February 18   File List Dialogs

MEASURE COM Port Global Update

You replaced an XP computer running your Perkin Elmer Lambda and need to change COM1 to COM3 in multiple Scan Method files. Click here to download a FilmStar BASIC program which globally re-assigns COM. Be sure to back up *.scw files before running the code!

March 9   Excel Filter Analysis

March 16   Si3N4 on Si Case Study

March 20  Edge Filter Calculator

May 20   Optical Glass
 Revised June 2

A large number (731) of optical glasses are included in FilmStar DESIGN/INDEX. Dispersion (*.itw) files deduced from manufacturers' data will be found in ..\Winfilm\Index subdirectories. When re-saved in ..\Winfilm\Index they appear in Setup.. Film Indices Dispersion. Your old Schott Glass directory is renamed to Schott Glass Old, thereby adding another 182 entries. Can't find the glass you need? Click here for a Hoya-Schott-Ohara-Hikari-Sumita-CDGM cross-reference (compiled by Hoya Glass).

Support lapsed? Data files converted to Excel macro-enabled workbooks can be downloaded here: Ohara, Schott, Hoya, CDGM. The workbooks create W,n,k arrays which can be pasted and manually saved in INDEX.

If absorption is not an issue (k~0), the Glass menu in DESIGN Setup.. Film Indices and INDEX Functions..Fit Index inserts Sellmeier coefficients.  Note: If you insert a glass as film material, Type is changed to massive. Verify values in INDEX or in DESIGN using BASIC Sub GetIndex. The code below copies a W,n,k array to the clipboard in Excel format.

Sub Main   ' for FilmStar DESIGN
    Dim xn!, xk!, w!, s$
    For w = 400 To 800 Step 10
        GetIndex 2, w, xn, xk  ' or GetIndex "SUB", w, xn, xk
        s$ = s$ & CStr(w) & vbTab & Format(xn, "0.00000") & _
            vbTab & Format(xk, "0.00000E+00") & vbCrLf
    Next w
    Clipboard s$ ' Paste into INDEX or Excel; compare with manufacturer's data
End Sub

More conveniently, DESIGN 2.61.4510 adds Plot in INDEX to View menus in the Film Indices dialog and Stack Editor. This is useful in detecting spurious n&k solutions or working with alloys or Double Dispersion.

Workbooks and n&k tables are provided without warranty. Use at your own risk and compare with original Ohara, Schott, Hoya, CDGM data and/or RefractiveIndex.INFO. Please notify FTG Software of any discrepancies.

June 11   More Optical Glass

~3K optical classes are included in the Synopsys Code V XML glass catalog. Why so many? The Schott catalog lists 120 current glass types while the Code V catalog adds obsolete types giving 497 total.

XML to Excel conversion provides access to these glasses. Click here for a version for up-do-date FilmStar users (auto INDEX plot and save). Not using FilmStar? Click here for a generic version giving W,n,k arrays. It is your responsibility to verify results. Manufacturer's Excel data converted to macro-enabled workbooks can be downloaded here: Ohara, Schott, Hoya, CDGM. A Hoya Glass cross-reference can be downloaded here (XLSM). Click here for the CDGM Glass Finder (web app), here for the CDGM cross-reference (PDF) and here for the Ohara cross-reference (PDF).

August 20   Batch Convert Spectra to Excel

Do you have a multitude of spectral files in text format (CSV, ASC, TXT, TAB, etc.). Unfortunately these are nonstandard formats not easily shared by colleagues and customers. The situation is further confused by the use of "," as decimal point, "." as thousands separator and ";" as list separator in European notation.

A new Excel workbook (FilmStar not required, revised Aug 26) provides a general and international solution for text files containing a two column list of wavelength (nm, µm, 1/cm) and response (%R/%T/OD) values. Response may be in the range 0-1 instead of 0-100%.

Files may contain column headers as well as additional rows above and possibly below spectral data. In our solution TEXT specifies text file wavelength type while XLSX specifies type for the converted Excel file. Check FilmStar format to add column headers [e.g. W (nm) %T]. Should TEXT R/T range from 0-1, X 100 multiplies values by 100.

