Stellar Time Keeping

© Brooke Clarke 2006 - 2007

Astronomical "seeing" limits what can be done.  The effect of seeing is similar to the star's position changing.  Seeing ranges from 0.5 arc seconds for an excellent location to 2 arc seconds for most locations to 5 arc seconds for poor city locations.

Star movement at zero degrees declination is 15 deg per hour or 1 arc second in 66 time milli seconds.  At other declinations the movement is slower.

So the best possible result is a timing accuracy of 33 milli seconds to around 100 ms.

The following table shows the measured variation in the Earth's rotation showing the daily change (dEpsilon) in ms (0.001") of time.
To have the same resolution as the ERSI table you need 100us time resolution but the best seeing is 330 to1,000 worse that that.

Earth Rotation Service Bulletin B lists UT1R-UTC for each day. (2007)
Date      MJD     x     y         UT1R-UTC UT1R-TAI dPsi dEpsilon
2007              "     "             s        s  0.001" 0.001"
(0h UTC)
JUL 8     54289 0.21518 0.39632 -0.161163 -33.161163 -62.2 -6.7
JUL 13    54294 0.21947 0.38319 -0.162176 -33.162176 -62.6 -7.1
JUL 18    54299 0.22589 0.37049 -0.162647 -33.162647 -62.9 -6.6
JUL 23    54304 0.22971 0.35703 -0.162754 -33.162754 -64.7 -6.4
JUL 28    54309 0.22925 0.34583 -0.162462 -33.162462 -65.3 -6.6
AUG 2     54314 0.22763 0.33130 -0.163459 -33.163459 -65.8 -6.2
AUG 7     54319 0.22785 0.31574 -0.164930 -33.164930 -66.7 -6.4
AUG 12    54324 0.22488 0.30185 -0.166586 -33.166586 -66.5 -6.6
AUG 17    54329 0.22113 0.28718 -0.168466 -33.168466 -67.4 -6.0
AUG 22    54334 0.21598 0.27252 -0.170614 -33.170614 -67.8 -6.4
AUG 27    54339 0.21213 0.25902 -0.173026 -33.173026 -67.8 -6.4
SEP 1     54344 0.20485 0.24686 -0.175802 -33.175802 -68.8 -6.3
SEP 6     54349 0.19825 0.23259 -0.178934 -33.178934 -68.4 -6.4
SEP 11    54354 0.19037 0.21835 -0.182414 -33.182414 -67.4 -5.7
SEP 16    54359 0.18113 0.20495 -0.186284 -33.186284 -68.4 -6.0
SEP 21    54364 0.17210 0.19350 -0.190507 -33.190507 -67.7 -6.5
SEP 26    54369 0.15845 0.18230 -0.195088 -33.195088 -67.7 -5.8

The change day to day is a few tesths of a ms so this method may work as a way of measuring seeing, but will not measure the earth's rotational period.

11 Apr 2009 - To get around the "seeing' degradation of knowing exactly when a star crosses the local meridian some form of averaging might be used.  Some possible ways:

• Use a number of photodiodes all in a line along the star image path and determine the time when each was crossed by fitting a curve to a plot of brightenss vs time and use all the crossing times to arrive at the meridian crossing time.  The answer would be known not long after the last photo diode was crossed.
• Use a star tracking telescope and record where it points prior to, during and after a meridian crossing as a time log.  Fit a line to the plot and solve for the crossing time.
• Use a number of telescopes, all seperated from each other by more than one foot so that the "seeing" is independent at each scope.  This would allow measuring the "seeing" for each pair of scopes so with three or more scopes the "seeing" of each would be known.  Then using the seeing as an inverse weighting combine the meridian crossing times of each to get a more accurate time.
• Use a lateral photo diode or X-Y CCD to plot the X-Y position of the star as it crosses the meridian.  This would not involve any moving parts so is probably more accurate than moving scope methods.  The star image needs to be less than 1/1000 of the resolution so that there are many points so a normal TV grade CCD is not good enough.

While looking into why the GPS satellites are in the orbits that they are I discovered a couple of the reasons are:
1) to bound the upper and lower limits of the Doppler shift.  Doppler shift was the basis of the Transit system and maintaining Doppler shift for GPS was a prime consideration.  Geosynchronous satellites would have near zero Doppler and so were not considered.
2) The GPS orbital period of 12 sidereal hours means that the ground tracks of each satellite is the same day to day.  If a satellite wanders more than a degree off track it's steered back on track.  This I hadn't known before.  Now there's a strong analogy between GPS satellites and stars.  That's to say if you were to look at a vertical line on you South meridian you would see the same GPS satellite cross at the same elevation angle every 12.0 sidereal hours.  If you knew when each satellite crossed the meridian you could measure the Earth's rotation period.

The time when any GPS satellite crosses your local meridian can be calculated from the ephemeris data for that satellite.   The ephemeris data that's broadcast from each satellite is only so good so if the broadcast data is used then there will be some error.  But the ground based reference stations monitor all the satellites and have very accurate ephemeris data a day or two later used for centimeter (2012 millimeter?) level land surveying.

Above is described how a 1 arc second angluar error in a star's position causes a 11 milisecond time error a similar analysis can be done for GPS positions.
For example a 1 cm error on the surface of the Earth with a radius of 6,378.1370 km at the equator a distance of 1 cm corresponds to an angle of 0.000090 deg or 0.32 arc seconds a little better than the 1 arc second "seeing".  So sub mm position accuracy is needed to get the Earth's period error below a millisecond.  This probably can be done using GPS signals and maybe using optical telescopes.  The image of a GPS sat may be much more stable than a star image because you could acually image the satellite.

New Idea 3 Plate Solving (Dec 2016)

Time Nuts mailing list messages Opening Question:
out of curiosity, are there any amateur/semi-pro experiments that can measure the length of the solar or sidereal day to sub-millisecond resolution?
To reproduce data like this: https://upload.wikimedia.org/wikipedia/commons/5/5b/Deviation_of_day_length_from_SI_day.svg

Something in the sky that goes "ping" every day - detected with a pointing accuracy of < 1ms/24h or <0.01 arc-seconds (!?). Or perhaps two satellite-dishes pointed at the sun and noise-correlation/interferometry??

