep itself can be observed.)
  
  
• The frequency response can be obtained by directly applying the same periodic pulse, but the S / N is lower. If you change from "User Waveform" to "Pulse" in the Waveform
  
  Combo Box of WG during the measurement by periodic sweep, you can check the improvement of S / N compared to the case where the impulse is directly input. However, the period of the "pulse" must be the same as the period sweep in the frequency / period combo box in advance.