1. DSLR Photometry Highlights
Hi everyone, this morning I’ll briefly describe some of my DSLR photometry work.
In particular highlighting some collaborations with professional astronomers who needed photometry to complement their spectroscopic radial velocity measurements.
The stars are all brighter than magnitude 8 or so, which is great for spectroscopy with metre–class telescopes but far too bright for photometry using professional CCD cameras on these telescopes.
A DSLR camera and standard telephoto lens or small telescope is ideally suited for bright star photometry.
I’ll also describe the equipment I’ve used and one or two problems.
Mark Blackford

2. Outline DSLR Photometry outline E1 Region standard stars
Camera problem?
η (eta) Muscae eclipsing binary
KX Velorum eclipsing binary
HD a new variable?
GG Lupi eclipsing binary
Conclusions
This is what I hope to cover but time is short so we’ll see how far we get.

3. On the next few slides I’ll summarise benefits and limitations of DSLR photometry.
For a comprehensive discussion I refer you to the American Association of Variable Star Observers DSLR Observing Manual.
I helped write the first edition of the Manual back in 2013.
Since then I’ve run three online courses for AAVSO and re-wrote most of the manual to address feedback from participants.
The latest version can be downloaded from the URL shown here.
The next course is scheduled for June.

4. DSLR Photometry Summary
Benefits:
Simultaneous RGB (transform to BVRc)
Don’t need filter wheel and filters
Wide field of view (several target and comparison stars)
Relatively cheap
Drawbacks:
14 bit ADC (12 bit for older models)
Non-photometric filters
Image must be defocused
Small image scale (close stars overlap in images)
Some key pros and cons compared to standard CCD photometry are shown on this slide.

5. DSLR Photometry Summary
Equipment:
Basic Canon DSLR body (APS-C sensor)
Medium fixed focal length telephoto lenses
80mm f6 refractor
Computer controlled GOTO equatorial mount
TheSkyX, Backyard EOS, Dimension 4, MaxIm DL
Analysis:
AIP4Win, Muniwin, MaxIm DL
Bespoke Excel spreadsheets (data reduction, ToM)
Peranso, VStar
This is the equipment and software I use. Others have used different camera brands, shorter or longer focal length lenses or telescopes.
Non-tracking mounts can be used but limit what can be done.
There are many software options other than those shown here.

6. Key Points DSLR photometry can be accurate and precise
Transformation of instrumental RGB magnitudes is possible for normal stars with approximately black-body spectra
Instrumental magnitudes of stars with strong emission and/or absorption features cannot be reliably transformed (e.g. novae, supernovae)
Useful measurements can still be made of these stars, e.g. time of minimum or maximum light, pulsation period
All cameras, be they DSLR or CCD, measure slightly different magnitudes due to differences in filter band passes and detector sensitivity. We call these instrumental magnitudes because they are specific to the instrumentation used.
Transformation is the process of converting instrumental magnitudes to a common scale.
Bright emission lines or dark absorption features in a star’s spectrum can make transformed magnitudes inaccurate. DSLR photometrists need to be aware of which types of stars will have these problems.

7. E1 Region Transformed Magnitudes
Now on to the first example which illustrates how accurate DSLR photometry can be if correct techniques are employed.
20 images were recorded in quick succession of the same 6.5 x 4.5 degree patch of sky centred on the E1 Standard Star Region which contains ~150 stars with precisely measured magnitudes in UBVRI filters.
These standard star fields were developed for photoelectric photometry calibrations but are particularly are useful for calibrating and validating DSLR photometry.
The slide shows my transformed DSLR measurements of 15 of the brighter standard stars. Precision (stdev) and accuracy (delta) are very good, comparable with most amateur CCD photometry.
delta = measured magnitude minus catalogue magnitude

8. Canon 450D + Nikkor 180mm f2.8 lens Celestron 8” EdgeHD Celestron CGEM mount Canon 600D + Canon 200mm f2.8 lens
For these measurements I used a Canon 450D camera and 180mm Nikkor lens piggy-backed on an equatorially mounted 8” SCT.
Some time later I managed to break the USB connector off the cameras’ motherboard when closing the observatory so bought new Canon 600D and 200mm telephoto lens.

