DSLR photometry, beginner's first impressions

Hi Heinz, It's great that you are venturing into DSLR photometry. Don't get discouraged. Observers have shown it can be done and very good data obtained. What I recommend is when taking all your frames, stack them, subtract the dark and divide the flat (if used) then work with the resulting single image to split out the green plane. I suggest trying that and seeing what you get. Continue to check that you get the proper magnitude difference between eta and zeta. If that is working you should get close to the proper value for epsilon. THe only thing I can think that may be a problem, but should not be that extreme is that epsilon will be at a slightly lower air mass (higher in the sky) than the other two stars and thus will appear a bit brighter. But it should not be that much brighter. Try some more measurements and let me know what you get. Good luck. Jeff

Hmmm.... I now read this presentation http://www.hposoft.com/EAur09/EAUR%20pdfs/%20%20DSLR%20Photometry.ppt and got the idea that I would have to apply a correction factor epsilon * (B-V) to adjust the measurement for the radically different color indices (B-V) of eps, eta and zeta Aur where epsilon is a camera-specific constant (not sure how to measure it, tho). Heinz

Hi Heinz, What DSLR are you using? In general, cameras of the same model have similar sensors, so the transformation coefficient that takes the DSLR green channel magnitude and converts it into a Johnson V magnitude will be about the same value. The way to determine it is to image a standard field (like M67) and plot the difference between your instrumental magnitude and the standard magnitude, versus the color of each standard star that you measured. This plot will look pretty much like a straight line, and the slope is that transformation coefficient for which you are looking. However, this coefficient is usually pretty small; Des Loughney found a value of 0.16 for his DSLR's green channel. The color difference between eps Aur and eta Aur is about 0.685, so it means that he added 0.11 mags to his estimate in order to obtain a Johnson V magnitude. While your Olympus will have a different coefficient, it will likely be about the same size. For your estimate of eps Aur of 2.6-2.7, this would mean your final estimate would be 2.7-2.8, which is still too bright. I'd think one of two remaining problems might be the cause: 1) the airmass/atmospheric extinction, as mentioned by Jeff, if you are observing near the horizon; 2) saturation, if your exposures are too long. You might try shorter exposures, and follow the procedure suggested by Jeff: stack first, then extract the green subimage for measurement. Arne

Hi Arne, Thanks for your comments. I'm using an Olympus E 420 , admittedly an entry model but the sensor should be the same as in some more expensive Olympus models so I guess in principle it should do. If anybody has some color calibration or transformation data for this model I'd be most interested, of course. Athmospherix extinction: Yes, that might be a problem. From my observation position & time, eps Aur is only about 20.4 deg above the horizon, eta Aur about 17.8 deg. Saturation is definitely not a problem, if anything it could be the opposite, or maybe undersampling. Other things to consider: Focus: I'm fighting with focus a bit, I've got two more or less reproducible focus settings. One is by turning the focus ring in manual mode to the very end, this turns each star into a rather huge defocused annulus with radius (center to outer edge) of around 10 pixels. Second option: When you power-off the camera in AF mode , the focus ring of the lens will be "parked" at a distinct position quite near the real infinity focus position. So you switch to AF, power down, switch to manual focus, power on , and voila. This gives only moderately defocused donuts with a smaller radius (maybe 4 px) and only a tiny hole in the middle. If you try to get the focus "perfect" you definitely get undersampling (most of the light is in just 2 or three pixels some time) and very short exposure time before saturation occurs (the DSLR has only 12 bits of resolution). Sensitivity: I was starting the experiments at 1600 ISO, which might not be wise (after reading the presentation linked above). So I'll try to use more moderate settings. At 1600 ISO, my guesstimation is that I get 1 ADU for every 10 electrons (judging from noise distribution). Field: With only 12 bits of dynamic range, I'm not sure if I have any other options for comparison stars within the same frame (using a 55mm lens, about equivalent to a 110 mm lens on an analog SLR). Lambda Aur is more similiar in color index, but even closer to the horizon and so much fainter so that it's not trivial to keep lambda and epsilon Aur in the linear range of the sensor. Bias: When shooting exposures with the minimal exposure time of 1/4000 sec, you can see that reading out the CMOS sensor produces some noise that tends to have patterns as horizontal lines (very faint of course, not solid lines :-) ). Problem is that the lines are in different locations with each shot (!!), so you can't hope to eliminate them with bias shots. I was thinking if correcting the final images by computing the median value for each line and then flatten the whole image line by line by adding / subtracting the line's median brightness difference with the overall median brightness Until now I've only had time to analyze one batch of frames (sharper focus and shorter exposure time) and will now analyse the rest (longer exposure and blurrier focus). If that doesn't work I'll try something like ISO400 or so next .... weather permitting (it's overcast now). It sure is fun to try to develop the right technique, tho. Thanks again Heinz

