A couple of years back, while on holiday I took a few photos of the night sky with my DSLR camera and the standard lenses. Doing this gives some appreciation of how the solar system moves in relation to our galaxy (the Milky Way). The most noticable effect is the streaking of the stars due to the rotation of the Earth. Since the camera and tripod are standing on the ground - they are fixed in relation to the earth and thus the camera's field of view will sweep very slowly across the sky.
Here's the clearest shot I managed to take on that trip:
![]()
You can see if you look closely that each star is elongated because of the combination of a 5 minute exposure and the distance that the Earth rotates in those 5 minutes.
I wanted to try to get a streak-free photo during this year's holiday (assuming the clouds stay at bay!). So I need to find a way to cancel out this phenomena. There are at least two options: 1) stop the Earth's rotation, or 2) rotate the camera in the opposite direction to compensate.
Option 1 presents some practical difficulties and is likely to have some environmental complications attached to it.
Option 2 seems ideal though.
Using a microcontroller (ATMEGA328P) and a stepper motor (plus driver board) I plan to build a motorised panning attachment for my tripod that will slowly rotate the camera in the opposite direction to the Earth's rotation and at the same speed as the Earth rotates.
How long is a day? Turns out that in this context (making the stars stand still from an Earth viewpoint) a day is not quite 24 hours long. We're used to our Earth day being governed by the illumination of the sun Sol.
When the Earth does a 360 degree revolution in relation to the stars, it has not yet done a full 360 degree revolution in relation to the sun because the Earth has move around its orbit by one 365th of a full orbit (approx). According to Wikipedia (http://en.wikipedia.org/wiki/Rotation_period or http://en.wikipedia.org/wiki/Day) the stellar Earth day is about 4 minutes shorter than the Solar day.
If you want to prove this yourself - pick a bright star and line something up with it accurately. Then come back *exactly* 24 hours later and see if it's still aligned. The star will be a tiny bit shifted off to one side (depending on which hemisphere you live in).
In the northern hemisphere an observer facing south will see the stars above moving from left to right as the Earth turns. My device will need to drive the camera from left to right in order to remain fixated on the same point in the night sky.
I plan to use the stepper motor to drive a reduction gearbox (something like 500:1) to drive the axle on which the camera will be mounted. So there are come calculations to do.
1 stellar day is 86164.098903691 seconds long on average.
So the Camera needs to do 1 full 360 degree turn in that time. The stepper motor I'm using has a 200 steps per revolution which equates to 1.8 degrees per step. However the driver I'll be using (Schmalzhaus EasyDriver v4.4) has the ability to microstep at 8 times better resolution than that, so effectively it's 1600 steps per revolution on the stepper shaft.
I have still to ascertain the actual ratio for my gearbox (it is not stated explicitly on the product). I'll have to open it up and count the teeth on the gear sequences. Once that is known I'll multiply this ratio by 1600 to get the number of ticks required to make the camera swing through a full 360 degrees. If this is a 500:1 ratio then it'll be 800,000 steps per camera revolution!!! That will be more than enough resolution for really nice smooth camera motion. A large ratio like that also helps reduce the strain on the stepper motor if a heavy camera or a long lens is used.
Ultimately I need to have the ATMEGA328P fire a step pulse every (86164.098903691 / 800000) == 0.107705 seconds. If I use a decent 1MHz oscillator as the microcontroller's clock source I can send a step pulse every 107705 clock cycles.
This design should be pretty easy from an electronics perspective, but it will require some metalworking to fabricate some aluminium parts to form the axle supports and camera attachment brackets.
I hope to do a full build log with parts list and schematics and mechanical drawings once I've finished it.
Here's the clearest shot I managed to take on that trip:
You can see if you look closely that each star is elongated because of the combination of a 5 minute exposure and the distance that the Earth rotates in those 5 minutes.
I wanted to try to get a streak-free photo during this year's holiday (assuming the clouds stay at bay!). So I need to find a way to cancel out this phenomena. There are at least two options: 1) stop the Earth's rotation, or 2) rotate the camera in the opposite direction to compensate.
Option 1 presents some practical difficulties and is likely to have some environmental complications attached to it.
Option 2 seems ideal though.
Using a microcontroller (ATMEGA328P) and a stepper motor (plus driver board) I plan to build a motorised panning attachment for my tripod that will slowly rotate the camera in the opposite direction to the Earth's rotation and at the same speed as the Earth rotates.
How long is a day? Turns out that in this context (making the stars stand still from an Earth viewpoint) a day is not quite 24 hours long. We're used to our Earth day being governed by the illumination of the sun Sol.
When the Earth does a 360 degree revolution in relation to the stars, it has not yet done a full 360 degree revolution in relation to the sun because the Earth has move around its orbit by one 365th of a full orbit (approx). According to Wikipedia (http://en.wikipedia.org/wiki/Rotation_period or http://en.wikipedia.org/wiki/Day) the stellar Earth day is about 4 minutes shorter than the Solar day.
If you want to prove this yourself - pick a bright star and line something up with it accurately. Then come back *exactly* 24 hours later and see if it's still aligned. The star will be a tiny bit shifted off to one side (depending on which hemisphere you live in).
In the northern hemisphere an observer facing south will see the stars above moving from left to right as the Earth turns. My device will need to drive the camera from left to right in order to remain fixated on the same point in the night sky.
I plan to use the stepper motor to drive a reduction gearbox (something like 500:1) to drive the axle on which the camera will be mounted. So there are come calculations to do.
1 stellar day is 86164.098903691 seconds long on average.
So the Camera needs to do 1 full 360 degree turn in that time. The stepper motor I'm using has a 200 steps per revolution which equates to 1.8 degrees per step. However the driver I'll be using (Schmalzhaus EasyDriver v4.4) has the ability to microstep at 8 times better resolution than that, so effectively it's 1600 steps per revolution on the stepper shaft.
I have still to ascertain the actual ratio for my gearbox (it is not stated explicitly on the product). I'll have to open it up and count the teeth on the gear sequences. Once that is known I'll multiply this ratio by 1600 to get the number of ticks required to make the camera swing through a full 360 degrees. If this is a 500:1 ratio then it'll be 800,000 steps per camera revolution!!! That will be more than enough resolution for really nice smooth camera motion. A large ratio like that also helps reduce the strain on the stepper motor if a heavy camera or a long lens is used.
Ultimately I need to have the ATMEGA328P fire a step pulse every (86164.098903691 / 800000) == 0.107705 seconds. If I use a decent 1MHz oscillator as the microcontroller's clock source I can send a step pulse every 107705 clock cycles.
This design should be pretty easy from an electronics perspective, but it will require some metalworking to fabricate some aluminium parts to form the axle supports and camera attachment brackets.
I hope to do a full build log with parts list and schematics and mechanical drawings once I've finished it.