9. Camera Problem? Obvious conclusion: Canon 600D is faulty
After about a year using the 600D I started seeing odd time series light curves like this from Nova Cen 2013 (V1369 Cen).
The red and blue magnitudes were oscillating out of phase with each other while the average of the two green channels was pretty constant. Clearly this wasn’t a real phenomenon happening on the star but an aberration of the camera.
Some nights the photometry was fine, other nights I’d get these oscillations.
Eventually I bought a new Canon 1100D DSLR and mothballed the faulty 600D. All was fine for several more months until, bugger me, the oscillations returned!
This time I could correlated it with trying to image a fainter target and using less defocus to make the star image stand out from sky background better. It dawned on me that perhaps this was the problem, not faulty cameras.
Obvious conclusion: Canon 600D is faulty
Logical solution: buy new DSLR (Canon 1100D)

10. What’s Really Going On Here? Insufficient De-Focus!
So I looked back at the Nova Cen 2013 photos and measured star image FWHM and centroid position over the time series.
The mount was tracking at sidereal rate but was not guided so periodic error and drift due to imperfect polar alignment caused the centroid to wander across the CCD as shown by the red line here.
Light was concentrated into only a few pixels so as the star image moved across the CCD most of the intensity fell on either a red, green or blue pixel with very little intensity in surrounding pixels.
This was what caused the red and blue channel oscillations. Each of the two green channels also oscillated but by averaging the two together the oscillations cancelled out.
Star centroid position over 55 minute time series
Periodic error and drift
Centroid sometimes on a red pixel, sometimes blue or green
Very little intensity in surrounding pixels

11. Sufficient De-Focus Adequate sampling of all three colours
The solution was to use greater defocus to ensure each star image covers several pixels of each colour.
Other benefits are that bright stars are less likely to saturate and longer exposures can be used to increase the total photon count and therefore improve counting statistics.
I believe some DSLR photometrists may be using too little de-focus in an attempt to reach fainter targets or use shorter exposures and are therefore compromising the accuracy of their measurements.
The silver lining from all this was that the 600D camera was not faulty afterall, so I now had two perfectly good DSLR’s for photometry.
Adequate sampling of all three colours
Avoid saturation of bright stars
Increase total photon count by increasing exposure time

12. η Muscae Multiple star system with at least 5 components:
η Mus A, eclipsing binary (Aa and Ab), V mag, d period
η Mus B, 58 arcsec, 7.3 V mag, ~200,000 year orbit
η Mus C, 3 arcsec, 10 J mag, ~3,000 year orbit
η Mus D, ~5.5 year orbit inferred from radial velocity measurements after accounting for radial velocity changes due to orbit of η Mus A
The next example is a bright eclipsing binary in a multiple star system that has at least 5 stars in total.
Roger Butland and Ed Budding have been monitoring this system spectroscopically for several years and deduced the existence of the fifth star, eta Mus D, from variations in radial velocity.

13. This is also detectable as variation of eclipse times of minimum
η Muscae
This plot shows the residual radial velocity (gamma velocity) measurements and a theoretical fit to the data after accounting for orbital velocities of components Aa and Ab.
They realised eclipse timing variations should also be detectable due to the light travel time effect.
But they were not able to make the measurements themselves. Tom Richards suggested DSLR photometry might be suitable so I started observing eta Muscae in 2011.
(E. Budding, R. Butland and M. Blackford (2013), PASA, 30, e037 doi: /pasa )
This is also detectable as variation of eclipse times of minimum

14. Light Travel Time Effect (LITE)
Observer D Aa Ab X
Components B and C are much further out (not shown) and their orbital periods are much too long to influence the radial velocity and eclipse time of minimum measurements.
Component D is several magnitudes fainter than Aab and too close to resolve in images
As stars Aab and star D orbit their centre of gravity the eclipsing binary changes distance from the observer
~2.4 day orbit
~5.5 year orbit
NOT TO SCALE
Stars B and C much further away

15. Light Travel Time Effect (LITE)
Observer D X Aa Ab
The extra distance means eclipses are observed several minutes later due to finite speed of light
Extra distance due to orbit of η Mus A
around the center of gravity with η Mus D

16. Canon 1100D + Canon 200mm f2.8 lens Canon 600D + Orion ED80T CF f6 Celestron CGEM mount
Early observations were made with the 450D and 180mm lens
Later I used the 600D and 200mm lens
Followed by the 1100D with 200mm lens
And most recently the 600D and 80mm f6 (480mm f.l.) refractor.

