New calibration light for spectrometers

The compact fluorescent light bulbs in the study work well for calibrating the spectrometers – providing good range of sharp lines across the range of the spectrometers.

However it is located in the study on a tall lamp stand and is not very portable.

Hence I purchased this small desk light for the purpose. The compact fluorescent bulbs turned out to be difficult to find as they are now outdated and replaced with LEDs. However, Wilkes had some on sale – I have tested the spectra of both old and new bulbs to ensure they match on principal lines and the match is virtually exact – see pictures below.

Great thing about this new light is that I can take it easily out into the field for calibrating telescope outside. Does need 240V but Ed’s marvellous battery box provides that – so no problems there.


Existing calibration light in study on tall stand and new portable desk light – bulbs in both are compact fluorescent lights:

CCDSPEC spectra of both types of bulb – amazing how similar they are given different manufacturers and bulb numbers. The wider peaks on the new bulb is due to over-saturation as the bulb is very bright and I was taking the spectra with CCDSPEC only 50cm away from the bulb. In practice this won’t be an issue as long as light placed some distance away. I may also purchase something to act as diffuser in front of light:

Existing CFL:

New CFL on small black flexible desk lamp base:

Spectroscopy from Lichfield with CCDSPEC – Damian and Andy

Damian and I went outside at my house in Lichfield and manually guided the spectroscope to get a bundle of spectrums of stars tonight. Once he had got the technique there was no stopping our boy Damian!

He calibrated the spectroscope using a compact fluorescent bulb – we used linear equation for figures for calibration file tonight.

Comparing images of spectra from brighter stars with online typical spectra – we could immediately find most of main lines generally within 2nm (20A) or real values. Wow!

Sky Watcher Equinox Pro 80mm, Manfrotto alt-az manual mount, hand guided, CCDSPEC, Nebulosity 4 software, QHY6 camera.

Note spectral classes:

Deneb A2 la
Altair A7V
Arcturus K1.1IIIFe
Vega A0V



N.B. With the peak of the Perseids tomorrow night and a reasonably clear night tonight we had expected to see a number of Perseid meteors…….Damian saw one in the nearly two hours we were outside and I missed even that one!


Calibration of the CCDSPEC/QHY6/Sky Watcher Equinox Pro 80mm/Nebulosity setup:

Calculation spreadsheet for higher order polynomials Excel Andrew Thornett 100818

Damian’s calibration below (his measurements circled):

Some spectra were amazing, showing incredible lines:



Spectral type A7V



Some spectra were not so amazing:

How to use the higher order co-efficients with PCSpectra

From: Kenneth Elliott
Sent: 10 August 2018 14:15
Subject: Re: 3rd order polynomials


Hi Andrew

This is how you use the higher order co-efficients with PCSpectra

Original (linear values entered into PCSpectra calibration file):

70.255, 0.6474

How to enter higher order co-efficients, x2, x3 and x4, into PCSpectra calibration file:

70.255, 0.6474, 0.001, 0.000000013, 0.00000000000121

Alternatively, scientific format can be used for co-efficients e.g.

7.0255E1, 6.474E-1, 1E-3, 1.3E-8, 1.21E-12

Pitfalls of using higher order co-efficients in PCSpectra:

There are two pitfalls

1: You need to have lots more lines if you use higher order fits and any fits outside the wavelength range you have got calibration lines may be wildly wrong as the equation has no constraints, so its fine with the Argon lamp.

2: you need lots of precision in the coefficients to get the accuracy, otherwise if it is multiplied by X**4 it could be wildly out.

How to improve accuracy/usefulness of higher order coefficients with PCSpectra:

Three different ways you could do this:

1. Astrosurf French site which has line idents and fitting programs, but I gave up trying to use it as, although the program runs in English, if there is an error then the error messages are in French.


2. Using STARLINK (UK) software – on Linux and needs a fair level of skill to get going.


3. The HST software both on Linux – on Linux and needs a fair level of skill to get going.



Developing a spreadsheet to calculate higher order polynomials for spectrometer calibration 10/8/2018

In the early hours of this morning, when I was having difficulty sleeping, I decided to have a go a seeing if I could do this (nothing on TV!) A few Google searches later and I found some good links of polynomial equation calculation and Microsoft Excel.

So this is my first attempt at working out polynomial equations, in orders from linear to 4th order – I am very proud! (Although please let me know if it is a load of rubbish!!)


