How to delete "ghost"-channels?

[flupa]

Hi all,

I have the following problem: After performing an FFT, Diadem (9.1) creates me several ghost channels: These channels are empty, have no name and their number is not equal to their index.

Now first of all: Why do does channels appear?

Or: How can I delete them? I tried the following:

for i=GlobUsedChn to 1 step -1

   if chnname(i)="" then

      chndelete(i)

   end if

next

But because the number and index are not equal that delets only some of them.

Regards

Patrick

PS: sorry for my edit-war

Message Edited by flupa on 03-28-2008 01:27 PM

[JoBa]

Hello Patrick,

do you perform the FFT in Analysis or Script module?

If in script, could you send me a code snippet that would let me reproduce the problem?

Thank you,
Johannes

NI Germany 649024

[flupa]

Hi Johannes,

I perform it in script. I wrote that script weeks ago and it worked fine. But now those ghost channels happen to appear.

Here is the code snipped after which the ghost channels appear:

' 4. step: calculate FFT

' ------------------------------------------------------------------------------

FFTINTERVUSER ="LengthStartOverl"

FFTINTERVPARA(1) =16

FFTINTERVPARA(2) =intervalLength

FFTINTERVPARA(3) =1

FFTINTERVOVERL =0 ' Overlap in %

FFTWNDFCT ="Hanning"

FFTWNDPARA =10

FFTWNDCHN ="time"

FFTWNDCORRECTTYP ="periodic"

FFTAVERAGETYPE ="No"

FFTAMPLFIRST ="Amplitude"

FFTAMPL =1

FFTAMPLTYPE ="Ampl.Peak"

FFTCALC =0

FFTAMPLEXT ="No"

FFTPHASE =0

FFTCEPSTRUM =0

Call CHNFFT1("time",P)

' 5. step: transfer function

' ------------------------------------------------------------------------------

FFTINTERVUSER ="LengthStartOverl"

FFTINTERVPARA(1) =16

FFTINTERVPARA(2) =intervalLength

FFTINTERVPARA(3) =1

FFTINTERVOVERL =0

FFTWNDFCT ="Hanning"

FFTWNDPARA =10

FFTWNDCHN ="time"

FFTWNDCORRECTTYP ="periodic"

FFTAVERAGETYPE ="No"

FFTAMPLFIRST ="Amplitude"

FFTAMPL =0

FFTAMPLTYPE ="Ampl.Peak"

FFTCROSSSPECTR =0

FFTCOHERENCE =0

FFTTRANSFCTTYPE ="Spectrum H1"

FFTCROSSPHASE =0

FFTTRANSPHASE =1

Call CHNFFT2("time",keyphasorChn,P)

call msgboxdisp("Index = " & ChnIndex(GlobUsedChn))

call msgboxdisp("GlobUsedChn = " & GlobUsedChn)

HTH, Thanks

Patrick

[JoBa]

Hello Patrick,

I wasn't able to reproduce the generation of "ghost" channels.

Could you attach a short script and a data file that causes those channels to be generated?

Thank you,
Johannes

AE, NI Germany

[flupa]

Hello Johannes,

Thank you for your effort. For me it is clear that you can not generate any ghost channels because I think the code is right and the first month I used it those channels did not appear. A colleague of mine said that those ghost channels often appear in Diadem when using FFT. His "solution" is to delete them afterwards...

That is was I tried now. Here is what I found out. Because their number was not identical to their index I was not able to delete them as long as their index number is higher than the number of maximum channels (memory management settings). I did now increased that value and I am now able to delete them with the follwing function:

For i=(GlobUsedChn*1.5) to 1 step -1
     if ChnName(i)="" then
         Chndelete(i)
     end if
next

I am not pleased with a workaround like that but it works... What might be problem with those ghost channels?

Regards
Patrick

PS: Anyway if those channels happen to appear regularly this should alert you NI-guys.

Message Edited by flupa on 04-02-2008 08:16 AM

[JoshRew]

I don't know if you're still having this issue, but for others that may be having it:

I've noticed that if you delete every value in a channel through the script (DataBlDel) commands, the channel loses it's name.

To demonstrate, clear your data portal, create a channel (named whatever you like) and put a single value in it.

Then go to the script tab and run:

Call

DatablDel("1",1,1)

The channel will no longer have a name...

This was done in DIAdem 11.1, I don't know about 2010 or older versions.

So if your code has any block deletes, that's where the "ghost channels" are probably coming from.

-Josh

[flupa]

Dear Josh,

We solved the problem with the workaround I mentioned above. We are currently transform the code to DIAdem 11.1 and we will see whether this issue will come up again.

Thanks,

Patrick