TDM Data organization

[YannickW]

Hello,

I have some question regarding the organization of the data in a TDM file. I use CVI 8  to generate the TDM file.

I am a complete beginner with Diadem and I am not sure if I save my data in the more efficient way.

Lets take an example with a bandwidth test of a digitizer.

The bandwidth is measured for each channel at different input range and different input impedance.  I want to save the frequency response and the measured bandwidth for each condition.
 I have organized the data like this

Root
- Custom property : UUT model
- Custom property : UUT Serial number

Channel group “Channel 1”

 - Custom property : UUT Temperature

- Data Array “Input Frequency”

- Data Array “50 Ohms - 100.0 mV”

- Custom Property : “Input range”
- Custom Property : “Input impedance”
- Custom Property : “Measured BW”

- Data Array “50 Ohms - 200.0 mV”

- Custom Property : “Input range”
- Custom Property : “Input impedance”
- Custom Property : “Measured BW”

 Channel group “Channel 2”
…
…

I would like to make some analyze like:
- To plot the body plot of a specified UUT
- To make a statistic of the “Measured BW” for a specified “UUT model”, “Input range” and “Input impedance”
 
Does this way of arranging my TDM file will let me do this kind of analyze of is there a best way to do it?

Best regards,
Yannick W.

[Brad_Turpin]

Hi YannickW,

Well, it is a pleasure to see such a detailed question, and you are wise to ask if the mapping of your data to the TDM model is optimal.  In general I like your approach.  It appears that you are using each TDM channel as a data record or struct with name-value property pairs attached to it that contain the actual measured data.  Yet you name each channel "array".  Do any of the channels actually have an array of unnamed values?  What is in the "Input Frequency" channel?

Given that you are measuring each channel of a multi-channel UUT, I think it makes sense to use the ChannelGroup hierarchy level as you have chosen to (UUT Channel Number).

You have the "UUT Temperature" property at the ChannelGroup level.  Why?  Does the UUT Temperature really change as you measure different UUT Channels, or is it really constant for each UUT measurement cycle?  If it's constant, then it belongs on the File level with the others.

Here's a current limitation of DIAdem that you might want to consider.  When you load more than 1 data file, say to compare 2 runs against each other, DIAdem only loads the File properties of the first file.  The ChannelGroup and Channel properties of both files make it into DIAdem, but the File properties of the second file do not.  So you might want to consider duplicating the File properties at the ChannelGroup level also.  If you do anticipate comparing multiple files together like this, then you will also want to add the UUT name to the beginning of the ChannelGroup Names.  Otherwise the "Channel 1" group from File 1 will collide with the "Channel 1" group from File 2, and File 2's will be called "Channel 11" instead of "Channel 1", because group names are auto-enumerated when they repeat.  In your case that would be ugly.  But neither the property duplication nor the "UUT XH7463G Channel 1" details are really needed if you're always going to be analyzing 1 file at a time, in isolation.

Do you have DIAdem 10?  You should.  The record-based approach with channel properties you have chosen is crying out for DataFinder access.  If you don't know what I'm talking about, ask me or check out www.ni.com/DIAdem.

By "body plot", did you mean "Bode plot"?  If so, how can you get a Bode plot from the data you've described in your TDM file? If not, what do you mean by "body plot"?  The short answer here is that any graphing you want to do in DIAdem needs to refer to channel values, not channel properties, so you might want to consider creating a "Graphs" or "Results" ChannelGroup with channels like "Frequency", "Transfer-Real", "Transfer-Imaginary" for a Bode plot, or even a ChannelGroup called "Statistics" with channels "UUT Name", "Input Range", "Input Impedance", "Measured Bandwidth".

Brad Turpin
DIAdem Product Support Engineer
National Instruments