Thousands of files can be converted in minutes. While files can have different numbers of data points, row structure defined by Range must be the same within a directory. Extn is not utilized when converting single files. Original files (e.g. TXT, CSV) are not altered or deleted, but prudence dictates that files be backed-up before utilizing the workbook. Suggestion: convert and verify a single file before proceeding. When converting multiple files, the first conversion is displayed for review and confirmation. Delimiters are

In addition to text files with two-column data, JCAMP-DX format is also supported. Since JCAMP includes format information, most choices are irrelevant and grayed-out as shown below.

When converting multiple files, the first conversion is displayed for review and confirmation. The workbook has not been tested with all possible files. Please e-mail samples of files incorrectly converted and include country setting if using "," as decimal point.

FilmStar users can convert MEASURE files with a BASIC macro based on the following. Contact FTG Software should modifications be required. Original files are not deleted. Note: nm limitation in FilmStar CSV.

' SpcConvert.bas for FilmStar MEASURE
' Converts FilmStar spectral file formats
' Paths must already exist
Const Path1$ = "C:\Winfilm\Spectra1\"
Const Ext1$ = ".dx" ' .csv/.dx/.spc/.xls/.xlsx    CSV: NM ONLY
Const Path2$ = "C:\Winfilm\Spectra2\"
Const Ext2$ = ".xlsx" ' .csv/.dx/.spc/.xls/.xlsx  CSV: NM ONLY

Sub Main
    k% = Len(Ext1$)
    s$ = Dir(Path1$ & "*" & Ext1$)
    Do
        i& = i& + 1
        FileOpen Path1$ & s$
        FileSave Path2$ & Left$(s$, Len(s$) - k%) & Ext2$
        s$ = Dir$()
        StatusLabel "Converting file #" & CStr(i&)
    Loop Until s$=""
    StatusLabel ""
End Sub

October 8   MEASURE (Scantraq) Response Bar

A response bar graph has been added to versions for PerkinElmer Lambda instruments. The graph is accessed via the Lambda Commander in Mpe983 and Setup..Response Bar in Mpe950 and Mpe1050.

Right click on the graph to adjust the bar's maximum value (default 100).

October 12   Excel (Office 365), Windows 10 Issues

MS Office includes 32-bit and 64-bit versions. 32-bit is recommended.

A user reported that FilmStar DESIGN BASIC macro ExcelRand.bas did not work in Windows 10. Upon testing, we found it to work in Windows 7 (64-bit) with both 32 and 64-bit Excel and Windows 10 with 32-bit Excel.

ExcelRand.bas inserts tolerancing spectra

At first, it appears that 64-bit Excel and Windows 10 are incompatible with FilmStar BASIC. The usual fix of deleting and resetting the reference fails because the reference does not appear in Edit..References. If you examine the .bas code in a text editor (Notepad++ recommended) you see the '#Reference line that is hidden in the FilmStar BASIC editor.

'#Reference {00020813-0000-0000-C000-000000000046}#1.9#0#C:\Program Files (x86)\Microsoft Office\Root\Office16\EXCEL.EXE#Microsoft Excel 16.0 Object Library
'ExcelRand.bas for FilmStar DESIGN (early binding)
...
Const xlFile$ = "c:\Winfilm\Excel 2007\RandData.xlsx"
...
Dim xlApp As Excel.Application
Dim xlBook As Excel.Workbook
Dim xlSpectra As Excel.Worksheet
Dim xlLayers As Excel.Worksheet
....
Set xlApp = New Excel.Application
...
AppActivate "RandData.xlsx"      ' Current title format
'AppActivate "Microsoft Excel"   ' Older Excel versions

Referring to the '#Reference line shown above, Excel 64-bit worked in Windows 10 when (x86) was deleted in a text editor. Edit..References lists it as a 'Bad reference', but it works. Apparently, Windows 7 substitutes 64-bit \Program Files\ for 32-bit \Program Files (x86)\ without the 'Bad reference' designation. As Windows is constantly updated, it is possible that our results might not be universal.

Such incompatibilities are avoided through late binding instead of early binding. Since coding is harder as intellisense becomes unavailable, an option is to develop code on a machine where early binding works and then implement late binding when distributing to others who might have various versions of Excel and/or Windows. Note that the BASIC code does not include a '#Reference line.