17. η Mus image
Prior to mm or 200mm lenses were used and Star B could not be fully separated in the defocused images so it was included in measurements.
The longer focal length of the 80mm refractor allowed Star B to be excluded from the measurement
Zoomed insets shows eta Mus A (brighter component) and eta Mus B

18. 2014 & 2016 V light curves Green: May 2014 Red: Feb 2016
Here are two primary eclipse time series. The green one at top was recorded with the 200mm lens in mid 2014 and includes the Star B.
The red light curve was collected in March 2016 with the 80mm refractor and does not include Star B so is a little dimmer than the green curve.
Time of minimum light in each curve was measured and compared with the expected time if no Light Travel Time Effect was present. That is, phase 1.0 on the x-axis.
The measured difference between 2014 and 2016 amounts to 480 seconds which is much greater than the estimated measurement error of +/ days or 25 sec.
Green: May 2014
Red: Feb 2016

19. Observed – Calculated (O-C) Diagram
This is a plot of theoretical light travel time based on the orbit of Star D that Ed and Roger deduced from their radial velocity measurements.
Blue squares are my measured values from eclipse times of minimum light. The first few years seem to fit reasonably well to the theoretical curve.
Red star is where we would expect a measurement made in early 2014.

20. O-C Diagram
However subsequent measurements in 2104, 15 and 16 clearly deviate from the expected values.
Ed and Roger are analysing more recent spectroscopy and together with my DSLR photometry should be able to refine the orbital parameters of Star D.
We plan to publish a brief update paper later in the year.
Refinement of star D orbit is required (period, ellipticity, inclination)

21. KX Velorum
Pavel Mayer, Charles University in Prague, Czech Republic EA eclipsing binary with an evolved early type component V magnitude day period Only previously observed eclipse was More photometry and spectroscopy required to better determine masses, radii, temperatures, etc.
Pavel Mayer at Charles University, Czech Republic asked if I could try to record a primary eclipse of this binary.
He had observed it spectroscopically with the FEROS echelle fibre spectrograph on the 2.2 m telescope ESO at La Silla.
It has a long period so there are only 4 or 5 opportunities each season, and some of those will be during daylight.
The only known eclipse observation was by Balona and Laing in 1986, and only part of the eclipse was recorded.
P. Mayer, H. Drechsel, and A. Irrgang, Astron. Astrophys. 565, 86 (2014)

22. January 2015, 7 hr 45 min time series
KX Velorum
KX Vel just happened to be in the field of view of another binary I’d been observing for Burcu Ozkardes and Ed Budding. So I went back and measured KX Vel in those images but no eclipse was recorded.
In January 2015 I successfully observed an egress from primary eclipse, shown in red here.
Clouds have thwarted all subsequent attempts.
January 2015, 7 hr 45 min time series

23. Looks like bottom of eclipse was observed
KX Velorum
Expanding the plot it appears that the eclipse bottom may have just been recorded.
Looks like bottom of eclipse was observed

24. Portable Rig Canon 1100D + Canon 200mm f2.8 lens iOptron ZEQ25 mount
To maximise chances of clear skies next season I plan to take this portable mount and camera west of the Great Divide for a few days either side of the predicted times.
So hopefully by next NACAA I can present a more complete primary eclipse light curve and Pavel Mayer can finish his study of KX Velorum.

25. HD 92607
Selected as potential comparison star for QZ Car, however it soon became clear that it was not constant. Not in the General Catalog of Variable Stars (GCVS) The SIMBAD database lists it as a non-variable star with spectral type O9II/III. A search of AAVSO Variable Star Index for variables near RA and Dec of HD showed that the star was found to be variable by the All Sky Automated Survey (ASAS) in I band. Classified as an eclipsing binary with day period.
On Friday in the Variable Stars South Symposium Ed Budding gave a talk on QZ Car, an eclipsing binary in a massive quadruple star system.
I’ve been making DSLR time series observations for this project since January 2015 using the 600D and 80 mm refractor.
HD was conveniently located near QZ Car and none of my planetarium programs showed it to be variable so I thought it might be useful as a comparison star. However analysis of the first time series showed it clearly was not constant.
A check of the General Catalog of Variable Stars (GCVS) and SIMBAD database failed to find a variable at that position. So I thought maybe I’d stumbled upon a new discovery.
But a search of AAVSO Variable Star Index for variables near the RA and Dec of HD showed that the star was found to be variable by the All Sky Automated Survey (ASAS) and classified as an eclipsing binary with day period.