Anders
-----------------

 from Chris A.

out of curiosity, are there any amateur/semi-pro experiments that can
measure the length of the solar or sidereal day to sub-millisecond
resolution?

Yes.  It is not hard at all to measure the Earth's rotational period, if all you needs is "sub millisecond"  It would get harder if you
cared about nanoseconds.

I worked on an amateur project with some others and while measuring the Earth was not the goal we had to know the Earth's rational period
to do the work.   The project was about stellar photometry.  But I leave that part out.....

Basically what we did was mount a camera made out of a small CCD sensor and a 135mm f/2.8 camera lens salvaged from  an old 35mm film camera.  The camera was fixed to the roof of my garage. (This was THE big cost saving feature:  The camera could not move.  The mount as fixed at one location in the sky forever, right at the equator)  I placed it in one end of a long wood crate and it looked up at the equator through a square hole on the upper end of the box.   The box provided some protection from the elements and provided a lot of light
shielding.

To measure Earth's rotation all you need to do in know exactly when you took an image and to have a GOOD catalog of star locations.  Let's say your image captures 200 stars.   They are rather blurry and each covers maybe 5 pixels but even so you compute the centroid of each "gaussian blob" and then do a least squares fit of all those centroids to the astrometric catalog.  The catalog is "good" to several milliacrseconds and with hundreds of centroids you can figure out were the camera was printed to a few  "MAS" (Milli Arc Seconds).  We
took many images every clear night for several years.    Hardware cost today is "not much" and you can use salvaged camera equipment  Almost all of the software is available for free. Certainly matching stares to catalog images is.  Yes the lens has geometric distortion and the CCD is likely not exactly 90 degrees to the optical axis but the software models this.  This is possible because millions of star
positions are known to insane levels of accuracy and if they appear in the "wrong" place in your image you can bet the cause is geometric distortion in your camera, especially after seeing the same error in hundreds of images.  We used narrow filters to limit the image to just one "color" so the chromatic aberration in the optics i not an issue, filters are cheap.

As part of our processing we time-tagged each image and also recored where the optical xis was pointed at.

So you'd need a small telescope or big camera lens and a camera that can be triggered by a computer and software.  Not really expensive.  I'd invest in the best used optics you can and get a monochrome camera.

Some people in the past century used transit telescopes to manually measure the time a star crossed a hairline in an eyepiece.  Then the next night to observe the same star again.  Now you know the length of the day (after you reduce the data)   Put you can measure a dozen stars every night and take an average.   In concept it is very simple.    But today we can measure a tens of thousands of stars per day from a suburban roof top.


Almost all other methods of measuring the Earth's rotation do not collect enough data.  You need tens or hundreds of thousands of data points. if you want to know the sidereal period to Time Nut standards.
Chris
----------------------

I think that you refer on prjects like Astrometry plate solving. I think one should got a reference to get a time reference instead of scope "pointing" reference, so, once one's got local coordinates in encoder positions, for example the values of the north pole with an alt/az mounting, can use a sub/arcsec plate solver to obtain good sidereal timing reference. using two encoders helps much.

The problem can be visibility of the reference points, however.
Best Regards,

Ilia
---------
Yes,  . . . basically correct.  But you save a ton of time and get better results if you simply bolt the telescope down to the Earth so that it can't move at all.   The aim point just needs to be "close" and then later you determine where it is aimed.    If you are only measuring period you don't need a surveyed location.   If measuring absolutely time you do.

Using a fixed mount is what makes this affordable by amateurs.  Epoxy the camera to a fixed masonry building.  This removes an unknown and dramatically simplifies the processing and also saves most of the cost witch is always the mechanical stuff.  One package of JB Weld epoxy replaces thousands of dollars of motors and encoders and precision gears.

With a fixed mount camera you have two kinds of "tine", that observed by the camera and a second from your GPSDO.  If they diverge then you deduce that it must be the Earth's rotation that changed.   But maybe you wonder of maybe the camera moved or some effect you forgot to remove.  So it is but to have some buddies running the same setup in different cities around the world and check that you all see the same results.   That is what we did.  It is FAR EASIER to do this kind of replication when the setup is very inexpensive.

Today you could build a camera for a LOT less then we did.  I'm thinking of a surplus used lens from a 35mm film camera.  A 250mm lens or so and a 3D printed plastic part that holds this to a cheap point and shoot camera.   We used epoxy to held the lens to the camera,  it meant you'd never be abler to take it apart again but it was going on a roof top, rain and all.

I'd like to measure the Earth's rotation using some type of optical system that uses either the Sun or stars.   In the book Splitting the Second by Tony Jones it mentiones that the Photographic Zenith Tube was replaced by the Danjon Astrolab.  It works at night by allowing a human observer to watch as stars cross a line of equal altitude.  It's very similar to the Dent Meridian Instrument opticallly.  Also has not just one time of concidence like the Dent but a number of them so that one event can be measured a number of times, allowing for averaging.  By using some fixed evevation angle you can measure many more stars than you can working straight up.

The good news about the Sun is that it's bright and so easy to detect.  The bad news is that the Earth's axis of rotation is not at right angles to the plane of the Earth's orbit around the Sun.  Hence the elevation of the Sun is always changing and over a year's time this amounts to a change of about 47 degrees, so a fixed telescope can not be used.  But stars always show up at the same elevation for each meridian crossing allowing for a fixed scope.

Years ago I looked into doing this using a rotating mirror turning on a shaft that was from a hard drive spindle, since at the time that was the lowest run out bearing you could get.  But I now think that looking at stars that are within say 5 degrees of the zenith could be done using a fixed scope and a mask with either radial or circular slits that were about 1 star image wide.  It may be possible to see the brighter stars during the daytime if the scope objective diameter is large enough and maybe with a blue cut filter.

8 Aug 2007 - Although seeing stars in the daytime is a bonus, just using a fixed scope pointing up to measure night time meridian crossings of stars would be all that's needed.  It may be that a low light level security CCTV camera is adaquate.  I'm looking at the PC164 now.

If mounted to a concrete pier looking straight up (as determined by a plumb line) even if there was some misalignment, if the scope stayed pointed in the same place the day to day timing would be accurate.

Note that navigation systems and timing systems are closely related.  A navigation system can be used where the location is the known and time is the unknown.

This is consistent with the performance of the MD1 and newer astro compasses.

