High Accuracy Magnetometer?

I am looking for a high accuracy magnetometer to measure the altitude and azimuth of my telescope the problem is that I cant find anything more accurate than the HMC6352 and it only has an accuracy of .5°and to give you an idea the moon is only .5667° (34 arc minutes) so that really wont suite my purpose, there may be one that is better (MicroMag 3-Axis Magnetometer) but the datasheet It doesn't supply the degree heading resolution only that it has a "resolution as low as 0.015uT". So does anyone know of a magnetometer that is accurate to 1-0.5 arc minutes (0.00833°) and cost less than $50?

So does anyone know of a magnetometer that...

If your goal is to improve directional accuracy / resolution, you have one set of answers;

If your goal is to improve directional accuracy / resolution for a category of highly inaccurate sensors, you have another set of answers.

I am looking for a high accuracy magnetometer to measure the altitude and azimuth of my telescope

I don't even have a vision of how a magnetometer could be used to measure the position of a telescope mount through it's alt and azi travel limits. I've only seen high resolution quadrature encoders used for such applications. Tell me how your sensor choice and setup would work assuming you could find the resolution needed.

Lefty

My initial thought of using a magnetometer was poor one, the best way to calculate the altitude and azimuth is to do a calibration off a known star and then use it as a reference point then all I have to do is calculate the deg from that point to any other stellar body and if I use a servo motor rather than a stepper I wont have to recalibrate through the night.

If you are moving the telescope any relative references will change . For the Due, Ive just combined the MP6050 6-axis code with the micromag3 code and some LSM303 to mix pure aa.x with M.x for a tilt corrected compass, with pitch roll angles. If you araldited the sensors to the telescope body this would give a frame of reference that was absolute. You would need to calibrate for the orientation difference between the sensors axis and the telescope but once done you should be able to run with it.
Ive just done a cut and paste job but really the micromag takes so long it needs to be on DMA which the ARM can do on the Micromags SPI.
It can also do I2C DMA, so the accel and gyro data could be run at 1KHz easily, with the compass putting in a good word every so often . The new Oculus headset runs its tracker at 1000Hz but i bet its not doing all 9 at that speed. there is no need.
If you can look at an exact place in space, when exactly will you look there. 65ns/"

bananas sorry i didnt answer your origional question, on accuracy of the micromag, it is measured by a count which can take 10's of ms . but for your application which is static you can take a very long average so i would expect to exceed the listed accuracy ! This is of course true for every sensor.