In my last post on observing, I introduced the recent upgrades to the quadrant. Unfortunately, the initial tests weren’t good. There was a significant error in the azimuth averaging $~2º$ at $180º$ azimuth. I’ve since determined the reason behind this and implemented what I hoped was a correction. And this recent observing session tested it.
Normally this is where I’d post a table of the data in the blog. But this brings me to another improvement, this time in data processing. Typically, results would get entered into a custom Java program I’d written just for this project. It would take all the individual observations, store the information in a database and do a bit of light statistics on it. In particular, for each star, it would take all observations for that object, average them, and determine the standard deviation. However, one of the features I’d wanted to build that never got done was to compare the calculated RA/Dec to the true positions.
Since that program was written1, I’ve significantly improved my Excel skills and it occurred to me that, instead of trying to write a complicated data storage and interface tool, everything could be managed quite easily in an Excel workbook. Or even better, a Google Sheet which would be accessible anywhere.
As such, the program that took almost a month to write has been entirely replaced with this Google Sheet2 in just a few days of work. This sheet also features a tab where the true RA/Dec of fixed objects can be input which feeds back to the main page for comparison3 allowing error to be calculated much more easily!
A set of pivot tables aggregates the data. I currently have one for date and one for the various objects, wherein both calculate the average RA4 and Dec as well as the standard deviation5.
So how was this most recent observing session?
The average error in RA was +0.33º with a standard deviation of 0.24º. In other words, I was pretty consistently aiming a third of a degree to the left. This is considerably lower than most other nights.
The average error in Dec was an astonishing -0.02 with a standard deviation of 0.26º. This ties for the second lowest average error in Dec.
Rarely have there been a few nights of comparable observing. Jan 4, 2019 was certainly better overall. However, this was the night that we switched, part way through the night, from using the observing program, in which I had a list of stars and when they crossed the meridian, to the simpler method of just capturing any star that we could see as it crossed the meridian, and then figuring out what it was later. Because much of the night’s data was with the former, which explicitly stated the time the object was on the meridian (which is equal to the RA), the data was artificially enhanced since it essentially eliminated this variable.
The only other one after that change, that was comparable, was Dec 21, 2019. This one still allowed for the determination of when an object was on the meridian as a free variable, but did still eliminated much of the uncertainty since the axis was fixed. No worry about systematic error with the scale.
Thus, to demonstrate that we can get impressively accurate data after adding the complication of another scale is a huge leap forward. These quadrant enhancements are also another huge leap in the amount of data that can be collected. No longer having to rely on the availability of assistants means any clear night can be used.
The azimuth scale offers two even more important factors. First, I’m no longer constrained to the meridian. Once I got into a groove, I was able to take a measurement every 2-3 minutes. In many cases previously, I was only getting 6-10 an hour unless it was a particularly dense portion of sky.
In addition, the quadrant has had a problem that objects within ~20º of the zenith were essentially impossible to target because it would require putting the observer’s head through the box for the plumb. But now, those same objects can be picked up when they’re lower in the sky.
Overall, I’m absolutely thrilled by the new possibilities this has raised. I’d been considering building some new instruments entirely, but the energy this has given me to continue with this one for some time will likely quash that for a good while.
- Which was over 2 years ago at this point.
- It appears that visitors to this sheet will see whatever filters I’ve applied. I’ll try to remember to leave it unfiltered when not using it, but I may forget from time to time. There’s also a few hidden columns that are intermediate steps for calculating things like the hour angle, or bringing in the latitude of the observation location.
- VLOOKUPs are your friend!
- In degrees.
- Another one of the fields I’ve included is the observer. Unfortunately, I didn’t keep very good records for much of my old data. So apologies to Padraig who doesn’t get much of the credit he should. I plan on trying to go back through the posts and tease out which nights he observed.