There are some U.S. patents on instruments that do this.

Eyeball Daytime Star Viewing

I think daytime star viewing is not something someone is likely to stumble on.  The three examples above probably have a common theme.  In order to see a star in the daytime with your eye the right combination of factors needs to be present. 

Light Gathering

One has to do with light gathering.  The larger the area used to gather the light the more sensitive is the result.  For example at dusk or dawn when it's too dark to see with your eyes alone you can see quite well using binoculars that have an exit pupil diameter of about 7 mm, i.e. matched to your dark adapted eye.  The exit pupil is the objective diameter divided by the power, so for example a 7x50 binocular has an exit pupil diameter of 50 mm / 7x = 7.1 mm.

If someone is in bright sunlight their iris closes down a a few mm diameter so they are not in a good spot to look for weak lights.  But if in a forest with deep shade, inside a building with a small opening to the sky or underground in a deep mine your pupil is more likely to be dilated.

If for example a telescope or binoculars were fitted with a blue block filter that did make a number of stars brighter than the background  you still could not see then if you were outdoors on a bright sunny day.  An extreme example would be binoculars fitted with a pair of Hoya O-58 that cuts slightly more than half the spectrum completely.  The sky may appear black to an eye that's out in bright sunlight.  The only way for this to work is from a dark daytime location or by using something like a gas mask to block the daylight and allow your eyes to dark adapt.

Magnification

Magnification has the effect of making a star (point source of light) appear brighter while making the background (diffuse or extended source of light) appear dimmer.  The problem is that at high magnification the field of view gets narrower and holding steady is harder as is pointing to where a star is located.

Example:  I've been bugging my neighbor Paul about daytime star watching, and he recently took a photo of  comet McNaught taken at noon.  He said it was difficult to focus because of the brightness.

The literature on star trackers makes it clear that it is possible to track a bright (Navigation) star in the daytime.

Many of these systems use a reticule (episcotister) or chopper in front of a Photo Multiplier Tube (PMT), or in newer instruments in front of a silicon diode.  The reticule can be used in two ways, in one it's just a mask to gate the star light and in the other it's a chopper to convert the light into pulses that can be AC amplifier and synchronously detected.  When it's just a mask the slit width is a little wider than needed for just a star.  This does two things, one: it makes a plot of brightness vs. time have a flat top like a pulse and two: if a planet passes through the slit it makes a different shaped curve with a pointed top since the slit width is narrower than the angular diameter of most things like the moon or planets.

If the area of the light sensitive element is much larger than a star image, as is typical of all these systems, then there will be some background noise caused by the area that is not receiving the starlight.  This results in a lower signal to noise ratio.

Optical Tube Assembly (OTA)

Stable Focus

For unattended operation the focus needs to be stable.  I doubt a amateur telescope can ever hold focus over the normal day to night temperature range encountered here in Northern California, let alone somewhere else where the temperature has much wider fluctuations.  I'm convinced that it's not only possible but not that difficult to design the tube assembly so that the focus is independent of temperature to a small part of the deepth of focus.

No mount is needed to the OTA can be mounted on a pier pointing near straight up.  This would be away from the house and other man made areas, i.e. as far from both the house, driveway and road as possible while still having some narrow view of the sky.

Focal Length

Focus Visible and Near Infrared

Silicon sensors "see" longer wavelengths than eyes.  It's very difficult to design a lens that can focus the different wavelengths of visible light at the same place and nearly impossible to do it for visible and near IR light.  The answer is to use an all reflecting optical system.  Mirrors do not have this problem.

Note that most amateur telescopes, like the SCT or "modified" RC use a glass corrector plate that acts as a lens and so are not all reflecting designs.

Primary Diameter and f-ratio

I've recently learned that the daytime sky is like an extended object in that the lower the f-ratio the more of the daytime sky background gets recorded in the camera.  But star images are point sources and the f-ratio does not effect their exposure.  So high f-ratio optical systems should be better for daytime star watching.  The other effect is that the f-ratio is almost equal to the star image size in microns.  So the f-ratio should not be too different from the pixel size of the camera.  Larger diameter primary optics gather light proportional to their unobstructed area, so larger is better, but also costs more.

Newtonian telescope (or other all reflecting design so it works well at near IR) mounted to concrete pillar.  Carbon fiber tube to help stabilize the focal length.
Mirror coatings may need to be customized for near IR reflection.
Scope City -  Newtonian OTAs - 6"-f6, 10"-f5, 12.5"-f5, 16"-f5, 6"-f8, 8"-f6 all fiberglass tubes
Vixen R200SS 8"-f4 Newtonian OTA - not info on tube material
Meade - truss-dobs - 8"-f6, 10"-f5, 12"-f5
Telescopes.com - Discovery series Truss Dobs -12.5"-f5, 15"-f4.2-f5, 17.5"-F, 24"-f5 - Truss or tube material not specified
JMI - 12.5"-f4.5
there are a number of carbon fiber truss dobs on the market.

5 April 2007 - Orion has the StarBlast 4.5" f4 dob scope for under $200.  It's getting good reviews mainly becuse of it good performance but for viewing and (in my opinion more important) imaging with silicon sensors.  This would make a great OTA for stellar timekeeping since it's the lowest f number I've seen on an off the shelf Newtonian scope.  The lower the F number the more curvature there is and the harder it is to figure the mirror.  The OTA uses a metal tube.  114 mm dia x 450 mm FL (FL is 17.7" so tube needs to be maybe 2 feet long to hold camera at prime focus)
14 April 2007 - Orion now is offering just a modified StarBlast OTA that good for use with CCD cameras.  The secondary mirror is larger and they've allowed for more infocus.  Sticker price $140 +s/h + rings

7 May 2007 - One test report says the plastic focusing housing is very flimsy, not suitable for some TV cameras.  Also if a TV camera is mounted at prime focus it will be out of the top end of the tube.  So if possible the best thing would be to get the 4.5"  f4 mirror and build it into a custom OTA with a longer tube.  I've asked Orion about getting the mirror. ans. no only scope.