'ExcelRandLB.bas for FilmStar DESIGN (late binding)
...
Const xlFile$ = "c:\Winfilm\Excel 2007\RandData.xlsx"
...
Dim xlApp As Object
Dim xlBook As Object
Dim xlSpectra As Object
Dim xlLayers As Object
...
Set xlApp = CreateObject("Excel.Application")
...
AppActivate "RandData.xlsx"      ' Current title format
'AppActivate "Microsoft Excel"   ' Older Excel versions

Previous related discussions: Excel Issues and Class References. Microsoft seems to be limiting 64-bit Office references to other Office applications in Windows 10. This could be a mistake or a security feature. If still using Windows XP: ExcelRandXP.bas, RandData.xls (copy to ..\Winfilm\Excel 97). Please contact FTG Software regarding any other issues raised by the latest versions of Office.

November 2   Stack Mode Issues

Stack Mode is intended for calculations involving many layers of many materials. While of little benefit to most users, Stack Mode is critical in the fabrication of certain devices.

Test example: StackTest.faw & StackTest.dmw   Copy to ..\Winfilm\Designs

A user reported issues in the following sequence: copy stack, expand repeated layers, paste original stack. In the DESIGN BASIC code below we use FSPlot to verify that the pasted original design is correctly evaluated. Reversed shows that we did not merely replot the same spectrum four times.

Sub Main          ' STACK MODE REQUIRED
    PlotClose     ' Close FSPlot if open
    PlotActivate  ' Activate FSPlot
    Calculate     ' Original (copy to clipboard)
    PlotNext
    StackAction 4 ' Copy design to clipboard
    StackAction 1 ' Expand repeated layers
    Calculate
    PlotNext
    StackAction 3 ' Reverse layers
    Calculate
    PlotNext
    AxesDraw
    StackAction 5
    Calculate     ' Pasted Original
    PlotNext
End Sub

(The same sequence is possible through menu commands.) As shown below the Reversed spectrum is different, as expected.

Stack Mode includes two similar 'reverse' functions: 1. Reverse layers (StackAction 3) which does not change AIR and SUB, 2. Reverse stack (StackAction 2) which switches AIR and SUB. In the second case Original and Reversed spectra are the same, as shown below (%T always the same, %R the same when k=0).

November 25   Collector Speed Issues

A user requested a more efficient Excel VBA method for importing data arrays as shown below. It was hoped that replacing a series of Collect calls with a single call would be speedier.

Blue text indicates spectra generated in DESIGN.

A custom Collection was implemented in BASIC Sub CollectThkVar. DESIGN 2.61.4552 or newer required. In addition, Subs CollectCopy and CollectSave make it easy to transfer spectra to Excel. The new functions are described in <F1> help.

Comparing calculation speeds for Sub Collect (repeated calls) and Sub CollectThkVar (single call) using the Excel workbook and sample design uploaded here, we could not discern any non-trivial speed differences.

Referring to RTvsThk.xlsm, note how the Collector is much faster (X3) if data is unformatted. Unfortunately, testing indicates that the Collector slows as it becomes bloated with data. While not an issue in typical use, the Collector bogs down with huge data sets. Even worse, CollectThkVar ultimately fails with the error shown below. Should that happen, click <Cancel> and delete Design1.exe in Task Manager. An easier way to do the same is the FTG Process Ender.

Another idea is to utilize GetRTData (Method #1 in RTvsThk.xlsm) and insert %R and/or %T directly into Excel cells, as shown in the following VBA snippet:

dBasic.SetLayers qm(), qmt$() ' Collector not utilized
For w = wMin To wMax Step wDelta
    kRow = kRow + 1
    dBasic.GetRTData w, 0, rs, rp, ts, tp, True
    Sheet1.Cells(kRow, kCol) = tp
Next w

While this method doesn't bog down, it is suboptimal. The fastest method (#7) utilizes BASIC's Calculate and Spectrum_Y and inserts data into cells as arrays. Differences between the fastest and slowest methods are a factor of ~30.

December 2   Bandwidth in Excel VBA

A user requires a simple method for computing center wavelength and bandwidth. We first thought to apply VBA techniques previously implemented in automation solutions, but decided it was more instructive to utilize worksheet functions instead, thus NO VBA.

Contact FTG Software for further explanation and/or to
request modifications meeting your specific requirements.

FSPlot Module verifies bandwidth calculations

FREE download: BW_NoVBA.xlsx. FilmStar is NOT required! Paste spectra (nm, µm, 1/cm) in A&B and adjust Level in G4. The worksheet is a great example of the useful (obscure?) INDIRECT function.

