My Bard Door Tracker
I decided to build a “Barn Door Tracker” so that i could have a portable mount for taking astrophotos with a wide angle lens. Something that i could setup quickly and was compact and light.
But the main purpose was a “make fun” project that involved some mechanical design and an opportunity to experiment with a hobby oriented micro-controller – the Adruino.
This design uses a straight rod driven by a stepper motor. The motor is controlled by an EasyDriver stepper controller and an Adruino Nano micro-controller. The micro-controller manages the speed of the stepper so it can vary the speed of the motor as the door opens, compensating for the geometry of the straight rod, single arm barn door. The variable speed motor therefore maintains a constant tracking rate for the full swing of the door arm from the closed starting position to the fully open, maximum travel position.
What is a Barn Door Tracker
For astrophotography, anything more than a quick snapshot requires some way to account for the earth’s rotation. With a camera on a fixed tripod, an exposure of more than about 20-30 seconds will result in stars that look anywhere from little ovals when using a wide angle lens to long streaks for a telephoto lens. The reason for this is the stars move across the sky as the earth turns, just at the sun moves across the sky in the daytime. And even though the motion isn’t obvious, like an hour hand on a clock, the stars do move and that motion shows up in a photograph.
The sophisticated solution is a motorized astronomy “mount” that is programmed to track the sky at the same rate as the earth turns. These mounts tend to be somewhat bulky but more to the point, expensive.
Enter the “Barn Door Tracker” – a simple and inexpensive mechanism to turn the camera in time with the earth’s rotation. According to Wikipedia the barn door tracker was introduced in 1975 by George Haig with plans published in the April ’75 edition of Sky & Telescope. The barn door tracker is also known as a “Haig mount” and also a “Scotch Mount”. (I don’t know where the latter term comes from.)
Original Barn Door Tracker Design
The original design consisted of two pieces of wood connected at one end with a simple door hinge. At the other end was a threaded rod and a hand crank.
Turning the crank caused the “door” to open. With some math, a watch and a little practice, the crack could be turned to move the door at the same rate the stars move across the sky. The tracker was mounted on a tripod and positioned so the axis of the hinge was aligned to the celestial North Pole. The camera was then mounted on the top board using a simple ball and socket or like mechanism. So with this setup, turning the crack at the correct rate moved the camera so it was always pointed at the object being photographed.
Motorized Barn Door Tracker
The next innovation – for obvious reasons – was to replace the hand crank with a motor. In the 70s, electric wall clocks were generally driven by 60hz synchronous motors. This made them easy to find or salvage from a discarded clock.
A synchronous motor has the advantage of turning at a very precise rate – as precise as the 60hz supply voltage. By carefully measuring the length of the barn door so that combined with the thread pitch of the drive rod, it was possible to use the standard 60hz synchronous motor to accurately track the sky. One disadvantage was the need for 60hz power (standard household current in NA).
Multi-Arm Barn Door Tracker
The more subtle disadvantage of the original tracker is that a straight rod driving the door doesn’t translate into a constant tracking rate. As the door opens wider, the constant rate at the drive rod results in an ever increasing tracking speed. At start up the door tracks the sky just fine, but after an hour or so the tracking is slightly fast. After a couple of hours it’s noticeable too fast.
So the next innovations were various configurations of the “double arm” tracker. The multi-arm geometry allowed the constant drive speed of the straight rod to produce a more or less constant tracking for a longer period of time. But at the cost of complexity and mechanical stability.
My Curved Rod Barn Door Tracker
The root of the problem is that a straight rod driving across the chord of an arch does not produce a constant motion along the circumference. So a more direct solution is to replace the straight rod with a curved rod – one with a curve that matches the arch of the barn door.
This solves the math such that the constant drive speed along the curved rod produces a constant tracking rate through all angles of the door.