28 May 2007 - Teleskop-Service has a nice f4 200mm Newtonian OTA for 500 euros. out of stock on the GEO 200 mm f4 (May '07)

31 May 2007 - OTA needs to keep imaging chip at the primary mirror focus.  The key specification may be the stability of the support structure.  Quartz rods or tubes may be an extreamly high stability material that could be used in a truss type structure.
GM Associates, Oakland, CA -
Quartz Scientific, Fairport Harbor, Ohio - Rods: 2 to 13 mm  dia, Tubes: about 150 sizes from 1x2 mm to 105x110 mm (4' lengths under 60 mm OD)
Newt Software - free to download runs on most flavors of WIndows
Newtonian Telescope Design Planner - IE and Netscape versions, for use with cameras provides % illumination vs. sensor size
Yazoo Mills - mailing tubes -
ProtoStar - Black Light Telescope Tubes with black flocking made from Phenolic-impregnated kraft paper
Hastings Pipe Co. - sells Aluminum tubing cut to order for telescope OTAs.

Depth of Focus

The Airy disk diameter is 2.44 x 0.00065mm x f so for the StarBlast f 4 mirror d = 2.6 um at the green peak of 659 nm. and 4 um at 1,000 nm near IR.  So the f # = Airy disk diameter in microns (um) when IR light is involved.

The Depth of Focus = 2 x f x d = 2 * 4 * 4 um = 32 um or 0.00125" or just over 1 mil.  The focusing method needs to be able to position the imaging chip in increments much smaller than 1 mil.  The change in tube length over temperature also needs to be will under 1 mil to avoid temperature defocusing.

Thermal Coefficient of Tube

Aluminum has abut 12.3 "/"/F*E-6.  The F.L. of the StarBlast mirror is 450 mm or about 17".  For a temperature range of +70 to +40 deg F or
 30 deg F change * 12.3E-6 * 17 = 0.0065 or 6.5 mils, way too much for a 1 mil depth of focus.

Thermal Compensated Tube

Pendulum clocks change rate with tempereature becuase the pendulum rod changes length.  But this can be conpensated by using two metals with different expansion coefficients such as steel and zinc.  John Harrison did this long ago.  If the two metals were just connected in a series straight line then there's no compensation.  An example of a compensating connection would be to have a steel tube longer than needed at the open end a smaller diameter zinc tube is connected at the top end to the steel tube.  The camera is mounted on the zinc tube down some distance from the end.  How far down depends on the relative coefficients of expansion.  If the coefficeients were the same the camera would need to be mounted inside the mirror.  The larger the difference in expansion coefficients the less length is needed to get the compensation.  Patent class 359 optical systems/820 Lens.with support..lens mounts...with temperature compensation.  359/820
1325936 Apparatus for Rendering the True or Apparent Focal Length of Objectives Independent of Changes of Temperature, 1911, 359/820 ; 219/121.6 compensates for both the lens mount and optical changes in the lens.

Coefficient of Theremal Expansion

Metal	in/in.oF x 10-6
Aluminum	12.3
Brass	10.4
Iron, forged	6.3
Phenolic-impregnated kraft paper	8
PVC thermoplastic	29
Delrin	6.8
Quartz, fused	0.33
Steel	7.3
Steel Stainless Austenitic (304)	9.6
Steel Stainless Ferritic (410)	5.5
Zinc	16.5

An aluminum tube 18" long would expand 18 * 12.3 *(bunch of stuff) and would be compensated by a length of PVC that was (18" * 12.3)/29 long or about 7.6" long.

Concern about Camera inside Tube because of heat waves This was brought up, but if a muffin fan is installed to blow outside air past the primary mirror and out the top of the tube as is the common practice for cooling off the mirror, then the heat gets blown away.  Seperating active voltage regulating circuits from the camera and placing them nearby, but outside the tube might also help.

Image size

Practical Calculations for the Newtonian Secondary Mirror -
CFF = 0.000433(f)³ and since the f# varies from maybe 3 to 12 the Coma Free Field will vary from 0.011 inches to 0.748"
I = arctan (ID / FL) for example a 10"-f6 scope will have a CFF of 0.094 (just under 1/10") and if the CCD chip will hold a 0.5 inch image (I.D.) then the angular field (I) will be ATN(0.5"/60") = 0.477454 degrees or 28.64 minutes.  But the Coma Free Field of view will be only ATN(0.094/60) or 0.08976 deg or 5.38 minutes of arc.

For the StarBlast f4 mirror CFF = 0.0277" or about 8% of the 1/3" CCD in a PC164 camera.
Working backwards to get a 1/3" CFF the f# needs to be about 9.1.
That may be a better match to the pixel size and would be better for daytime star viewing.  But makes for a longer tube, i.e. a 4" diameter mirror will focus at 36" or about twice as long as the f4 tube.  But still within the size range that's easy to ship or move.  Also flatter mirrors are easier to find/make.

Discovery Telescopes - makes 6" f8 mirror $189

Newtonian Telescope Design Planner - an on line calculator
A calculation of Airy disks for various telescopes by David Whysong

C-Mount Lens

The thread is 1.0" x 32tpi. 
The Cine mount expects the flange face to focal plane distance to be 17.5mm.  So a C-mount lens works on a C-mount camera.
If  C-mount lens is going to be used with a CS-mount camera an adapter ring 5 mm thick needs to be inserted.  This means that almost all new cameras use CS mounts so they can be used with either type of lens and are typically shipped with the adapter ring installed.

CS (C-Short) mounting is a newer standard.
The CS mount uses the same thread but the back lens flange to focal plane distance is 12.5mm.  A CS lens can only be used on a CS camera mount, not a C mount.

The CCD or CMOS imaging chips that are typically in CS mount TV cameras are typically 1/4, 1/3 or 1/2" nominal sizes.

My thought is that adding a filter that cuts blue light (a red filter) will increase the contrast of a star.  As the wavelength of the cut gets longer more blue light is cut but also there's less total light.  So there's a sweet spot where the contrast is the highest.  A shorter wavelength filter lets in more blue light lowering contrast.  Any longer wavelength filter cuts more star light lowering contrast.

Visible light is in the 400 to 700 nm range and silicon chips can see in the 200 to 1200 nm range.  So a silicon chip will have more contrast than an eye since it can see IR light from the star.  The quantum efficiency of a Silicon sensor is very high.  Much better than camera film or the eye.  Blue is around 400 nm and is the predominant sky color.

