Technical Issues - FilmStar News 2017
January 16 Photopic Response Curves
A user requested an
automatic way to plot photopic transmittance as a function of wavelength.
This is readily accomplished manually in Excel, but tedious. Our
solution combines the FilmStar Workbook, FilmStar BASIC and the
FSPlot Module. It works as well for photopic reflectance.
Referring to Setup...Parameters, we utilize Transmittance vs. Wavelength Range. This allows us to set graph range 350-800 even though CIE calculations are 380-780. The example uses CIE 1931 values (1 nm interval) and is easily modified. Users requiring help with FilmStar BASIC or Workbook should contact FTG Software.
January 17 Auto-Optimize in Workbook
Here's a simple illustrative example. Download AutoWbOpti.zip; unzip and copy as follows (assuming default directories).
You should get a similar graph after clicking <Ctrl+Z> for Zoom Mode. Changing the Workbook Macro from CALCULATE; ($D$1) to something like BasExec x=5; gives the BASIC Conflict message. The Workbook may be the most challenging module in FilmStar; users are invited to contact FTG Software for an online run-through via Glance Networks.
January 20 Zero Layers
As shown above, there is a great difference in the spectra when 14 layers (including a massive 1 mm Si substrate) are deleted. But as Zoom Mode shows below, the spectrum at short wavelengths is not affected by the removal, thus verifying the algorithm.
Previous versions failed when Side 2 Compensation <> FWD ignore Side 2. That is still the case, but now Zero Layers automatically changes Side 2 Compensation if necessary, thereby eliminating error messages. Substrates should be treated as massive layers; designs with parentheses are not supported. Most users will never require Zero Layers, but those that do consider it to be a very important capability.
February 17 MONITOR Turning Point Adjustment
In some cases it is useful for MONITOR to approximate a near-exact QWOT by a full QWOT. This is accomplished with new BASIC Sub TurnPtAdjust. By default only the last layer on a chip is modified, but that option can be changed as shown below. The magenta color indicates modified layers.
TurnPtAdjust can be
assigned to a macro button or automated by specifying it as a Calculation Macro
(new) in Tools.. Macro
Commands. The simplest macro is BasExec TurnPtAdjust. New
BASIC Sub WsSetNum and Sub WsSetText offer further customization. The
new functions are described in an updated help file.
February 23 MEASURE (Scantraq) Saving Two File Types
A user requested the ability to automatically save spectra in two formats. Click here to download BASIC code that does the job. Copy to your C:\Winfilm\Basic32 directory. In order to automatically save scans, enter the following in the Setup.. Scan Macros editor. Substitute Scantraq for Winfilm as required.
When there is no Scan Macros editor, as is the case for instruments such as Ocean Optics or FTIR, the code is specified at the end of the Sample Scan macro in the Control Macros dialog (part of the Setup.. Scan Method sequence).
Sub Main .... .... BasChain "C:\Winfilm\Basic32\Save_CSV&SPC.bas" End Sub
You can also enter the following in Tools.. Macro Commands, in this case assigning to Macro Button 4.
March 6 Identifying bad data
Measuring %R/%T for a fused silica substrate seems easy, but maybe not! A user, intending to deduce n&k vs. wavelength, sent us data collected with a PerkinElmer Lambda 1050 with a goniometer accessory and UV Winlab software. As this is a well-known material, expected values are easy to calculate. Comparing theory (SiO2 Sopra.itw) and measurements in FSPlot's View.. Zoom Mode <Ctrl+Z>, we note problems from 300-400 and 800-900 nm. There is no better way to identify bad data. Such comparison will be difficult (impossible??) in UV Winlab.
Transmittance results were disappointedly similar. Was integration time (scan speed) optimal? Was the tungsten lamp aligned properly, stabilized, and not approaching burnout? Operator error? 0% baseline compensation? Would FilmStar smoothing help? This includes the ability to smooth before normalization, thereby minimizing noise. That would not, however, improve 300-400 and 800-900 nm regions. Reasonable n&k results from 400-800 nm were deduced with INDEX Calculate n,k Algorithm #4.
Failure to examine data before hitting <Calculate> often leads to wasted hours, frustration, and possible crashes. If in doubt, send us your data (in one Excel file please). We note, however, that the accuracy required for n&k analysis typically exceeds that required for QA pass/fail specifications. When plotted on a 0-100% scale the above spectra give a different impression.
March 20 Preserving design format
Since solutions based on Notepad or Word/Excel may be inconvenient, 2.61.3980 adds a new option to the Groups editor: Edit...Lock Design. In this mode designs can be edited and evaluated but are locked against reprocessing, thus appearing as originally typed when saved and reopened. Optimization and other features are disabled in this mode. The status bar adds Design Lock when activated.