Download Polynomial Calculator Spreadsheet:

Calculation spreadsheet for higher order polynomials Excel Andrew Thornett 100818

Screenshot of 3rd order polynomial calculation version of above spreadsheet, compared to Dr Elliott’s linear calulation spreadsheet for same data:

First attempt at calibrating DIY Spectrometer using Compact Fluorescent bulb

The problem with this calibration is that relationship between pixels on x-axis and wavelength is not a simple y=Ax+B for slit-based spectrometers. I used a variation of Dr Elliott’s calibration Excel spreadsheet here which does assume that this relationship holds. I need to create a spreadsheet that allows generation of higher order equations. The DIY Spectrometer control panel will accept two higher orders if available, improving accuracy.

My calibration today used a linear equation.


Click on link below to download calibration files from today, including my calibration Excel spreadsheet for generating the equation and A & B from readings off the spectrum of a calibration light source:

Calibrating DIY Spectrometer 090818@1400

Screenshot from Spectrum Studio (below) (control software for DIY Spectrometer) – the spectrum is one I have taken of a Compact Fluorescent Light Bulb. I am attempting to identify the elements of a line on the spectrum using the line identification function in the software – yellow line.

Note that spectrum covers a narrower range than CCDSPEC as the defraction grating is 1800 lines/mm in the DIY Spectrometer rather than CCDSPEC’s 600 lines/mm.

Comparison spectrum taken approx. 10 days ago with CCDSPEC (below):

Compact Fiber Coupled CCD Spectrometer Kit (DIY) 1800 l/mm (“DIY Spectrometer”)

Sold by Science-Surplus, these are fiber-coupled spectrometers sold on ebay. They provide quite high resolution at very cheap price – the big question is how it is going to be attached to my telescope but that is a challenge for another day!


Description of instrument from Science-Surplus:

This is a reconditioned compact fiber-coupled CCD spectrometer. It is a crossed Czerny-Turner design originally manufactured by B&W Tek, Model BTC-110S. It is an OEM instrument formerly used in a medical device and is sold “as-is.” The spectrometer includes an 1800 l/mm grating and the CCD detector is not actively cooled. The spectrometer is not aligned. However, it is “pre-aligned” so that the unit measures the spectral range of about 500 nm to 700 nm with a resolution typically better than 2-3 nm, The grating can be rotated to measure wavelengths below 365 or up to 700 nm (in first order) with a 200 nm total range. An order-sorting filter is typically not included. The spectral resolution can be 1 nm when the alignment is optimized. Other gratings can be purchased for different spectral coverage. Aligning the spectrometer is not trivial.

The kit includes the spectrometer, a custom enclosure (shown in the photo), a fiber optic collection cable (SMA), 5 volts DC power supply, and a serial port communication (D9) communication cable. The kit also includes access to our proprietary Spectrum Studio ©2018 software to read out the spectra, calibrate the wavelength scale, record and/or average the data with or without dark subtraction.

Spectrometer Characteristics

  • UV-NIR (coverage is typically 200 nm. Min/max wavelengths are 365 and 700 nm)
  • 1 nm Spectral Resolution (optimized)
  • 16 bit Digitizer
  • RMS read noise <50 counts (typical)
  • Sony ILX511 linear CCD detector array
  • ~350 ms Readout Time
  • 20 Hz analog spectral readout possible
  • overall dimensions: 5.75 x 3.75 x 1.75 inches

Software Description

  • Spectrum Studio, ©2018 Science-Surplus available as a download
  • Windows XP, Vista, 7, 8, and 10 compatible
  • Integration time from 50 to 65535 ms
  • Average 1 to 1,000,000,000 scans
  • Single scan or continuous scanning
  • Save and load spectra (csv file format)
  • Dark signal subtraction
  • User input coefficients for calibrating the wavelength scale
  • Instructions for creating your own custom software interface

System Requirements

  • Windows .NET framework 3.5 (in English)
  • Serial Communication
  • 9-pin Comm Port or USB to Serial adapter
  • [Note added by Andrew Thornett = works well through USB 2.0 with USB-serial adapter on my Windows 10 Dell 5 5000 laptop]

What’s in the box

  • Compact CCD Spectrometer
  • SMA fiber optic patch cable, 0.5 m length
  • 5V wallplug power supply
  • 9-pin Serial Communication Cable

Resources on sodium vapour discharge lamps (street lamps)

One of my next tasks to do with my spectrometer is to take a spectrum of the street lamps outside our house.

Here are some resources on low pressure and high pressure sodium vapour lamps found around Lichfield streets:









Why different image acquisition software programmes result in need to recalibrate CCDSPEC/QHY6

This follows on from Andrew’s post about needing to recalibrate his CCDSPEC/QHY6 when he changed image acquisition software from Nebulosity 4 to EZCAP:

Re-calibrating CCDSPEC/QHY6 combination for photos taken with EZCAP Image Acquisition software rather than Nebulosity software 4/8/2018

Regarding why recalibration is necessary when image acquisition software is changed:

Hi Andrew

It’s quite simple really.