Ah yes......!!! With frames that had longer integration time (stacked 10 x 1 sec) and doing color split after stacking, I get much better results (magnitude difference = 0.29 ). Still some headroom before saturation should set in (max ADU for Eps Aur in averaged stacked image is only ca 1000 ). The imaging was done at Aug 23 around 23:00 UT, judging from other reports maybe I was quite lucky to get some reference shots before the visible start of the eclipse. Looking forward to new measurements. Heinz EDIT: And I just discovered that the strange result from the first series of frames was probably caused by a combination of alignment and dark frame problem during stacking. When doing the alignment manually and with the correct set of dark frames .... the result was 0.3 mag difference for that series as well. Lesson learnt: always check visually what comes out of those automatic processing steps :-)

Hi Heinz, Good to hear that you've found most of your problem. When I'm stacking images, especially at the beginning of a project, I usually look at every sub-image to ensure it looks ok, and inspect the result as well to see that the noise was properly reduced. I also often measure a couple of the shorter exposures individually to see that the photometry looks good. Much of good photometry is a combination of "art" and "perseverence". You learn what works, and you apply your lessons every time thereafter. One of the best lessons is to take multiple sets. If you are stacking 10 images, then take 30 so that you can create 3 stacks. Then compare all three results - they better be pretty close to one another. Arne

Hi!

Yet another stack of frames yielded a 0.3 mag diff to eta Aur, so I guess except for calibration for color index, I'm now getting close to the point where I can make reports.

Just to give an impression, here's a JPEGcrop of the stacked, calibrated and color filtered (G) image I used with AIP4WIN: Just for the sake of completeness, the rather giant FITS file this was taken from can be downloaded as well if you are patient www.tickuhr.de/epsAur/green-10x2.5_-_dark_20092308T223000.fits But if anything it's probably the header that could be interesting: ============================== SIMPLE = T / file does conform to FITS standard BITPIX = -32 / number of bits per data pixel NAXIS = 2 / number of data axes NAXIS1 = 3720 / length of data axis 1 NAXIS2 = 2800 / length of data axis 2 EXTEND = T / FITS dataset may contain extensions BZERO = 0.00000000000E00 / Real = FITS_value * BSCALE + BZERO BSCALE = 1.00000000000E00 / OBJECT = '*' INSTRUME= 'OLYMPUS E-420' EXPTIME = 2.5000 / Exposure (integration) in seconds DATE-LOC= '2009-08-24T01:30:55' /LT FOCALLEN 55.0 mm FOCALRAT f/4.0 ISOSPEED 1600.0 ASA COMMENT -------------------------------- COMMENT History with AIP4Win v2.3.0, Registered to Heinz-Bernd Eggenstein COMMENT Process date 08.26.09 HISTORY Converted from raw DSLR image using AHD Bayer interpolation HISTORY Camera-provided white balance HISTORY AIP4WIN V2 AutoProcess Tool HISTORY Original images: P8241975.ORF - P8241985.ORF HISTORY Calibrated HISTORY Aligned using 2 stars with manual star tracking HISTORY Average of 10 images HISTORY Green channel extracted COMMENT Pre_Mul(0) = 2,11946 COMMENT Pre_Mul(1) = 0,93841 COMMENT Pre_Mul(2) = 1,1367 COMMENT Pre_Mul(3) = 0,0 COMMENT Cam_Mul(0) = 1,96875 COMMENT Cam_Mul(1) = 1,0 COMMENT Cam_Mul(2) = 1,39844 COMMENT Cam_Mul(3) = 0,0 COMMENT Filter Array = RGGBRGGBRGGBRGGB END ============= CS Heinz

