[lister6520]

Is there any way to 'manually' calibrate the compass of the O-GPS1? The auto calibration seems to work fine when out in the open but in the presence of metal structures it doesn't quite get it right with the result that stars still have trails, even if much shorter than they would without the O-GPS1. I wish it were possible to point it at Polaris and press some button to tell it precisely where North is as that would solve the problem, provided of course that the North star is visible.

For me this would make a big difference as the way things are I cannot get a reliable North when shooting from home - problem being the steel mesh in the roof which alters the magnetic field by several degrees.

This is a stack of 8 40 second exposures. Light pollution is quite bad so I had to darken the sky considerably in PP.

From the blur of the TV antenna one can clearly see the O-GPS1 doing its thing but not well enough as can be seen from the trails of the stars. It is not it's fault as in open areas it works fine, but I'm quite sure manual calibration would easily solve this problem where there is a 'polluted' magnetic field.

[Stone G.]

Well, in principle you could do as the compass designers for (iron)ships had to do: Counterbalance the deviation caused by the vessel by means of soft iron balls and rod magnets around the magnetic compass. But I have my doubts you could make it work in practice.

Establishing the direction to true geograpical north at your (fixed) observing site shouldn't be much of a problem using some characteristic landmark and Google Maps.

[lister6520]

What you say about the compensating balls and magnets should work but if I'm not mistaken getting all the components in the right place to calibrate the compass used to be a highly skilled job which I don't think I would be capable of doing.

Actually establishing the true North is the easy part - I just need to centre the North star in the viewfinder. The problem is how do I tell the O-GPS1 that it is now facing directly North rather than the 8 degrees or so West of North that it thinks it is. That is something which should be very easy to provide in the camera firmware but it can of course only be done by Pentax.

[AstroDave]

I just took a quick look at the O-GPS1 review and the comments about the astrotracer mode.

I'm not quite sure what it actually does, but I am a professional astronomer, so maybe we can figure out what you need to do.

Did you do the calibration? The unit needs to know where it is, the direction the camera is pointing, and the focal length of your lens (that sets the scale that is must move the sensor to take care of earth rotation).

I would hope that you can just tell it what direction you are pointing, which should be fairly easy to establish. As has already been discussed, letting the compass determine north can be tricky. Conversely, if you are not being too badly affected by the steel in your building, the unit may be able to figure out from its GPS-determined location what is called the local deviation (i.e. the difference between true north and magnetic north). Perhaps your unit is trying to compensate twice: you told it a correction, and it is figuring out its own correction?

My biggest question (to the O-GPS1 manual?) would be how does it know the true orientation of the camera with respect to the sky. Just saying you are pointing e.g. "east" is not enough. You need to know the camera tilt (forward/backward as well as sideways), as well. I think the K-5 can get at least some of this info from its internal tilt meter, but I'm not sure how far over that it can measure. In general, the field of view will rotate during an exposure, unless you have you camera already on an equatorial mount. Unless sensor motion can take out a rotation as well as a steady drift (up/down or left/right) you won't get rid of the star trails.

Does the extent of the trailing depend on where you are pointing your camera? i.e. try pointing pretty much due south (and maybe tilted up towards the celestial equator) - that would be the easiest motion to compensate.

[Stone G.]

I would hope that you can just tell it what direction you are pointing, which should be fairly easy to establish. As has already been discussed, letting the compass determine north can be tricky. Conversely, if you are not being too badly affected by the steel in your building, the unit may be able to figure out from its GPS-determined location what is called the local deviation

This is the very core of the OP's question:

1) The lens focal length is read from the lens or manually input in the case of old lenses;
2)The inclinometers and accelerometers in the camera body tells the system what the tilt of the system and sensor is;
3) The direction to true geographical north is determined by the reading of the magnetic compass in the O-GPS1 and the geograpichal position as determined by the GPS in the O-GPS1 and then some knowledge about the magnetic declination at the actual position. The O-GPS1 litterature does not tell us from where that information comes - it may be in tabular form in the firmware or come via the GPS sattellites???

But the system cannot handle highly localized magnetic deviations and the O-GPS1 Manual explicitly tells us that. And I agree with the OP: It would indeed be convenient if there were means in the firmware to manually input corrections for such localized deviations.

Actually establishing the true North is the easy part - I just need to centre the North star in the viewfinder. The problem is how do I tell the O-GPS1 that it is now facing directly North rather than the 8 degrees or so West of North that it thinks it is. That is something which should be very easy to provide in the camera firmware but it can of course only be done by Pentax.

Actually, you don't need visibility of the North Star. If you have a piece of freeware such as Cartes du Ciel/Sky Chart you can use any star, because the software will give you the azimuth in real time and you can then make up the difference between the true azimuth and the compass reading from the O-GPS1. One must of course enter the proper geographical position to the software but that we have with great accuracy from the O-GPS1 itself.

[lister6520]

@Stone G.
we're in sync, you understood me 100%.

Regarding the magnetic declination, GPS units usually have this built in to them as they are capable of telling you direction of motion both in terms of magnetic north as well as true North. I guess it must be a table stored inside the GPS, possibly at relatively low resolution and then using interpolation. Even a very old Garmin GPS II+ I have does this so the table is necessarily quite compact as back then memory was very limited in such devices.

On Android devices the GPS module also provides the magnetic dip (I'm not sure if that's the correct name), that is the inclination of the field line to the vertical, as well as the exact field strength that should be present at any location on Earth. This is used mostly to give a warning when the magnetic field is distorted as any distortion is likely to cause variations not only in the horizontal direction but also in its vertical direction as well as the magnitude of the field. So if any of these doesn't match what the tables say it should be at that particular geographic location then a warning is issued as it would then be reasonable to assume that even the horizontal direction is inaccurate.

This latter check however does not seem to be implemented in the O-GPS1 as once it is calibrated I never get a warning even if I move it very close to large metal objects which cause the compass indication to go completely wrong.

While writing this post it just occurred to me that I can probably use the smartphone to look for the 'cleanest' spot on the roof and locate the camera right there. Now I just need to wait for a clear night to see whether that will help.

[Stone G.]

Your/our problem with local magnetic deviations and the O-GPS1 compass' sensitivity to that gave me a silly idea - much like the ship compass I showed above:

Why not make a sliding device with slightly magnetic objects to the left and right of the O-GPS1? I dont know how well it will work in practise, but here's some kind of proof-of-concept:

I simply placed my camera+O-GPS1 on my desk, pointing due south (180 degrees as the electronic compass would have it) and then approached the O-GPS1 with a screwdriver from the right and from the left:



Moving from afar to touching the hindmost right and left sides with the screwdriver, I got some interesting results:


Click on image to see small video on my homepage.

As can be seen in the video, I could actually enforce from 0 to about +10 degrees deviations at varying distances from the right side and about from 0 to -20 degrees from the left side.

Hmmm......Will it work in practice or not? I don't know, but I do think i will pursue this possibility further!