At right is the first version of my barn door tracker which had a curved rod. I decided to make it out of aluminum rather than wood mainly because i liked the look of it. I also thought that the door arms could be made more compact than with wood and still retain the necessary stiffness.
Aluminum is actually quite easy to work with – basic carpentry tools and a good file are sufficient to produce acceptable results. I also have a lightweight tabletop lathe which is adequate for turning small bearings and connectors.
The motor is a stepper driven by an “EasyDriver” card which in turn is controlled by an Arduino Nano micro-controller.
But the mechanical design is also more complicated than the simple one arm, straight rod tracker.
The basic straight rod design drives a threaded rod through a drive nut mounted in the upper arm. The other end of the rod simple presses against the lower arm and is free to slide as the door opens. The motor enhancement is just an extension of the rod and therefore drives the rod directly. Very simple and easy to build.
A curved rod cannot be attached directly to the motor, so gears are required to transfer the motor rotation to a drive nut on the curved rod.
Curving the rod is also a bit of a challenge. In terms of tracking accuracy, the radius just needs to be close. But the mechanical design requires the curve to be very close the radius of the door arm. Otherwise the gears will bind as the door opens. To allow for the radius to be slightly off, the upper support for the curved rod was made to float, but only in one direction. If the support was simply loose, then the curved rod flopped to the side – which also created binding. So a pin through the rod created a pivot point which allowed the rod to move in the required direction but kept it from flopping sideways.
And gears are bad! Ok, gears aren’t bad, but getting a gear which is perfectly round and mounted precisely centred on it’s axis is a challenge when using basic carpentry tools. When dealing with arc-second tracking and um pixels, it’s not good enough to just get a good fit. Things have to be very accurate!
The tracking measurements showed a significant error with a peak-to-peak error of 76″ (21″ rms). I could see the guide star wobbling back and forth in the guide app (Metaguide) so i didn’t bother with a test photograph with the DSLR.
I thought the tracking was bad enough to to warrant a redesign!
I tried to implement a Periodic Error Correction (PEC) using the micro controller. I thought it would be a straight forward task to include a PEC table that determined how much the motor needed to speed up or slow down to compensate for the PE. I used PECPrep to filter the PE for just the dominant period which corresponded to the period of the large gear on the drive rod. Then created an excel spreadsheet to convert the tracking deviation into tracking speed. However, i could not make any significant improvement. Either my math or programming was off or my understanding of how to implement PEC.
My Straight Door Barn Tracker with Variable Drive Speed
Version 2 goes back to the mechanically simpler straight rod design but uses the micro-controller to adjust the drive speed as the barn door opens. I found the idea on the web, so i can’t take credit for that novelty.
The conversion to a straight rod required very little mechanical redesign. Since the micro-controller determines the rate the motor turns at, the arm length is not critical – it just needs to be accounted for in the math to determine the step rate.
A new mounting was required for the motor to allow it to tilt as the door opens. The pivot allows for about 60deg of opening which translates to 4hrs of tracking. The drive rod is connected to the motor shaft with a simple in-line coupler. The coupler is threaded but a set-screw is still required to hold the rod in place. I choose to have the rod push the door open which means the coupler tends to unscrew the drive rod. (This happened during a test when the set-screw was loose.)
Similarly, the drive nut on the lower arm needs to pivot as the door opens. The brass drive nut has pins that fit into the slots in the lower arm. This component was machined from a length of hexagonal brass stock i had lying around. The “keepers” overlap the pins and hold the nut in place. Gravity would probably have sufficed, but i wanted it to be secure. The wings are secured in place by fancy knurled knobs (purchased) that are easy to turn even with gloves on. This arrangement allows for the drive nut to be easily freed from the lower arm and quickly turned back to the starting position.
As well, it makes it easy to dismantle the tracker – meaning removing the drive rod. Allowing the drive rod to be easily removed makes the whole package more portable. I can toss the dismantled tracker in the trunk with the camping gear and not worry about something getting bent.
