Cut Out Module

[maffos]

Hello All...

I am using DasyLab 13 Pro. In my worksheet, I have the output of a Transfer Function module (Add-ons) going into a Y-t chart to display transfer function magnitude / phase vs. frequency. The transfer function module gives a double-sided frequency spectrum which I'd like to truncate. Let's say the block size is 32768 and I am trying to keep the first 12800 points. I tried this at first with a "cut out" module but when I do this the block size of the output blocks is still 32768 so I get overlapping blocks. I could use the "replace with zeros" option in the "cut out" module but ideally I'd like to have output blocks that have a block size of 12800 and have the correct frequency abscissa. Is there an easy way to do this?

Thanks!

[CJ_Butler__DASYLab_Pro]

In DASYLab 13.0 and DASYLab 2016, the Cut Out works as you expect. 

My input block is 8192. I configured it for 0 to 5000, with NO replace with zeros, and the output block size is 5001. 

You may need to select the setting to keep the original time stamp.

[maffos]

CJ... That is same as the setup that I have but the output does not work out correctly. . In this example if I set the block size to 8192 and the Cut Out module to 5000, I get an output with a block size of 8192 and a "Real Block Size" of 5001 (as seen in the Info Area). (I don't know what the difference between Block Size and Real Block Size is...) When I try to graph this or save it to a file, the frequency vector is 8192 points in length w/ the original frequency range and the data is 5001 points in length. What I did to look at this is to run a Generator outputting 4V of noise into a Digital Filter set to ~ 1/2 my sample rate. I then ran that into an FFT module and connected that to a Y-t chart. I also ran this signal into the Cut Out module and then into another Y-t chart. The Y-t chart on the Cut Out module shows the garbled spectra. I tried this with the setting to keep the original time stamp and to not keep the original time stamp and it does not work in either case. Thanks for your help! Matt

[CJ_Butler__DASYLab_Pro]

Can you attach that worksheet? 

[maffos]

CJ - See attached.

Thanks!

[CJ_Butler__DASYLab_Pro]

Ok... I see the behavior. 

[maffos]

What I have *been* doing is exporting the whole thing and manually deleting the unwanted rows. The replace by zero option is OK but I'd rather get rid of the rows entirely! I have tried a couple of different approaches but they all end up messing up the time (frequency) data...

[CJ_Butler__DASYLab_Pro]

It turns out, that may be the best thing. 

The sampling rate defines the range for the x-axis (e.g. 2000 Hz) and the block size acts as resolution ("points-per-millimeter") for the signal that always "fills“ the entire space of the x-axis. We are in frequency domain, not in time domain.

When you cut the block, it does not alter the sampling rate, so the x-axis of the Y/t chart is not modified.

When you reduce the number of samples, the range of the x-axis needs to be filled with samples, so the Y/t chart collects samples to show a record of data that fills the range.

When you reduce a block, the short block flag is set to indicate that it is not the original block size. The Relay module does this too. The short block size is not guaranteed to be equal for each block. For example, one block could be 1000 sample wide, the next 20, the following 4000, etc. As a result, the range of the x-axis of Y/t chart would be jumping between different ranges…

Visualisation in Y/t-chart:

You need to use the Zoom function to change the x-axis range to something smaller that the maximum range defined by the sampling rate.

Saving to file:

Use the „Data window“ to reduce the amount of samples written to the file.

Do not show the output of the Data window in a Y/t chart: the x-axis has the range defined by the sampling rate, the data-windowed samples still fill the entire range – but that is… not correct.

This may be a solid case for post-processing with a different program. 

[maffos]

CJ - thanks for your help. I'm glad to know what is really going on here!