DTT Imported Spectra Bugs
Multiple Bugs? If so, they might need to be split into multiple tickets.
Jeff Kissell wrote:
When importing .txt files that have been exported from other DTT sessions (I've only played with "power spectra" ASDs thus far), there are two bugs: (1) After import, the session does not retain the import if one restarts the measurement. Here's the scenario:
- I import a .txt ASD for comparison against a measurement taken in the past grabbed from the frames
- I realize the time stamp is wrong, so I change the time stamp, and start a new measurement.
- Imported spectrum disappears from the choices of what to plot. There is a "DTT Secrets" work around -- If you save the imported ASD as a reference, it will remain an option to plot after restarting the data.
(2) One cannot calibrate imported ASDs. One can go through all the motions in the calibration pop up window, but upon closing the window, it doesn't actually calibrate the ASD.
Example importable text file to play with: /ligo/svncommon/SusSVN/sus/trunk/HLTS/L1/Common/Data/2014-11-27_L1ISIHAM25_PerformanceASDs.txt
Which you can import as 6 channels (with the "Default" Conv.), named (i.e. what you type in the "A:" fields for columns 0 through 5) L1:SUS-PR3_M1_ISIINF_X_OUT_DQ L1:SUS-PR3_M1_ISIINF_RY_OUT_DQ L1:SUS-PR3_M1_ISIINF_RZ_OUT_DQ L1:SUS-PR3_M1_ISIINF_Y_OUT_DQ L1:SUS-PR3_M1_ISIINF_RX_OUT_DQ L1:SUS-PR3_M1_ISIINF_RZ_OUT_DQ
Other import params are all the default: Input Type: ASCII Input Format: XY and Zero Time Binary Option: Little Endian XML Option: Read All, Replace Existing, and Include Calibration
Data Type: Power Spectrum Data Range: Start Index = 0, Max Length = 1000000, Bin = 1
For the calibration, the raw text file is already in nm/rtHz or nrad/rtHz, so the calibration is a simple 1e-9 [m/nm] or [rad/nrad]. So, for example, enter in a calibration for L1:SUS-PR3_M1_ISIINF_X_OUT_DQ as (highlight that channel, hit "new") Reference: (some name, it doesn't matter) Unit: m Time: (some large time in the past) (hit "OK", select Pole/Zero tab, select "pole-zero" box, select "default" and enter in a) Gain: 1e-9 (with no poles or zeros, hit "set" or "OK") and watch how the channel doesn't change. It would start out as 1e3 [nm] at 0.1 [Hz], and after calibration should be 1e-6 [m] at 0.1 [Hz].
(3) (a) The time stamp, number of averages, and the effective binwidth of the measurement doesn't get imported, and (b) causes those identifiers to disappear from the plot -- even after selecting the cursor back on a non-imported trace. This is most likely because the ASCII export doesn't store any of this information.
ref bugzilla: https://bugzilla.ligo-wa.caltech.edu/bugzilla3/show_bug.cgi?id=768
Some work done by Jim Batch:
Some of this seems like expected behavior given that the ASCII file only contains XY data. There's no association between the ASCII data and any channel, other than what you enter when you import the data. So if you want to save the imported data for comparison with fresh data, you'd have to save as a reference.
The Calibration is a bit unexpected, I'll look in to that.
As you point out in 3, the ASCII export doesn't store the settings so they can't be restored. Only the XML format would do that.
I was able to apply calibrations to the imported ASCII data. It does appear that the combination of items checked or clicked, and the order of doing so matters (probably more than it should).
What I did:
- start diaggui
- File->Import...
- Fill in 6 names for columns 0-5, any name, set Data Type to Power Spectrum.
- Click Import.
- Select the data file, click Open. In main diaggui display -
- Click Result tab
- Display the 6 channels as power spectrum
- Click Calibration...
- Select 1st channel (not [Default-TF] or [Default])
- Click Default check box
- Click New
- Fill in some name in the Reference field
- Enter m as Unit:
- Click OK, leaving the Time as the default (6/1/1980, 0:00:00)
- Click Pole/Zeros tab
- Click Pole-zero check box
- Enter 1e-9 in Gain: field
- Click Set
- Click OK
At this point, the calibration is applied.
I did NOT change the time in the new calibration box. If you do change the time, the calibration won't be applied to the displayed data. Since the time is not part of the imported data, it's defaulting to June 1, 1980, 0:00:00.