Hi Heinz, I'm a bit surprised - the image shows donuts for the out of focus stars, as you mentioned earlier. However, most DSLRs that I know about use standard camera lenses, which are just multi-element refractors. An out of focus star should be Gaussian in profile, without the central hole. Usually you only get central holes when you have some system with a secondary mirror in the way - a Schmidt Cassegrain telescope, a catadioptric camera lens, etc. This one also looks somewhat out of collimation, as one side of the ring is brighter. However, it does look like you have found the right field. eta Aur is V=3.172, so this would make your observation, without transformation, to be about 2.87. Jeff's estimate is about 3.07 for recent measurements, so it looks like you need to add about 0.20mag to your values to correct them for the difference between your passband (Olympus green) and the standard Johnson V. This would imply a transformation coefficient of about 0.28, which is not outside a reasonable possibility. See what images taken on subsequent nights give you. Good start! Arne

Hi! Yeah, I was wondering about the donuts as well....they are visible in live view and individual frames, so they are "real" and not an artefact of whathever post-processing. It's a simple SIGMA55-200 mm zoom lens, definitely a refractor. The image is more like a gaussian disk when you are approaching the optimal focus point from the "near" end and becomes a donut when you focus past the infinity focus point. I hope this doesn't mess up AIP4Win's measurements, as this is the easiest way to get a reproducible focus setting each night (just turn the focus ring all the way). At longer focal lengths the donuts also blur into disks. Go figure. CS Heinz

Ihave a theory about the donuts. After inspecting some of the shots before the color split is done, I get the idea that this is caused by chromatic aberration: Out of focus the chromatic aberration seems to split the spectrum across the disk, and if the part that appears at the center is not that bright or filtered away when the color split is applied, you get donuts. Does it make sense ? CS Heinz

Heinz, I think that the magnitude estimates that you are getting are due to atmospheric extinction ( including some haze ) if you are observing epsilon at an altitude of 20 degrees. I recently observed epsilon at an altitude of 25 degrees. The correction using eta as a comparison was around 0.08 ( subtracted from the apparent V magnitude of epsilon ). At 30 degrees altitude the correction was about 0.02. Last night I looked at epsilon when it was high in the sky ( from Edinburgh, Scotland ) and got good results. The estimate was: 29th August 2009: JD 5072.656: 3.052V - Standard error: 0.006. I use a Canon 450D DSLR with a fixed 85 mm lens, undriven, on a tripod. The camera settings are 5 seconds, ISO 200 and f 5. I obtain five sets of ten images. Each set is average stacked with AIP4WIN. The Green channel of the average stack is then analysed with the photometric tool. Once one gets familiar with the procedure the time it takes to get 50 images and analyse them is about 45 minutes. Des Loughney

Hi Des, Thanks for your interesting comments. May I ask how you manage focus, do you defocus or isn't that necessary with your longer focal length and ISO 200 sensitivity? Thanks, Heinz

Hello Heinz, Ideally, the images should be slightly defocused. However, I have found that there is no need to do this with exposures of 5 seconds with an undriven camera and a 85 mm lens. The rotation of the Earth, at that exposure, slightly blurs the images. Therefore, when I am getting the camera ready to take images I get the target star in sharp focus so that the images are not too blurred. I have marked the point on the lens at which infinity focus occurs. This can vary slightly as the ambient temperature changes so I have a number of points. The variation in infinity focus is a little bit greater with a larger lens such as the Canon 200 mm. I use this to do photometry on stars of down to magnitude 10. I believe that the variation in infinity focus is the due the expansion/ contraction of the lenses. 5 seconds is the maximum undriven exposure with the 85 mm lens except near the celestial poles where a longer exposure is possible ( 6/7 seconds ). I try, as a rule, to use a five second exposure as it minimises scintillation problems. I hope that this is helpful. Des

