block duration , integration time and settling time

[KT123]

From my knowledge of DFT, the sample interval must cover complete number of cycles in order to have a accurate DFT output.

For swept sine measurement, will the settings of block duration , integration time and settling time affect the DFT output?

Am I right that block duration actually determine the number of sample values to be used in measurement analysis? If i set my sampling freq = 51200Khz and block duration = 100ms, the number of sample will be 5120 samples.  How do i know that 5120 samples covers complete cycles of the input signal?

Thank you.

[Ymahardh]

Hi,

Yes correct, to get an accurate DFT measurement you have to use at least sample rate that can cover complete number of cycles. Commonly, you are using FFT technique to measure DFT. Larger block duration will give more frequency accuracy on frequency reponse. Meanwhile, when you performs single-tone measurement at each test frequency. It is recommended to use block duration which is small enough to give reasonable test time and large enough so it does not put the test computer at risk of being unable to continuously generate and read the signals. In summary, when you set smaller block size, the faster swept sine can transition from one test frequency to the next. To differentiate FFTand Swept sine technique, you can refer to this link: http://zone.ni.com/reference/en-XX/help/372416A-01/svtconcepts/swept_sine_vs_fft/

Yes, you can say block duration determine the numbe of samples in measurement. As I said, higher block duration will achieve higher frequency resolution. In order to get an accurate measurement and avoid aliasing, it is required to have at least a sample rate equals to 2 times of your frequency input signal and you can increase your block duration to get higher frequency resolution. You may refer to this link: http://zone.ni.com/reference/en-XX/help/372416A-01/svtconcepts/fft_funda/

Getting to know your input cycle or frequency depends on what you are measuring. Thus, I would be helpful if you can share what you are doing right now. 

Thanks

Yasrof 

NI Application Engineer

[KT123]

Thank you Ymahardh.

I am trying to test if the swept sine measurement VI provided in sound and vibration toolkit is accurate enough as compared to the actual analog measuring device.  Hence, i am now trying to understand the swept sine measurement VI.  However, my fundamental of signal analysis is not strong which had given me a hard time understanding it. May i know if there is any recommended book or online doc to understand more on signal analysis (specifically freq response, thd, icn and etc.)

Thank you.

[Ymahardh]

Hi,

I assume when you say accurate measurement, you are talking about frequency analysis. Both FFT and Swept sine may give you an accurate frequency analysis. It depends on what your objective is. This is about a short of fundamental regarding swept sine measurment: http://zone.ni.com/reference/en-XX/help/372416A-01/svtconcepts/swept_sine_overview/

You may refer to this link for more info regarding signal processing material: http://zone.ni.com/devzone/cda/tut/p/id/5564

This is one recommended book that is using labview: http://www.amazon.com/gp/product/0071444920

Last but not least, you can just search into NI website to look for particular term in signal processing. 

Regards

Yasrof 

NI Engineer

Please help me to give a kudo with a simple click 😄

[Ymahardh]

Oh ya, please let me know what hardware you are using and what measurement you need to achieve. I may be able to suggest ho you can get accurate measurement. Thanks.

Regards

Yasrof

[KT123]

Hardware used: NI cDAQ 9172, NI9215, NI 9263

Measurement to achieve: frequency response, THD and ICN

Frequency response:

Using analog measuring unit(function gen and oscilloscope),

For frequency of 1khz, the gain calculated is -4.7db

Using swept sine measurement VI,

For frequency of 1khz, the gain calculated is -6.11db

Thank you.

[Ymahardh]

Hi,

To get accurate data measurement without any aliasing, you should at least set your sampling rate at 2x of your frequency measurement. In this case of 1kH measurement, then your sample rate should be more or equal to 2kHZ. For your, NI 9215 is able to achieve maximum 100 KS/s sampling rate. 

You can find signal processing and analysis fundamental at NI website: 

 Signal Processing and Analysis

Use swept sine technique to see single tone frequency response and use FFT for braodband frequency reposnse. 

Please let me know if there is anymore that I can help. Thanks

Regards

Yasrof

NI 

[KT123]

Dear Ymahardh,

My setting used are: sampling freq - 51200hz

                                  block size - 100ms

the swept sine measurement freq response gain are still different from the calculated gain where readings are measured by the analog measuring unit. May i know what are the possibilties of the difference in the readings?

Thank you