So the chart to the left may be misleading in that both the eye an silicon sensor have both been normalized to 100 %.  It would be more meaningful so show them in terms of star magnitude numbers.  But that's tricky because you need to add an optical system in front of the sensor to have equivalent systems.




Edmund Optics - Hoya R-72 - passes some red and IR, or Hoya RM-90 passes only IR, not visible red.  The IR series filters may not be the best choice to minimize the blue sky background and get the highest contrast on a silicon sensor.  The O-58 (Orange 580 nm) Sharp Cut Filter has less than 0.001% transmission at the 550 nm peak for human visible light yet passes 590 through 2,400 nm.  This may be a better choice.
Hoya Optics - only sells polished 50 x 50 x 2.5 mm or 165 x 165 x 2.5 mm or unpolished 165 x 165 x 4-5 mm filters.
The CM500 is the filter used in front of a silicon sensor to give it human eye color balance.  At the H-alpha wavelength of 656.281 nm this filter has a transmission factor of about 0.14.  A silicon sensor is about 7.4 times more sensitive than the eye at H-alpha.

The "Balloon-borne Large-Aperture Submillimeter Telescope” (BLAST)
"BLAST Autonomous Daytime Star Cameras"

SBIG - Red and Infrared Sensitivity - side by side comparisons of what a silicon sensor sees in the near IR
Radiance and Transmission Models - but called "IR clutter assessment", not visible light? - software for sky brightness & transmission
Spectral Sciences, Inc. -  computer code for atmosphere reflection and transmission -

Things that can be measured to learn about the sky background.

Background Brightness

Seeing

Long IR Temperature (Clouds)

Straight Up

Other Places to Point Scope

There are a number of things that might be used as the image sensor.

Single Photo Diode

CCD Camera

The type of TV camera used for star occultation timing, not the integrating type.
My PC164C web page.
Super Circuits - PC164C-EX - does it have built-in IR cut filter? Ans: No, it's IR capable as is
    Notes(0) on PC164C and 1004X CCD Low Light camera's - The PC164C is more sensitive than the PC164-EX
    The Ideal Occultation Telescope - mounts TV camera at prime focus eliminating the secondary mirror.
    Video Astronomy - First experiences with a PC164 camera - photos of a Gain Modified camera
The PC180XS uses the same Sony chip as the PC164C and has a light blocking area of 1 sq inch where the stock PC164C blocks about 1.5 sq in.  If the aluminum case is turned down to 1.25" diameter the area would be 1.22 sq inches.

A purpose made camera using either of the Sony chips would have a much smaller light blocking area.  The chip has an area of about 0.2 sq in.
This would be a long skinny PCB with the chip mounted at 90 deg on the end.  One way to accomplish that would be to use two PCBs where the spacing was such that the distance across both boards was 11.4 mm (0.449") then 8 of the chips leads could be soldered the top board and the other 8 leads to the bottom board.

For astronomical use the PC-164 is typically modified by adding a manual gain control.  This way the stock Automatic Gain Control will not try to make the image average to gray.  For better timing it would also be good if instead of a free running oscillator for the syncronization pulse timeing the camera ran from an external frequency standard.

Pixel size and plate scale calculator by Stan Moore

PC164 Major Chips:

Guidance of Sony semiconductor Datasheet - CCD Image Sensor(ICX) - CCD Camera System(CXA, CXD) - Video(CXA,CXD)
Sony CCD Image Chips -
ICX254AL - 1/3" EIA HAD EXview (replaces ICX054) 510 x 492 pixels 9.6 x 7.5 um (12 um pixel diag)
IC258AL - 1/3" - P164C-EX not as sensitive as plain version.
Sony CCD Camera System Chips -
CXD2463R - Timing Controller for CCD Camera
CXA1310AQ - Single Chip Processing for CCD Monochrome Camera
( CXD1267AN ? in CCD data sheet) - CCD Vertical Clock Driver

The ICX285AL still camera chip may be a better choice since it have more pixels (1392 x 1040) that are smaller (6.45x6.45um).  Note most Newtonian scopes have f numbers around 5 so this is a very good fit for pixel size.  BUT the near IR is not as good as the ICX254 and the readout timing is a little slower (15f/s instead of 30 f/s).

A custom hardware video processor could be made that would time stamp star crossings.  Probably the camera would be rotated so that stars moved either along or at right angles to the scan lines.

By using a sync separator IC even and odd fields can be identified as well as vertical and horizontal sync pulses.  At the end of each horizontal line a peak detector is read and zeroed.  This is no problem for a micro controller in terms of speed.  If both a peak and a valley detector looked at each line then (peak-valley) is how bright an object is.  If the camera is rotated so that stars move across (not along) scan lines  Then a single star can be tracked by scan line number (taking into account even and odd fields).  This method would provide a star meridian crossing time quantized by the time for a scan or about 62 us.  Note that for the most sensitivity a star image should be about the same as a pixel in size.

Finer time resolution can be had by rotating the camera so that a star moves along a scan line and looking at the time when it crosses the meridian (which would be at a fixed time from the start of the scan line.  This is much harder to do since it requires either video speed signal processing or a threshold to be triggered (this might work since there would be two times, the rising edge and the falling edge).  It would also be good to have peak and valley detectors to help in establishing the threshold.

Star Movement

International Occultation Timing Association (IOTA) - uses real time video cameras combined with date + time stamp on each field.
Drift-Scan Timing of Astroid Occulations - Scanalyzer to process image intensity
On the Beep... - genereates an audio beep for audio recording
KIWI Percision Timestamp Utility - PC program, not On Screen Display http://www.geocities.com/kiwi_36_nz/kiwi/kiwi.htm
VNG UC GPS Time Receiver -
KIWI-OSD, Video overlay of GPS precision timestamps - How to use the KIWI OSD video time inserter - US Sales - A detailed look at KIWI OSD video timestamps -
Horita - GPS Video Time Code products -
GPSPACE - GPS Positioning from Active Control System Clocks and Ephemerides - GPS post processing software - the PPP service is good to about 0.02 meters (under an inch) with a static 24 hour observation.  The idea is to get a good position fix from GPS.
Video_EXposure_Analyzer VEXA - uses microprocessor to blink LED as a tool to show the beginning and the end of the optical exposure within every single video field of a PAL or NTSC video camera
The IOTA Occultation Camera   (IOC) - Design and application of a fast computerized CCD camera system for recording of astronomical events - 2002 status?
Welcome at the website of Gerhard Dangl - Video_EXposure_Analyzer VEXA - turns on and off LEDs that are recorded in the video frame to see where in the video frame the shutter is open.  Also the exposure timing edges of a field may overlap into an adjacent field. - Measurements of exposure and internal delay on video cameras for use with Video Time Inserter -

A customized PC164 that has manual gain control and external sync input might be a good thing. 