26. Eta Car
QZ Car
HD 92607 HD
92607

27. This is the ASAS V filter light curve acquired over about 7 years.
It appears to be the only available data on this EB prior to my DSLR observations.
HD ASAS-3 V Light Curve

28. Less scatter, primary ~0.01 mag deeper, slight O’Connell effect
My V filter light curve collected over the past year shows less scatter making it possible to identify a slightly deeper primary eclipse.
There is also a small O’Connell effect where the maxima are different, usually attributed to dark or bright spots on one of the stars.
Less scatter, primary ~0.01 mag deeper, slight O’Connell effect

29. B and R data acquired simultaneously, although somewhat noisier
An added bonus of DSLR photometry is that B and R filter data are recorded simultaneously with the V filter data.
These are somewhat more scattered because the camera is less sensitive to these wavelengths and there are two green pixels for every red and blue pixel.
Never-the-less, the three light curves can be analysed with modelling software to determine physical characteristics of the binary including mass, luminosities and temperatures of the two stars and orbital inclination.
B and R data acquired simultaneously, although somewhat noisier

30. GG Lupi
Absolute Parameters of Young Stars: GG Lup and μ1 Sco, E. Budding, R. Butland, M. Blackford, MNRAS, Volume 448, Issue 4, p
EB type detached eclipsing binary
Magnitude range 5.56 – 6.19 V
Period days
Elliptical orbit with eccentricity ~0.15
Period of apsidal motion ~100 years
This is another bright eclipsing binary that Ed, Roger and I have been observing.
Results of light curve and radial velocity modelling were published early last year. The modelling resulted in accurate masses, radii and temperatures of the two stars in the system.
But we didn’t look too closely at the apsidal motion of the orbit in that paper.

31. Apsidal motion
The precession of the periastron of a binary system resulting from tidal gravitational moments.
Precise times of minima for eccentric eclipsing binaries can be used to accurately determine the period of apsidal motion, providing a test of the effects of General Relativity.
The mean internal structure constant of the stars can also be estimated.
As the orbit precesses the phase at which the secondary eclipse occurs relative to the primary eclipse will change.

32. Observer B A
When the long orbital axis is aligned with the observer secondary eclipses are at phase 0.5 after primary eclipses.
Time from primary eclipse to secondary eclipse is the same as time from secondary eclipse to primary eclipse

33. Time from primary eclipse to secondary eclipse NOT the same as time from secondary eclipse to primary eclipse
Observer B A
When the long orbital axis is perpendicular to the observer secondary eclipses are at maximum displacement from phase 0.5.

34. GG Lupi 1985 Strömgren y light curve
This phased light curve was recorded with the Danish 0.5m telescope at European Southern Observatory at La Silla, Chile.
The secondary is at phase 0.5 and the duration is significantly long than the primary eclipse.
(J. Andersen, J.V. Clausen, and A. Gimenez, Astron. Astrophys. 277, (1993))
Secondary at phase 0.5, Primary 8.8% of light curve, secondary 12.6%

35. GG Lupi 2012 DSLR light curves
I collected this phased light curve in 2012.
The secondary was at that time at phase 0.4 and both eclipses have the same duration.
(Absolute parameters of young stars: GG Lup and μ1 Sco,
MNRAS (April 21, 2015) 448 (4): )
Secondary at phase 0.4, Primary 10.3% of light curve, secondary 10.3%

36. 2012 & 2014 V light curves Green: May 2012 Red: July 2014
In 2014 the time between primary and secondary eclipses had changed by about 240 seconds compared with the 2012 measurements.
So the effects of apsidal motion are quite apparent in only two years.
Green: May 2012
Red: July 2014

37. O-C diagram Petr Zasche, Charles University in Prague, Czech Republic
Using all available times of minimum from electronic photometry (PEP, CCD and DSLR) we can construct this Observed – Calculated Diagram.
Petr Zasche at Charles University in Prague kindly used his MatLab program to analyse this data to refine the orbital parameters, including ellipticity and apsidal period.
More observations every two years will help to refine these parameters even further.
Petr Zasche, Charles University in Prague, Czech Republic
JD0 = (± 0.016) HJD
Period = (± ) days (Sidereal)
Period = (± ) days (Anomalistic)
Ellipticity = (± 0.046), Apsidal period = (± ?)
More observations every couple of years

38. Canon 1100D + Canon 200mm f2.8 lens Canon 600D + Orion ED80T CF f6 Celestron 8” EdgeHD Soon to be equipped with a CCD and filter wheel Paramount MX+
In January I replaced my CGEM mount with a Paramount MX+ which can carry a lot more weight and has much better pointing and tracking accuracy.
So now all my photometry options are available on the one mount making it easier to switch between them depending on which target is being imaged.

39. All observing so far from here

40. Soon from here

41. Conclusions DSLR photometry can be both accurate and precise
Targets with strong emission or absorption spectra are not suitable for transformed magnitudes but useful observations are still possible (e.g. novae and supernovae, period analysis, times of minimum or maximum)
Amateurs with modest equipment can photometrically monitor stars that are too bright for professional telescopes
Pro-Am collaborations can be very rewarding