Why filter wheels

My automated scope far from my office and home may measure 80 targets a night in 3 filters each. Not sure how you do that without a filter wheel.

Many filter wheels are plain nuisance when doing photometry. In the past, most of the ones designed for amateur astronomers had awful positioning accuracy. I measured accuracy of +-1 mm for 5 position filter wheel from SBIG (sold for ST-6) and +-0.2 mm for Optec IFW. Now compare that with typical pixel sizes of sensors...

So I find your question to be completely justified.

However, most often one needs colour information (i.e. observations in more than one filter) to properly transform measurement results to some standard system. Yes, there are occasions when this is not (very) critical, and then one can just fix a filter to the camera (and preferably to the telescope as well ;-) ).

There another solution how to avoid filter wheels, too. Namely - a multi-channel instrument. Such devices were built already at the times of photographic plates (e.g. Tartu Observatory, Estonia, had two identical astrocameras with Pg and Pv filters that imaged the sky simultaneously).

Similar concepts have been used for photoelectric photometry, too. For example:
Four-Channel Photometer at the Danish 1.54-m (https://www.eso.org/public/teles-instr/lasilla/danish154/four-channel-photometer/). IMHO it is still operational.

Or in the era of CCD-s:
"BUSCA: A Telescope Instrumentation for Simultaneous Imaging in 4 Optical Bands" (https://arxiv.org/abs/astro-ph/9903177)

Another question is their mechanical complexity and .. well.. .cost. :-) For BUSCA - 4 large-format back-illuminated CCDs...

Actually, having two-channel simultaneous CCD/CMOS photometer can be quite feasible for amateurs, too. An instrument selector with suitable (fixed?) dichroic filter (that is passing more or less all the light in some wavelength range while reflecting in other(s), they can be bought from e.g.. Edmund Optics or Thorlabs etc) and two CCD/CMOS cameras with attached photometric filters. E.g. those relatively large format ZWO ones that AAVSO is evaluating for BSM right now. When light loss of ~2x can be tolerated, simple semi-transparent mirror can be used as well.

Tõnis,
wishing the best remaining day of 2018!

It is relatively easy to build a 2-channel system, such as that by Smith and Genet:

http://adsabs.harvard.edu/abs/2006SASS...25...87S

As Tõnis mentioned, the major cost in dual-channel systems used to be the cameras, but now you can purchase inexpensive digital cameras.  For APASS, we used twin OTAs on a single mount, another method.  However, multi-channel systems are more expensive and more complex than just putting a filter wheel in front of your normal digital camera.  I'm also evaluating the use of one-shot-color cameras for scientific photometry, and should have some results later this spring.

For many projects, a single filter is adequate, and by using a single filter screwed into the camera nosepiece, you eliminate complexity and cost.  Many people start out using an unfiltered camera, progress to using a single filter, and end up with a filter wheel and many filters.  With each step, you have more projects available for which you can contribute useful information.

Arne

My sense is: one-shot-color cameras may have a place in multi-band photometry but they will be quite limited by their fixed relative sensitivity between bands. Whereas with a filter wheel I can control relative star and band exposure times at will, precisely, and over a very wide range. Indeed I have to doubt that one-shot cameras will find use in multiband photometry of red stars: there can be no one exposure time to measure multiple colors with good S/N but without saturation or other nonlinearity. If you have to accommodate with multiple exposures, you lose the advantage over filter wheels. Hopefully one-shots will find use in sun-temperature eclipsers etc, and maybe for bright stars where little time is lost by the need for multiple (short) exposures.

Separately and for the record: Many modern filter wheels, including mine from SBIG, register within 1 CCD pixel, all night long and for years without maintenance. To warn against filter wheels because someone once used a bad one is like warning against aspirin because Tylenol once hurt somebody decades ago. If one's photometry doesn't need a filter wheel, excellent, simpler is better. But if you do, modern filter wheels are for sure more reliable than most observatories' electrical or internet services.

Arne, thank you for nice reference!

Eric, you're absolutely right about filter wheels, I was (deliberately!) exagerrating. We have currently that Optec IFW in use, and for mmag level work, I can observe just in one filter per night (or two, when morning twilight flats can be taken), because that ~0.2 mm is already slightly too much shift. At the same time, other photometer has Apogee AFW50-9 positions with sub-pixel accuracy (13.5 um pixels) because of mechanical stops. Our former SBIG and still-in-use Optec ones have Hall sensor based position detection, that is not very accurate. I suspect that many modern low-end wheels may have the same design. IMHO when ever possible, for high quality work a filter wheel with mechanical positioning should be selected.

Additional problem, that may arise from imprecise positioning of filters is variable vignetting and variable amount of scattered light as function of (x,y) on sensor. That second issue is very (I mean VERY) inconvenient to handle.

Happy new year and many clear nights to everyone!

Tõnis