Another reason for making the tracker out of aluminum was i thought i could make a sturdier hinge than i could buy in the hardware store. I think i accomplished that. The pin for the hinge is a piece of 5mm diameter stainless steel i salvaged from a discarded printer. I used my tabletop lathe to bore a hole in the centre of a length 1/2″ aluminum square stock. A file was used to round the corners.
The last component is the controller. It consists of an Arduino Nano micro-controller and an EasyDriver driver specifically designed for small bipolar motors. These components are mounted on a standard perf-board and enclosed in a discard iPod packaging case.
The hand controller includes:
- three buttons for reverse, run/stop, fast forward
- power LED which shows the tracker status (stopped, tracking ffw/rev)
- a second LED (mounted on the Nano) to show additional control info
- an analog POT to adjust the tracking speed
- on/off switch
- 12v power supply standard jack
In order to compensate for the different tracking speeds vs drive speeds at different angles, the motor speed (step interval) needs to vary as the door opens. The Arduino atmeg-328 isn’t great for float arithmetic and it would be a time consuming calculation to do the trig functions. So again from the example on the web, the controller uses a table to look up the drive speed for different door angles. One thing great about stepper motors is the controller knows precisely how many steps it’s made (assuming nothing jams). So the easiest way to track the door angle is to count the number of steps and use the step count as the index into the drive speed table. An excel spreadsheet was used to calculate the door angle for each table entry (step count) and then from the door angle calculate the required drive speed for that angle. Changing drive speed about once ever 0.5deg or so maintains a sufficiently accurate tracking rate, even for a fully open door.
I wire wrapped the connections onto a standard perf-board. A wire wrap tool i found out is expensive, so i made one using the 5mm stainless steel rod from the printer. This is actually version 2 of the hand controller as well. I fried the first version completely destroying the EasyDriver and disabling one of the digital i/o pins on the Nano. Since i had soldered the connections on version 1 of the controller, it made it very difficult to salvage the still working components. So version 2 is wire wrapped.
I was a bit surprised when i measured the tracking that there was a significant periodic error of 37″ peak-to-peak (10″ rms). This is much better than the with the curved rod but still some concern. The dominate period is the motor period so the likely culprit is the drive rod. A p2p error of 37″ translates to about 0.002″ along the drive rod so it doesn’t take much of a mechanical error to show up in the tracking. Further investigation is required to isolate the problem.
Again i made an attempt at re-implementing PEC. And again the same disappointing results. So i moved on to a real test.
The first light test was done on 2016-02-22, a night with a full moon and wispy clouds. The first target was Sirius. It’s easy to find in the view finder and easy to focus on. I use Backyard EOS so focusing is actually pretty easy. Framing with the lisn’t weight ball and socket is a little unsteady. But since i was using a 28mm lens, i didn’t exactly have to be precise.
This is a single frame with a mild stretch applied and no calibration frames. It’s a tight crop of the centre of the field, but at the full resolution. M41, The Little Beehive, shows up nicely. Tracking is pretty good. I think the major culprit is polar alignment. I have yet to align the polar alignment scope. I can also fine tune the tracking rate but i’ll do this after more tests with a proper PA.
The next target was Orion with a 5min exposure.
Again the tracking is pretty good with less than adequate PA showing up with the north-south elongated stars. This is a single frame with a moderate crop of just Orion and again the full pixel resolution. The processing, as above, is a mild stretch, no darks or flats.
i doubt i’ll use this mount for anything more than 5min exposures. For longer exposures i’ll use my telescope mount and probably guide, even if imaging with a wide angle lens.
A barn door tracker used with a wide angle lens or even a mid-range telephoto lens is a viable alternative to an expensive and bulky telescope mount.
This project was a lot of fun to do. I purposely choose a complex implementation just for the fun of it. A useful barn door tracker can also be constructed from simple parts using basic carpentry skills, especially when considering a manual crank.