View Single Post
  #10  
Old 07-03-2018, 09:03 PM
Wavytone
Registered User

Wavytone is offline
 
Join Date: Jul 2008
Location: Killara, Sydney
Posts: 4,147
By all means mess with it as a simple DIY experiment, but its really only suitable for wide-angle shots with short lenses, and its the sort of thing I'd make with my son to learn from, in his teens.

Simple geometry - the linear error at the focal plane equals the linear error of whatever pushes the barn door, multiplied by the ratio of the lens focal length divided by the length of the barn door. Lets say I'm using a 15mm lens, and the barn door is 15cm long, so the ratio is 0.1. Also lets suppose the camera sensor pixel spacing is 3 microns (typical for APSC or m4/3).

Also suppose we would like to keep errors in the image below, say 3 pixels for nice star images without trails or blobs. 3 pixels is about 10 microns.

This means the tracking mechanism must be accurate to 10 x 10 = 100 microns, or 0.1mm. Not impossible for a homemade mechanism made by someone with a lot of skill, but highly improbable relying on a bent metal strip guesstimated by eyeball. To that add whatever errors are in the screw thread that drives it.

Now suppose you want to put a 300mm telephoto lens on. The ratio of lens/barn door is now 2, twenty times greater than previous. It means that for the same tolerance in the image of 3 pixels, the accuracy of the tracking mechanism must be 20 times better, 5 microns. This is way beyond what can be done without a machine shop and for that kind of precision you'd find it easier to buy an equatorial mount.

There's an electronic issue, too, which will be familiar to anyone who has dabbled in building a DIY drive. If you are relying on a simple resistor-capacitor network and CMOS oscillator (eg a 7555) to govern the frequency of the signal to the motor, you will find this is not particularly stable and random variations occur - bad enough to be visible rather like "periodic errors" in gears except that its not periodic, and can't be predicted - and really annoying.

The only real solution that really works is to use timing regulated by a quartz crystal clock, wth the frequency divided down to whatever you need one way or another (a microcontroller is a good way). Which is what is in most telescope control handsets.

Last edited by Wavytone; 07-03-2018 at 09:40 PM.
Reply With Quote