Opinion: The filter industry seems stuck with QA methods developed before computers were invented. Qualifying filters by 'spectral geometry' results in good filters being rejected and poor filters accepted. We welcome your comments!

December 11   BASIC Calculate Revision

Consider a quarter wave stack (1H 1L)10 with nH = 2.3, nL = 1.46, kL = 1000. Clicking Calculate or pressing <F9> gives the Overflow error. BASIC Sub Calculate, however, ignores the error with erroneous results. The issue has been addressed in DESIGN 2.61.4560 by replacing Sub Calculate() with Function Calculate() and adding Property DesignError$ as illustrate below.

If Not Calculate() Then
    MsgBox DesignError$, vbCritical, "Calculate"
    End
End If

or when running DESIGN from Excel VBA...

If Not dBasic.Calculate() Then
    MsgBox dBasic.DesignError$, vbCritical, "Calculate"
    Set dBasic = Nothing
    GoTo DoTheNeedful
End If

Previous code using Calculate will work without modification.

December 14   Veeco Spector OMS Support

A new version of SpectorOMS.bas adds Cauchy dispersion. CAUCHYM is the built-in DESIGN/INDEX function: n = A + B/W^2 + C/W^4, k = 0 (W in µm).

' SpectorOMS.bas for FilmStar DESIGN
' Copyright 2020 FTG Software Associates
' Creates coating files (*.fsd) for Veeco Spector OMS
' Version 1 Layers only
' Version 2 adds Cauchy dispersion

' ***************** Change as required **************************
Const fPath$ = PgmPath & "\Spector" ' usually c:\Winfilm\Spector
Const Title$ = "Spector OMS Coating File"
Const Version = 2
' ***************************************************************

Designs (SpectorTest.faw utilized here) are specified in optical thickness with QWOT=1.0. The BASIC code checks for these conditions. A Version 1 recipe eliminates SUBSTRATE: and MATERIAL: lines but is otherwise identical.

ITISPECTORDESIGN
VERSION:2.0
DESIGNWAVELENGTH:550.00
SUBSTRATE:CAUCHY:1.5047:3.98748E-03:3.02956E-05
MATERIAL:L:CAUCHY:1.3784:.0016959:2.34207E-05
MATERIAL:H:CAUCHY:2.07845:.020082:1.11098E-03
LAYERS:20
LAYER:1:H:0.72725
LAYER:2:L:0.75417
LAYER:3:H:1.00000
LAYER:4:L:1.00000
   ...
LAYER:17:H:1.10767
LAYER:18:L:0.21573
LAYER:19:H:0.62411
LAYER:20:L:1.26478

December 21   Spectrophotometer Noise Issues

Spectrophotometer detector/grating/lamp changes can be troublesome. A new MEASURE option excludes noisy data with BASIC Function ExcludeRange. DataConvert optionally inserts missing data points.

Sub Main  ' Convert and change data spacing
    ExcludeRange 850, 970
    DataConvert 700, 1000, 2
End Sub

Sub Main  ' Convert using Sample scan range
    If ExcludeRange(850, 970) Then DataConvert
End Sub

Examining the BK7 graph below (from Lambda 900), it appears that the IR-Vis transition has negligible noise with sufficient signal averaging time.

A closer look (FSPlot Zoom Mode <Ctrl+Z>) shows that noise is not completely eliminated. Whether or not that is important depends on specifications.

Since the spectrum of BK7 is flat from 850 to 970 nm, we can safely replace noisy data with interpolated points. ExcludeRange can be applied repeatedly for multiple ranges. Other options to reduce noise:

1. Cary machines include a Signal to Noise Mode which (we assume) relates averaging time to signal strength. That should reduce noise while optimizing throughput.

2. PE Lambda 1050 adds an InGaAs near-IR detector which considerably reduces noise.

3. Lambda 900-1050: Expert Mode allows custom Slit Width and Signal Averaging Time vs. wavelength. One could, for example, increase values in the range 850 to 970.

4. A spectroscopist suggested utilizing a high reflector instead of BK7 for the 100% Reference scan. That requires its reflectance to be known (STAN-SSH-NIST). No traceable standard? A VW (absolute) reflectance accessory could also be utilized.

Comments from readers with Lambda and/or Cary UV/Vis/NIR instruments are welcome.

December 23   Electric Field Algorithm