The camera should not be mounted to a conventional rack type focusing mount but rather to a custom made "C" threaded mount so that the focus can be locked down.  Note that although it's easy to focus a scope on a star at night it's not possible to focus in the daytime with a scope on a mount that points it straight up.  (maybe a mirror to get a view of a distant land based object would work.)  The mount also needs to have a provision to allow the camera to be rotated and then locked without changing the focus.  Since the mount will be fixed it can be designed so the that front of the TV camera is just outside the tube ID.  This will allow minimizing the diameter of the secondary mirror.

Another, maybe better option, is to mount the camera on a Newtonian scope where the secondary mirror is normally mounted.  So there would only be one optical component, the parabolic mirror.  Some calculation needs to be done comparing the diameter of the camera to the diameter of an optimal secondary mirror for the primary mirror diameter and focal length.  It may be a good idea to repackage the camera to minimize it's area that's blocking the field of view, i.e. a long skinny camera would be much better than a short wide camera.  The small (1/3") video chip in the PC164C might lend it's self to a long skinny design.

There are also obstruction free Newtonian scopes whre the secondary mirror is off to the side.  It could be replaced with the camera using a new longer and larger diameter tube.

Celestial Navigation Data for Assumed Position and Time - his web page allows you to see where the navigation stars are.

List of 57 Navigation stars -

The Mag column gives  you an idea of how bright the star is (sun = -27, very limit of human eye +6). 
S.H.A. is the Siderial Hour Angle and is the UTC1 time when the star crosses the zero degree longitude line.
Dec is the stars elevation in the celestial reference system.  0 would be in the plane of the earth's equator, 90 near the north star.
So for stars that are straight up their dec will be 90 - <your lat) or in my case near N 50.809838 deg.
There are 23 stars marked with the asterick meaning they are Prominent in the Northern hemisphere.

Vega has a brightness magnitude of zero (quite bright) and a declination of +39 deg (about my lat).
USNO rise, meridian transit, set times for a list of objects.

Vega is not only bright but is almost overhead for me.

One of the things that MICA can do is Calculate\Configurations\Sky Map.  Just now (6/24/07 11 am) eps Per was very close to the zenith according to MICA.  One of the parameters is what magnitude stars to show.  I have that set for -30 (sun) to +3 (bright stars).  The planets also can be shown.  So the current map shows the sun, moon, Mercury, Venus & Saturn being up.  Right clicking the star at the zenith and selecting Object Info shows:
Name: eps Per
R.A. +03h 58.3m
Dec   +40º 01.9'
Azimuth   +15º 18.1'
Zenith Distance +00º 52.4'
Magnitude +02.89

A spread sheet of the 17 mag 3 or brighter stars that come close to the zenith and calculating their average declination shows -1.01 degrees.
Going back and calculating the spread of declinations centered on -1,01 degrees (i.e. instead of aiming the scope straight up it's tilted down a degree) shows +0.3 to -0.3 or a 0.6 degree field of view.

A PC164 at prime focus in the StarBalster has a 0.64 degree field of view.

Drift Scan is an astronomical imaging method where the camera is fixed and aligned in rotation so that a star moves along a horizontal row of pixels.  By choosing the camera clock rate to match the movement of the star field the light from the star gets integrated.  The exposure time depends on the focal length of the lens.

Information about TASS, The Amateur Sky Survey. - polar mounted CCD camera using 50 mm film camera lens makes exposures lasting a few minutes then stored on computer.
Stardial - fixed camera aimed at zero degrees declination with a field of view of 5 x 8 degrees.  Uses the drift scan idea, see antimated gif.  But since doing a full frame with a 50 mm F.L. lens would take a very long time, only a small number of rows are used.
Use Audine for acquire images -
Observing Lunar Occultations with CCD´s by drift-scan imaging  -
The Application of the TDI Method to Observations of Lunar Occultations -
CCD drift-scan imaging lunar occultations: a feasible approach for sub-meter class telescopes - paper.pdf
Drift Scan Imaging  -

The older ones used photo multiplier tubes.   I expect that more modern versions use silicon diodes to achieve higher quantum efficiency.  There seems to be a limit around 1 milli second for the time resolution you can achieve using a TV type CCD imaging chip.  The PEP can resolve time 1,000 times better, i.e. into the micro second area.

Typically a narrow bandpass optical filter is used for timing critical applications to get a faster response time since the light in band stop regions does not degrade the signal to noise of the desired light.

Optec  - Photometers - the SSP-5A photomultiplier aimed mainly at star brightness using a color wheel can resolve 1 milli second.

The Automatic Astro Compass was used in the B-52 bomber to provide celestial navigation by tracking a star.  Astro tracker patents are very similar to those on this page.  But note that the astro compass was for use in a plane and most of these patents are for use in a spinning spacecraft, although some are for use on the spinning earth.  The newer versions of star trackers were able to track stars 24 hours a day, i.e. they could see stars in the daytime.

Land surveyors have used star and sun sights since the invention of the telescope to find North and the Lattitude and with an accurate watch Longitude.  In some of the books it's mentioned that you can see some stars in the daytime.

Weems, Captain P. V. H., "TIMEKEEPING .", NAVIGATION, Journal of The Institute of Navigation, Vol. 3, No. 4, 1952-1953, pp. 117-120.
"The rotation of the earth, though now known to be variable, is the basis of time." So this was known at least a decade before the change in definition of the second.

Patents Containing: "Star Tracking System".  Some of these are for Earth based applications and some are fore space based apps.