Hi! Ok, I have made some more experiments using longer exposure time and different sensitivity, and the result match far better what you'd expect. And I've tried to estimate the correction for color index eps*(B-V), even tho extinction/air-mass seems to play a bigger role for my observation window. When picking stars with a similar air mass index, comparable magnitude but different color index and adjusting eps so that diffential magnitudes match published values, I get quite a good fit put the correction factor eps comes out having a different sign! E.g. it's more like -0.08, as compared to +0.16 for your Canon. Would that be sensible at all or is it carved in stone for some technical reason that epsilon is always positive? Thanks Heinz

Heinz,Here is the procedure for working out the correction using my transformation coefficient of 0.15 ( not 0.16):"Epsilon Aurigae has a (B-V) colour index of +0.537. The comparison star I use is eta Aur which has a (B-V) of - 0.148. The total difference in (B-V) is therefore 0.537 plus 0.148 which gives a total of 0.685. When you multiply this figure by the TC ( 0.15) you get 0.1025. This is what I add to my green channel magnitude estimates to yield an accurate V magnitude."If I find by using eta Aur as the comparison that the V magnitude of epsilon is 3.15 then a correction has to be made. Epsilon is 0.1025 brighter which means that this value is subtracted from 3.15 to give 3.048V.Des

Hi Des, I see, this is the way I do it as well. The doubt I'm having is with my result of the correction factor (0.15 in your case) in the first place. What I did was this: 1) measure a few stars in the same field (there are some near lambda Aur that seem to be still bright enough to get a linear response from the sensors and should have similar extinction amaong them) 2) plot the difference of the published amd measured magnitudes (y axis) vs. their published (B-V) values (x-axis). It's reasonably close to a straight line +/- measurement uncertainties 3) do linear regression on the data points, ==> the slope should be the correction factor. As a sanity check, test this correction factor on eta and zeta Aur and see if the corrected relative magnitude is closer to the published value than before the correction. It is!!! Now it seems that my correction factor has a different sign : -0.08 ! If my factor is indeed correct, I would have to subtract (!) 0.05 mag from my raw estimate (now around 3.1 with longer exposure and probably less haze than before when I got lower values, so my final estimate would be close to 3.05 for eps Aur on Aug 31st) Arne mentioned that this correction factor should be similar for different DSLR sensors, that's while I'm a bit skeptical about my calibration attempt, tho. CS Heinz

Heinz,Thanks for the email. Your estimate for epsilon on 31st August sounds about right ie other observers seem to be making epsilon to be around 3.05 or 3.06 at present. I have also said that the transformation coefficient should be similar for Canon models. This is an assumption which may be unwarranted. I have only studied Canon 350D and Canon 450D models. I do not know whether the infra red filters in Canon cameras vary or if these filters are similar in other camera makes. I wonder if the precise nature of the filters affects the TC.I would be grateful if you could help me test this. I would be grateful if you could do five sets of ten images of eta Aur and zeta Aur. You should stack each set and analyse the stacked Green channel image.Can you let me know the five resulting estimates of the difference in V magnitude and what you understand to be the V magnitude of eta and zeta. I will compare your results with my own results on hopefully similar dates. The stars should be above 35 degrees altitude.Des Loughney

Hi Des,35 deg alt. currently means about 2 am at where I live, so this should be possible. However the weather forecast doesn't look too good for the next couple of days so it may take some time before I get a clear sky again. CSHeinz

