2 March 2019 at 5:29 pm #574268
Not entirely sure whether I should post this, but what the hell.
A couple of nights ago I tried for my very first exoplanetary transit. The sky, the electronics and my incompetence meant it was an utter failure, though perhaps a comedy of errors might be more accurate.
Two targets were plausible. The observatory was opened up well in advance of the first ingress but the mount played silly buggers and wouldn’t track.It wasn’t cajoled back into action until 10 minutes after the prediction. On to plan(et) B. Plenty of time, 50 minutes, and all the calibration images, selection of autoguider star, and so on went swimmingly. The first few dozen 15-second images looked fine so I left the system to its own devices for a while. On returning I discovered that I’d forgotten to tell the dome to track the scope, which was now looking at 5% sky and 95% dome. Shortly after fixing that, the computer crashed.
As expected, the data is junk. Even the good images had a poor SNR because of the extremely dusty sky we’ve been having here for the last couple of weeks. The predicted depth of the transit was 0.9%, or 9 millimags. A precision of 1 millimag requires a SNR of at least 1000.
I truly hope that I’ve not deterred anyone else from having a go. I intend to try again when there’s a good chance of a clear sky lasting for at least three or four hours.2 March 2019 at 5:49 pm #580779Lars LindhardParticipant
Good that you write these things.
We all make mistakes, often silly ones like your dome experience. My logbook has plenty of observations with errors like yours – it is easy to look away at the right – or rather wrong – moment.
I think we all learn from postings like yours, and this helps us to remember the basic things, check the equipment during the day, start up in good time, make things running and keep an eye on them etc.
Better luck next time.3 March 2019 at 12:18 am #580781Eric WatkinsParticipant
I had similar problems when I first started on exo-planets. Initially I took longer in getting underway than I had anticipated and hence did not have enough egress lead in time, sometimes I would even start imaging after the egress point. This past week the auto guider has decided not too auto correct tracking for some reason, even though I had changed nothing from the working sessions before. Irrespective of any number of calibration sequences I now make it goes of on it’s drunken man’s walk about.
Incidentally, which exo are you trying.
Eric8 March 2019 at 11:29 pm #580822
“Incidentally, which exo are you trying.” Perhaps that is best kept firmly swept under the carpet for the moment.
Now trying again with WASP-65b. Predicted ingress is 23:52 with a depth of 0.0138 magnitudes. After much careful calibration 50 second exposures produce a SNR of~900 on an unfiltered CCD and imaging started at 23:00. My guess is that should give adequate precision (around 1.1mmag all being well) at a useful cadence. OK, it’s not in a standard photometric band but it’s close enough to GAIA-G and will do for a first attempt. I’ll be happy if anything decent looking shows up in the light curve.
Seeing is appalling right now, with stellar images at 8-15 arcsec FWHM. Strong north-easterlies blowing over the Canaries and as they pass over the caldera and cumbre a great deal of turbulence ensues down here at 760m altitude. The lucky guys at El Roque are above all that at 2600m. Perhaps it will improve later during the night but knowing my luck, it won’t.9 March 2019 at 3:22 pm #580829Robin LeadbeaterParticipant
Since you have plenty of photons such that stochastic precision is not a problem it could be worth considering a choice of filter to minimise the systematic errors which are likely to dominate at the mmag level. Bruce Gary’s on line book has some suggestions
Robin9 March 2019 at 5:10 pm #580830
Thanks. Initial measurements suggest that the actual precision is around 3.5mmag. Seeing was so bad that the aperture is 12 pixels in radius, and at 1.4arcsec/pixel that corresponds to a disk half a minute across! FWHM is markedly smaller, of course, but still far too big. Perhaps observing in Sloan-r might be an improvement.
It’s even windier today and I’ve serious doubts whether it’s worth even opening the dome. Incidentally, the gusts last night were strong enough to make the scope flap around beyond the ability of the autoguider to track correctly when observing at low altitude late in the session. I lost around 5% of the data to that cause.10 March 2019 at 6:10 pm #580840
This is the raw light curve. Purple shows the measure fluxes and their error bars. Green lines connect the points. A period of missing data and the existence of two (high measured flux) outliers is readily apparent.
The only processing performed is raw photometry with APT and a Perl script to normalize the photon flux to a comparison star. Absolutely no smoothing, outlier-removal or any other kind of data munging has been performed. The x-axis is HJD-2458551. The predicted ingress was for 2019-03-08T23:52 and egress at 2019-03-09T02:36. In JD these are 2458551 + (0.4944, 0.6108) respectively.
To me it appears that the ingress was picked up nicely. At about mid-eclipse I found that the dome wasn’t tracking properly so the data there is either noisy or missing. Towards the end of the eclipse the target had sank to an altitude of <35 degrees and further adjustment of the dome was neede (the slit exposes either the top 60 degrees or the bottom 60 degrees). In addition, the high winds made the scope flap around so much that the autoguider couldn’t cope and several more images were lost. The eye of faith suggests egress was detected but that may just be wishful thinking.
Detailed analysis will take place in due course.
- You must be logged in to reply to this topic.