tdms save fft cluster

[aartjan]

Hi all,

I've been using TDMS quite a bit lately and I am wondering who may have a nice solution for the foolowing issue:

A graph accepts the cluster output of the FFT VI's in the measurement analysis palette. The cluster is of type dbl (f0), dbl (df), array of dbl (waveform magnitudes).

Unfortunately TDMS write does not accept this dat type. A quick solution is to convert the cluster to a waveform, which I like. Now I just wonder, how to best 'represent' an FFT as an timedomain waveform. I later also read the data in Diadem, and it would be nice if I could make it recognize that the x scale is not time, but frequency bins instead.

Thanks for your suggestions

[Jarrod_S.]

DIAdem itself seems to store frequency channels as slightly altered waveforms, so you're on the right track by converting your frequency domain data into a waveform, even if it doesn't really make sense as such at first. When you calculate an FFT in DIAdem, the results are stored in a waveform-type channel with a few properties set specially.

For instance, check out the properties for this FFT result:

The "wf.." properties all denote waveform channels. Simply wiring a waveform into a TDMS Write function will automatically create the first four wf properties in this list, so if you want DIAdem to treat your data more like frequency data, try manually adding those last few properties with appropriate values. You can use TDMS Set Properties to do so. The property name should be exactly like you see in the picture.

Message Edited by Jarrod S. on 02-15-2007 10:37 AM

[aartjan]

Yes, I came to the same conclusion and made the attached vi. It gives me the option to write the FFT as a waveform or as two writes (data and properties).

What bugs me, though, is that I have to reverse engineer all these properties myself. I could use a whitepaper on all the NI predefined properties of datatypes such as waveform.

Aart-Jan