Hello For some of you it might be interesting to read how Christian Buil does it with exoplanets: PHOTOMETRY WITH HIGH PRECISION DIGITAL CAMERAS Le cas de HD 189733 http://www.astrosurf.com/~buil/exoplanet/phot.htm The case of HD 189733 (english translation) http://translate.google.com/translate?js=y&hl=en&ie=UTF-8&u=http%3A%2F%2Fwww.astrosurf.com%2F%7Ebuil%2Fexoplanet%2Fphot.htm&sl=fr&tl=en EXTRASOLAR PLANET OBSERVATIONS BY RADIAL VELOCITY METHOD (HD 189733 b, HD 195019 b, 51 Peg b, tau Bootis b) http://www.astrosurf.com/buil/extrasolar/obs.htm With photometry just for the HD 189733 host star. But what now can be done by amateurs with the echelle sprectrographs is fascinating. Even some pro ob- servatories have not as good spectrographs (just larger scopes). Tentative detection of the extrasolar planet of the HD209458 system http://www.astrosurf.com/~buil/us/exop/209458.htm Here one can see an unusual method to handle much faintercomp stars. This is of course nothing anybody wouldexpect from the CS practicipants withtheir DSLRs. CS Wolfgang

The best time to do observing visual or photometry is when the star system is at the meridian. At this time of year it will not reach the meridian before sunrise. In October it should be on the meridian around midnight. For now, estimate how long your observing session is, determine astronomical twilight and back track the amount of observing time plus a few minutes. This will put the star system as high in the sky as possible. When calibrating your camera you want extinction to be minimal and ideally have it figured in. If you do not figure in your extinction you MUST make your measurements as high in the sky as possible to minimize extinction effect.You might even want to wait until the star are on the meridian and calibrate then. The observational data will not change, but your reduced data can be updated with the latest calibration.Have fun.JeffHopkins Phoenix ObservatoryPhoenix, Arizona USAphxjeff@hposoft.com

Heinz,I having been doing some work so that we can get an idea of the transformation coefficient (TC) of your camera. We can look at the difference in magnitudebetween eta Aur and zeta Aur. I did 50 images of these stars last night. I analysed their green channel images in five sets of ten. Eta is the comparison star at 3.18V and zeta aur the target. The (B-V) of eta is - 0.148 and the (B-V) of zeta is 1.154. The total (B-V) difference is 1.302. My TC is 0.15. The correction that needs to be applied to my green channel estimates is the (B-V) difference times the TC or 1.302 times 0.15 which equals 0.195 magnitude.The average of the five green channel estimates from the average stacked image ( five sets of ten ) I found to be 0.744. This is what is called the average instrumental differential magnitude. To get the differential V magnitude you subtract the correction of 0.195 from 0.744 to get 0.549. This is then added to 3.18 ( the magnitude of eta Aur ) to give a magnitude of 3.729V for zeta. The standard error for this estimate was 0.005. This was the value at 1.00 UT on 5/9/09.If you are able to do the same thing ie get five green channel estimates of the instrumental differential magnitude and let me know the average we should get a good idea of your TC.Des

Hi Des,No luck here so far, just rain and clouds and haze... maybe tonight...CUHeinz