2981843 STAR-TRACKING SYSTEM, 1947 - track star in bright sky background, chopping improves background rejection, IR pass filter rejecting blue sky background
2947872 STAR TRACKING SYSTEM, 1956 - piror art systems wasted 50 to 75% of light in shutters
3259751 STAR TRACKING SYSTEM - 1962- reticle quadrents 1 & 3 black gives PWM signal
3053984 STAR TRACKING SYSTEM - 1951- Day or Night star tracker - reducing FOV & sensor, red or IR lead sulphide, 50% light throughput, minimum sensor area
3194966 PHOTOSENSITIVE STAR TRACKING SYSTEM - 1961
3177366 PHOTOSENSITIVE STAR TRACKING SYSTEM - 1960
3165632 STAR-TRACKING SYSTEM USING A FREQUENCY MODULATED CARRIER WAVE - 1950 - nutating image FM when off boresight
6158694 Spacecraft inertial attitude and rate sensor control system -1998 - spinning scope to despin satellite
3194949 AUTOMATIC ONE-STAR TRACKING NAVIGATION DEVICE - 1965 - tracks Sun
6252627 Star tracker detector having a partial memory section - 1999 348/311; 348/314 uses CCD to detect stars, but only stores needed information, not full video field
3080484 ELECTROOPTICAL LIGHT-DETECTING APPARATUS - 1951 nutating, Day or Night Star Track tube techonlogy
3024699 LIGHT MODULATION SYSTEM - 1962 raster + offset shutter
3015457 Azimuth Control in a Guidance System - 1962 same system as 3027841
3181812 AIRCRAFT SEXTANT MOUNTING - 1965 same system as 3027841
3027841 GUIDANCE SYSTEM - 1962 very complex mechanics Fig 66 master control board probably a bomber nav system
2940171 ANGLE MEASUREMENT - 1960 - uses mag tape to form angle encoder
3048352 AUTOMATIC CELESTIAL NAVIGATION AND GUIDANCE SYSTEM - 1962 mechanical & tube
3215913 VARIABLE TIME-CONSTANT SERVO- MECHANISM SYSTEMS - 1962 servo bandwidth issues
3002097 DISPERSION SCANNER - 1961 - 4 telescopes
2949030 GYRASCOPICALLY STABILIZED OPTICAL SYSTEM PLATFORM - 1960 -
2941080 ASTROMETRICAL MEANS AND METHOD - 1960- detecting freq in two different bands - only works on bright stars
2966823 TRACKING TELESCOPE WITH DUAL FIELD OPTICAL SYSTEM - filed 1948 issued 1961 - dual magnification system
2923202 Dual Field Optical System - 1960 same system as 2966823
3006236 Apparatus for Astronomical Navigation, Michaud, Oct 31, 1961, 356/139.02 ; 356/139.05; 356/139.06; 356/147; 356/149 -
3739175 PHOTO SENSITIVE STAR SENSING ARRAY - 1973 uses two line sensors, but no info on pixel size
3381133 SCANNING DEVICE FOR TRACKER USING CONCENTRIC PHOTOSENSITIVE - 1968 Bulseye semi detector & nutating image - cancels out background gradient! center dot = 0.005" dia, I.D. of outer ring=0.009 and O.D. of ring= 0.020"  For the Stellar Timekeeping application no position a moving part is not desirable.  But using a second identical diode that sees dark would be good if there's any termperatrure effects that need  to be removed.
2958783 SCANNER - 1960 50% waste chopper
3244886 LIGHT MODULATION SYSTEM FOR PHOTOSENSITIVE TRACKING DEVICE - 1966
3244896 STAR TRACKER SCANNING SYSTEM USING A CIRCULAR SCANNING PATTERN AND A SQUARE APERTURE - 1966 no moving parts, very wide field of view probably night only operation
3241444 TORSIONAL LIGHT MODULATING MECHANISM - 1966 specal alloy, permanent magents & coil driven at reasonant freq of bar
3192824 SCANNING SYSTEM FOR LIGHT TRACKING DEVICE - 1965 uses dove prism to rotate image 90 thus able to scan 2 axix by rotation prisim 45 deg.  so uses 1/2 the mechanical parts needed for 2 axis scanning.
3251261 STELLAR ABERRASCOPE  1966 two back to back telescopes measure star aberration to determine spacecraft velocity
3443099 SIGNAL VERIFYING DEVICE BY- 1969 filter noise and false signals
3527951 LIGHT MODULATION SYSTEM - 1970 oscillating reed moves scanning slit
3527950 LIGHT MODULATION SYSTEM USING AN OSCILLATING REED SCANNER- 1970 oscillating reed moves scanning slit
3544221 QUARTZ MODULATED MIRROR SMALL ANGLE DETECTION DEVICE 1970 quartz rod nutates secondary mirror (TRW) Quartz rod is driven at reasonance, much better than motors.
2850939 ADJUSTMENT MEANS FOR OPTICAL ELEMENT - 1958 means to center scanning disk in star tracker
6012000 Simplified onboard attitude control based on star sensing - 2000 spacecraft position & orientation
3729260 INTERFEROMETRIC ROTATION SENSOR - 1973 a TV camera sees an interferance patters, for example a number of black and white bars which change as a point light source moves in it's field of view.  VERY COOL.
3827807 Star Scanner, Fletcher, Aug 6, 1974, 356/139.02 ; 250/206.2; 33/268; 356/147
5927653 Two-stage reusable earth-to-orbit aerospace vehicle and transport system
2946893 SCANNER FOR OPTICAL SYSTEMS- 1960 magnetically coupled not gears
3437814 SCANNER DRIVING PHOTOSENSOR WITH SIMPLE HARMONIC MOTION
5978716 Satellite imaging control system for non-repeatable error
3436635 PULSE WIDTH MODULATED SERVO DRIVE CONTROL SYSTEM
3398345 DUAL CHANNEL TRIGISTOR OUTPUT STAGE MOTOR SPEED AND REVERSING CONTROL SYSTEM
3401324 TIMING NETWORK FOR A MODULATED SERVO DRIVE CONTROL, SYSTEM
5207408 Stabilized air supported structure
3095541 AMPLITUDE RANGE AND AS FUNCTION OF TIME
3465229 METHOD OK CONTROLLING A DIRECT CURRENT MOTOR
3378745 RATE FEEDBACK LOOP NETWORK
3486100 PULSE WIDTH MODULATOR NETWORK
3465236 TIMING MEANS INCLUDING FIRST AND SECOND TIMING NETWORKS TO SELECTIVELY GATE TURN-ON DEVICES IN ...
3295010 IMAGE DISSECTOR WITH FIELD MESH NEAR PHOTOCATHODE
3191038 HORIZON SENSOR FOR SATELLITE ATTITUDE CONTROL
6275677 Method and apparatus for managing a constellation of satellites in low earth orbit
3436636 DIFFERENTIAL PREAMPLIFIER NETWORK FOR A SAMPLE-DATA MOTOR SPEED CONTROL