Although the spectrum appears in exactly the same location on the chip, what comes out depends on how he data is clocked out and that depends on the software. Some chips have non sensitive pixels which are clicked out first and disregarded whilst other programs read out the whole chip which may be bigger than the imaging part.

So the pixel number in both X and Y can be different. Hence, it is essential to do the calibration and observations with the same readout program

Hope this helps.


PS I designed and built a CCD camera Electronics and software 30 years ago

Reflection on experience of doing spectroscopy so far

I have now been the proud and excited possessor of a spectroscope for about a month. If you have been following my posts, you will see they clearly demonstrate my intense interest and excitement in this new area (for me at least).

Statting with the fortunatel purchase of a competitively priced and very well designed system thayvallowed me to generste spectrums virtually immediately, it also allowed me to obtain star slwctra quickly and so i did not become put off by a significant learning curve.

Thay learning, i have found, does exist and can be frustrating but rewarding too when i get something to work.

I have discovered:

I can get spectra of daytime objects eaaily and quickly by just pointing the instrument at the ibject in question. This is rally easy to do and will work well for outreach. Unlike other astronomy branches, this one can be done in the day!

I can obtain spectra of bright stars with my 80mm Equinox Pro on alt-az mount also easily and it is not too much if an issue tk hand guide the telelscope to obtain these spectra. This is not mentioned in the manual and very little about it on the internet where guided exposures are the thing but my experience shows it is effective for bright objects which can easily be seen in the 10mm eyepiece in the guiding port.

The eyepiece in the guiding port simply needs to be focused on the slit – do this in the day. Then at night focus the star to a small round dot – as small as possible. When it is over the slit the star divides in two and spectrum evident in the guiding eyepiece. This is only when over the slit. The eyepiece in the guiding port has small field of view. It is really important to ensure the finder is aligned on the scope. I might add illuminated cross hairs to help here. Not on that yet.

The situation becomes more complocayed once i start to calibrate the spectrometer. Calibration helps as it enables accurate determination of the wavelength of peaks on the spectrum amd i can consistently obtain accuracy of calibration, it appears, within 1nm. Compact fluorescent bulbs seen to be good source of light for this exercise. They come with diffuser as part of the design (the glass cover is milky). They are small and portable. I could do with making a portable rig including such a bulb to take into the field as i have found that calonration needs to be done each time I change the telescope or even the image acquisition software! Basically I suspect it is best done every night the spectrometer is used. Comparison spectrum of fluorescent lights are available from Wikipedia together with information on the wavelengths of peaks on the spectrum. Although calibration is tine intensive, it ks rewarding and leads to exciting results when you can identify peaks correspnding to specific molecules or elements.

Although using the spectrometer hand guided with 80mm Equinox Pro was relatively easy, I have found that changing to the 120mm Equinox on EQ6 Pro mount is a lot more complex and time consuming and frustrating although potentially a lot more rewarding. Changing to this set up opens up the possibility of taking spectra of much fainter and difficult to find objects and taking much longer exposures including planetary nebulae and faint stars. I have only used this set up once to date (last night), and this experience demonstrated to me that the driven moint does allow me to take significantly longer exposures without having to keep moving the scope with fine control on handset even when mount is poorly aligned. I found that using the spectrometer eyepiece to align the mount was quite difficult as the eyepiece uses light from the reflection on the highly polished metal around the slit. This is a lot less efficient than a high quality mirror with costing ps and so the image is much fainter than I would normally see in an eyepiece in a diagonal. Two possible solutions to this problem for the future are:

1. Next time I could align the mount without the CCDSPEC in the scope and then insert the CCDSPEC. As the CCDSPEC will need to be calibrated and its weight needs to be taken into consideration when balancing the scope on the mount, in practice this will mean that I need to mount the telescope on the mount together with CCDSPEC in the focuser, calibrate the CCDSPEC/QHY6 whilst attached to the scope, take the CCDSPEC out, then align the mount without rhe CCDSPEC in situ, then reinsert the CCDSPEC and do observations. This leads to quite a lengthy and complex set up process and hence I can see the advantage of having an observatory with fixed and permanently aligned telescope for spectroscopy. In addition, I am worried that the mount will have difficulty being aligned without the weight of the CCDSPEC in the focuser due to balance having been achieved with the CCDSPEC in place.

2. A better suggestion has been made by Damian. He has suggested that rather than taking the CCDSPEC on and off to align the telescope, a preferable option would be to use the excellent new laser pointer made for us by Ed as part of the three star alignment of the mount. Great idea! In addition, changing the eyepiece on both the CCDSPEC and finder scopes to illuminated reticule versions will help.