Finally, I had a chance to get some decent observation time...even tho the moon was so bright you could almost read the papers in the middle of the night, and there was some haze, the values came out to be quite stable. Here are the raw instrumental magnitudes (V is eps Aur, C1 is eta Aur and C2 is zeta Aur) from the AIP4Win log. (I shot 5 series of 10 frames each, with dark frames for calibration) Series 1: Comp Star_X Star_Y ADU_Max Aper_PV Aper_Px Ann_ADU Ann_Px Sky_ADU Star_ADU Mag ±Sigma Var 1816,116 1125,939 4901,1 713267, 610 101651,8 273 372,351 486132,8 -12,272 0,0013 C1 2095,899 1591,051 5185,6 691709, 611 107991,3 272 397,027 449125 -12,186 0,0014 C2 2219,428 1497,735 2822,37 481493, 619 106310,4 271 392,289 238665,6 -11,499 0,0025 Series 2: Comp Star_X Star_Y ADU_Max Aper_PV Aper_Px Ann_ADU Ann_Px Sky_ADU Star_ADU Mag ±Sigma Var 2068,671 894,321 4648,67 664492, 616 86306,9 272 317,305 469032,5 -12,233 0,0013 C1 2362,405 1350,635 5073,78 655760, 616 93478,5 275 339,922 446368,2 -12,179 0,0014 C2 2483,068 1253,569 2811,98 438692, 617 89951,4 272 330,704 234648,1 -11,481 0,0025 Series 3: Comp Star_X Star_Y ADU_Max Aper_PV Aper_Px Ann_ADU Ann_Px Sky_ADU Star_ADU Mag ±Sigma Var 2020,651 767,591 5835,87 790076, 618 108389,2 275 394,143 546495,6 -12,399 0,0011 C1 2318,702 1221,340 6787,11 796406, 618 107689,6 272 395,918 551728,9 -12,409 0,0011 C2 2438,498 1123,171 3589,45 533237, 618 106033,7 272 389,83 292321,8 -11,719 0,0021 Series 4: Comp Star_X Star_Y ADU_Max Aper_PV Aper_Px Ann_ADU Ann_Px Sky_ADU Star_ADU Mag ±Sigma Var 2338,423 943,760 6391,24 666751, 451 54640,1 188 290,639 535673,4 -12,377 0,0010 C1 2646,854 1389,961 6809,03 655379, 449 57119,6 188 303,828 518960,3 -12,342 0,0010 C2 2764,171 1288,990 3560,05 407335, 449 55382,9 188 294,59 275063,7 -11,653 0,0019 Series 5: Var 2484,805 738,352 6461,54 692495, 452 49821,7 184 270,77 570107,3 -12,445 0,0010 C1 2803,022 1178,228 7157,04 681823, 450 53796,2 189 284,636 553736,6 -12,413 0,0010 C2 2918,188 1074,661 3597,21 415417, 454 50374,6 185 272,295 291795,2 -11,717 0,0018 Ok, so let's write M = Mr + eps *(B-V) + C where Mr is the raw instrument magnitude (B-V) is the color index and M is the (published) magnitude value then if we use the magnitude values for eta and zeta Aur, we can solve a system of 2 linear equations in two variables (eps and C) to get an idea of the correction factor eps and the offset C Problem: what is the magnitude (outside of eclipse) of zeta Aur in the first place? Different sources will give different values, the 3.729 that you got seems to be on the bright end, tho. E.g. Wikipedia gives 3.77, on AAVSO charts it's 3.8, but on "stellarium" it's even 3.65 (!). I assumed 3.77 and for the 5 series of frames, I got these values of eps (or TC) for my Olympus E-420 -0.0745 -0.0829 -0.0768 -0.0760 -0.0814 Now, choosing the eps and C constants so that eta Aur and zeta Aur magnitudes are a perfect fit, the magnitude values of eps Aur would come out at : 3.04 3.07 3.14 3.09 3.09 with an average of 3.09 and an SD of 0.036 I guess one should get better results under more favorable conditions (less moon light and haze) CU Heinz