5012081 Strapdown stellar sensor and holographic lens therefor
3165632 STAR-TRACKING SYSTEM USING A FREQUENCY MODULATED CARRIER WAVE
3497696 STAR TRACKER ASSEMBLY USING MANGIN MIRROR AND FIELD SCANNING DEVICE
3381133 SCANNING DEVICE FOR TRACKER USING CONCENTRIC PHOTOSENSITIVE
3388629 DETECTION MEANS FOR STAR TRACKER
3439427 METHOD FOR NAVIGATING A SPACE VEHICLE
2981842 AUTOMATIC GRID SCANNING TRACKER
3088033 AUTOMATIC MULTIPLE GRID SCANNING TRACKER
3053984 STAR TRACKING SYSTEM
3080484 ELECTROOPTICAL LIGHT-DETECTIKG APPARATUS
3048352 AUTOMATIC CELESTIAL NAVIGATION AND GUIDANCE SYSTEM - ref:
3072794 ROTATING WHEEL SCANNER - Daytime Star Tracker: small spot size, small exit pupil.  It's important that the sensor area is about the size of the star image and not much larger, i.e. the illuminated area should approach 100%.  If the illuminated area is <1% then there will be a lot of noise.
3297877 CAPACITIVE COUPLING FOR A PHOTOCATHODE IN A POSITION INDICATING DEVICE - photo cathode with circuitry can read spot position like star tracker or missile seeker
2763177 TAYLOR SOLAR AND STELLAR TRACKER - can be configured for sun or stars using same prisims (mirror side or thrugh glass)

3551870 Photoconductive This Film Cell Responding to a Broad Spectral Range of Light Input, -3447234 realted pat, 3388629, 2765385,                      3013232,  3208022, 3284252 - taylored for poor blue response to imporve daytime star tracking

3447234 PHOTOCONDUCTIVE THIN FILM CELL RESPOND-ING TO A BROAD SPECTRAL RANGE OF LIGHT INPUT

3610936 Apparatus for Determining the Position of a Discrete Target Occuring within a field of view, Fried, 250/206 ; 250/214.1; 250/233; 250/237G; 250/237R; 356/147; 359/235

6060702 Low-cost light-weight star tracking telescope, May 9, 2000, 250/203.6; 359/399
5206499 Strapdown stellar sensor and holographic multiple field of view telescope therefor, Apr 27, 1993, 250/203.6; 250/216; 359/20; 359/399

3981588 Means and method for determining meridian location and azimuth  September 21, 1976 356/139.02 ; 250/206.3; 33/268; 356/139.06 search 356/141,152 250/23R 33/268

Calls:

3521071	Electro-Optical Aparatus for developing an effect representitave of the atttitude of the aparatus relative to that of a source of radiant energy (maybe a star tracker)	250/206 ; 250/203.3; 250/233; 356/139.02; 356/139.03	July 1970
3571567	Apparatus which Determines Lattitude and Longitude form the Deriuatives of two Coordinates of a Star	701/300 ; 250/203.5; 250/203.6; 33/268; 701/222	March 1971
3591260	Constant Time Response Scanner by CDC	359/235 ; 250/203.7; 356/139.02; 356/140; 356/148	July 1971
3713740	Astronomic Survey Apparatus and Method by CDC location within 100 feet and North within 10 arc seconds "Of course, once the position of the sensor is known, the invention can also be utilized to detect radiation from celestial sources having unknown positions and the position or orbital parameters of these sources can be calculated."	356/139.02 ; 250/203.6; 250/237R	January 1973
3717413	Sun Sensing System for a Flying Body (for a spinning satellite)	356/139.02 ; 244/1R; 244/168; 250/203.4; 33/264; 356/147	February 1973

Referenced by:

4840490	Laser position measurement and alignment
4710619	Apparatus for generating a signal providing information regarding a radiating source, especially an infrared source

3290933 Navigation Systems  73/178R ; 250/203.1; 250/237R; 33/268

Hoya - IR Filters - the RM90 or RM100 make the sky appear black to the eye, but pass IR, thus are DANGEROUS becasue if you look toward the Sun your eye will not see anything and the pupil will open but a large dose of IR gets into your eye and may blind  you.  These may be too extreme.

Wild, Heerbrugg, T4 - pier mounted surveying scope for astronomical observations. A pin prick on a paper tape recording Chronograph generated by an encoder on the T4 along with a second channel from a Chronometer can be used to accurately determine Longitude.

IEEE Xplore - Study of star image detecting technology in daytime strong background -

7349803 Daytime stellar imager, Mikhail Belenkii, Donald G. Bruns, Vincent A Rye, Timothy Brinkley, Trex Enterprises Corp, 2005-04-15 - uses H (1.6 microns) InGaAs) sensor w/PE cooling or K (2.2 microns) HgCdTe more costly cooling band IR  "..evaluated the star statistics in the visible I-band by using the Catalog of Positions for Infrared Stellar Sources and in the H-band and K-band by using the 2-Micron All Sky Survey catalog. Both of these catalogs are well known and are available on the Internet. "
7349804 Daytime stellar imager, Mikhail Belenkii, Donald G. Bruns, Vincent A Rye, Timothy Brinkley, Trex Enterprises Corp, 2005-04-15 -
7447591 Daytime stellar imager for attitude determination,Mikhail Belenkii, David Sandler, Donald Bruns, Eric Korevaar, Trex Enterprises Corp, 2005-09-26 - maybe up to 3 stars in 1 deg FOV.
20150042793A1 Celestial Compass with sky polarization, Mikhail Belenkii, Lawrence Sverdrup, Vladimir Kolinko, Trex Enterprises Corp, 2013-08-12 - Marine Corps contract M67854-12-C-6501.   Telecentric fisheye lens see all of sky and horizontal telescope makes sighting.  Uses known position of Sun or Moon to figure out bearing.