Oh great, first the forum software removes all line breaks from my post, and then it won't let me edit it for some spam filter trigger ....next try:Finally, I had a chance to get some decent observation time...even tho the moon was so bright you could almost read the papers in the middle of the night, and there was some haze, the values came out to be quite stable. Here are the raw instrumental magnitudes (V is eps Aur, C1 is eta Aur and C2 is zeta Aur) from the AIP4Win log. (I shot 5 series of 10 frames each, with dark frames for calibration) Series 1: Comp Star_X Star_Y ADU_Max Aper_PV Aper_Px Ann_ADU Ann_Px Sky_ADU Star_ADU Mag ±Sigma Var 1816,116 1125,939 4901,1 713267, 610 101651,8 273 372,351 486132,8 -12,272 0,0013 C1 2095,899 1591,051 5185,6 691709, 611 107991,3 272 397,027 449125 -12,186 0,0014 C2 2219,428 1497,735 2822,37 481493, 619 106310,4 271 392,289 238665,6 -11,499 0,0025 Series 2: Comp Star_X Star_Y ADU_Max Aper_PV Aper_Px Ann_ADU Ann_Px Sky_ADU Star_ADU Mag ±Sigma Var 2068,671 894,321 4648,67 664492, 616 86306,9 272 317,305 469032,5 -12,233 0,0013 C1 2362,405 1350,635 5073,78 655760, 616 93478,5 275 339,922 446368,2 -12,179 0,0014 C2 2483,068 1253,569 2811,98 438692, 617 89951,4 272 330,704 234648,1 -11,481 0,0025 Series 3: Comp Star_X Star_Y ADU_Max Aper_PV Aper_Px Ann_ADU Ann_Px Sky_ADU Star_ADU Mag ±Sigma Var 2020,651 767,591 5835,87 790076, 618 108389,2 275 394,143 546495,6 -12,399 0,0011 C1 2318,702 1221,340 6787,11 796406, 618 107689,6 272 395,918 551728,9 -12,409 0,0011 C2 2438,498 1123,171 3589,45 533237, 618 106033,7 272 389,83 292321,8 -11,719 0,0021 Series 4: Comp Star_X Star_Y ADU_Max Aper_PV Aper_Px Ann_ADU Ann_Px Sky_ADU Star_ADU Mag ±Sigma Var 2338,423 943,760 6391,24 666751, 451 54640,1 188 290,639 535673,4 -12,377 0,0010 C1 2646,854 1389,961 6809,03 655379, 449 57119,6 188 303,828 518960,3 -12,342 0,0010 C2 2764,171 1288,990 3560,05 407335, 449 55382,9 188 294,59 275063,7 -11,653 0,0019 Series 5: Var 2484,805 738,352 6461,54 692495, 452 49821,7 184 270,77 570107,3 -12,445 0,0010 C1 2803,022 1178,228 7157,04 681823, 450 53796,2 189 284,636 553736,6 -12,413 0,0010 C2 2918,188 1074,661 3597,21 415417, 454 50374,6 185 272,295 291795,2 -11,717 0,0018 Ok, so let's write M = Mr + eps *(B-V) + C where Mr is the raw instrument magnitude (B-V) is the color index and M is the (published) magnitude value then if we use the magnitude values for eta and zeta Aur, we can solve a system of 2 linear equations in two variables (eps and C) to get an idea of the correction factor eps and the offset C Problem: what is the magnitude (outside of eclipse) of zeta Aur in the first place? Different sources will give different values, the 3.729 that you got seems to be on the bright end, tho. E.g. Wikipedia gives 3.77, on AAVSO charts it's 3.8, but on "stellarium" it's even 3.65 (!). I assumed 3.77 and for the 5 series of frames, I got these values of eps (or TC) for my Olympus E-420 -0.0745 -0.0829 -0.0768 -0.0760 -0.0814 Now, choosing the eps and C constants so that eta Aur and zeta Aur magnitudes are a perfect fit, the magnitude values of eps Aur would come out at : 3.04 3.07 3.14 3.09 3.09 with an average of 3.09 and an SD of 0.036 I guess one should get better results under more favorable conditions (less moon light and haze) CU Heinz

Heinz,Thanks for the work which is interesting. It establishes that there is a different transformation coefficient for an Olympus E-420 compared with the Canon 350D/450D. It would be good if anyone out there can provide an explanation for the difference.There are a number of different estimates of the out of eclipse magnitude of zeta Aur. Earlier in the year a number of us did a study of the March eclipse of zeta. We looked at the system before and after the eclipse. Whatever the published magnitudes of zeta we found the actual magnitude, in 2008/2009, to be around 3.73V. I would be interested in the outcome of your calculations assuming that 3.73V was the correct magnitude.I was able to make an estimate of epsilon last night. It was:10th September 2009- JD 5084.51- 3.096V - Standard Error 0.004. This is obviously consistent with your estimate.Best Wishes,Des

Hi Des,That's interesting. If I assume mag 3.73V for zeta Aur and solve the equaltions to get the color correction factor etc, I get brighter magnitudes for eps Aur:3.023.053.123.073.07and color correction factors close to -0.11 instead of -0.08 ... so it does make quite a difference what actual V magnitude is assumed for zeta. I guess I need to go out again and find a nice field of stars near the zenith where there are stars of about the same magnitude but different color to get more data points. But even with the revised values, the Olympus and Canon correction factors are quite different.I would feel more confident with this whole color correction if I could deal with the raw green filter values from the Bayer array, but somehow the AIP4win driver for the Olympus raw format won't let me extract the raw bayer matrix, so I have to use this de-bayerization, white-balancing and who knows what happens there. It would definitly be interesting to hear from other DSLR users what kind of corrections they have to make. At least it seems to give consistent results fro my DSLR, just the calibration needs some fine tuning.CSHeinz

Exciting .... on Sept 19th I got about 3.2 .. 3.24 for eps Aur when assuming 3.73 for zeta. It was about time :-) But ... if a make a measurement with RGB-G filter, and correct for color index to what it woulkd be for V filter, how should I report ? - Report the uncorrected value and specify RGB-G - Report the corrected value and specify V band (ebven tho I really measured with RGB-G) - Report the corrected value and specify RGB-G, but add a comment that value is corrected and how it is corrected? Hmm..... CS Heinz

Heinz,I made epsilon to be 3.180 on the 19th September. Jeff made 3.22 last night. As I understand he AAVSO specifications if you send in an estimate derived from the green channel image and it is not transformed ( using your transformation coefficient ) then it is labelled as a TG estimate ( see < http://www.citizensky.org/content/photometry>). If you are satisfied that using your transformation coefficient ( and a careful methodology ) you are now producing good V estimates then you can describe your estimate as a V estimate. The only way to be reasonably certain that you are producing good V estimates is by comparing your results with people who have been getting good V estimates for a while. For a number of reasons, including the fact that few comparison stars are completely invariable, I figure I am getting consistent V estimates when my results are regularly within + or - 0.02 magnitude of other reliable observers. I think that it is very difficult to refine estimates ( including V estimates ) without being part of an observing team - which is what Citizen Sky is all about.Des

As I have written in another post I have also done som testing using my Canon 450D for photometry. I have used rougley the same methods thatis described here but also counted for air mass. The formula then is something like this: M = Mr + eps * (B-V) + k' * AM + CI use 20 different stars round epsilon for solving eps and k' using regression analysin in Excel. My hope was to get consistend values for the coefficients so I don't have to use all 20 stars. But so far they have wobbeld a bit so I have stucked with doing all stars and compute eps and k' every time.I don't know if there is some fundamental difference in using RGB instead of only G, other than you got a different value on eps? I have tried both methods and think I got the best result using RGB. When I use RGB I bin 2x2 pixels with 1 R, 1 B and 2 G toghter to one new pixel which get the combined value of the 4 pixels. My coefficients for eps-G is about -0.055 and -0.075 and for eps-RGB about -0.085 to -0.105. k' is about -0.15 to -0.35I have used magnitud values from the Tych2 catalog for my 20 compairsion stars. Tycho use Vt and Bt magnitudes but they could be transformed to V and B with: V = Vt - 0.09 * (Bt-Vt) and B-V = 0.85 * (Bt -Vt). Tycho gives the following values for some stars:Eta: V=3.138 B-V=-0.141Zeta: V=3.767 B-V=1.146Lambda: V=4.694 B-V=0.592Rho: 5.198 B-V=-0.125In my final estimate of epsilon I use V=4.705 for lambda though, as I suppose this is the official AAVSO value.

A couple of tips. When determining the extinction coefficient k' use stars with color index close to zero (B-V =0.00) through varying air mass from near the horizon to near the zenith. When determining epsilon (V band color transformation coefficient) use stars with color index varying from -1.xx to +1.xx near the zenith only. Jeff Counting Photons Hopkins Phoenix Observatory phxjeff@hposoft.com

HiWhich (B-V) do you use for eps Aur ?+0.54 mag B-V ?CS Wolfgang

Hi Wolfgang, The (B-V) of epsilon Aurigae varies. I have measure it from around +0.50 to +0.59 with most (B-V) > +0.55 Jeff Counting Photons Hopkins Phoenix Observatory phxjeff@hposoft.com