Steve's 'Barn Door' - electric (relay) drive

Although the Barn Door was originally intended to be operated (i.e. 'driven') manually, the ultimate goal was always to Motorise it.

What is required is a Motor that, with suitable gearing, will turn the Drive nut at exactly 1 rpm.

Ignoring questions of build accuracy (and polar alignment) and assuming correct gearing etc. eventually the tracking accuracy will come down to achieving, and maintaining, the correct Motor speed.

This is easier said than done :-)

A simple DC Motor might keep 'reasonable' speed on the living room floor but, exposed to a variable voltage battery supply (as the battery runs down) and variable ambient temperature (such as the night air in the back garden), the speed will be all over the place.

Fitting a suitable voltage 'regulator' could allow the voltage (and thus the speed) to be set to some fixed value. However each night we would still have to wait for the circuit to stabilise (i.e. 'cool down') and then 'calibrate' the drive (i.e. adjust  the voltage regulator until the required speed is obtained).

Saying 'calibrate the drive' is also easier said than done. First we do a polar alignment, then pick 'any star' and engage the drive. If the star 'drifts' East/West, this may be assumed to be due to incorrect drive speed, however it  could also be due to errors in the drive bolt 'angle of bend' or errors in the gearing (teeth spacing) or even changes in drive friction / motor loading.

Even if we manage to set the speed 'correctly', there is a good chance that it will 'wander', for example as the drive nut encounters more or less resistance and places more or less load on the motor (which then slows down or speeds up accordingly).

Plainly, with no 'absolute reference' (and no feedback allowing comparison against that reference), any errors in drive speed will accumulate during the entire Exposure time. Even assuming we achieve 5% accuracy, after 10 mins we could be up to half a turn (half a minute) out.

Of course a Stepper Motor can be used, but these are rather more difficult to control (even so, see later).

What is needed, then, is some sort of 'automatic speed adjustment' system.

If the Drive nut position is monitored and compared to an accurate Clock reference, then clever electronics can be used to vary the Motor voltage to obtain the required speed and compensate for any 'drift'. This will not compensate for errors in drive bolt 'angle of bend' but will eliminate the need for  tedious Motor voltage calibration and allows the drive gearing accuracy to be ignored (essentially, any old gearing will do, since any gearing errors are eliminated by monitoring the actual Nut turning).

Complicated electronics are a bit beyond me = so I have adopted a simple approach based on  switches and Relays (see below) that anyone should be able to follow !

The basic 'clockwork' (or "hurry up and wait") Drive

The Drive nut is rotated by a simple geared DC motor powered from a (rechargeable) battery pack. Cheap gearing can be obtained by purchasing Meccano gear sets on eBay. The Meccano Motor can even be used, driven from a 6v sealed lead-acid battery.

A Relay is wired up to supply the power to the Motor.

A pair of switches are then wired up to control the Relay (and hence the Motor). Motor On is triggered from a Clock - Motor Off from the Drive Nut itself.

Even at high magnifications (long exposures) it is acceptable to be 'out' by 5 to 10 seconds (in time) per minute, SO LONG AS the error does not accumulate.

So, in the basic circuit, 'Motor On' is generated once per minute (by the second hand of a Clock sweeping past some switch) - and 'Motor Off' is generated by a switch on the Drive Nut, once per full rotation.

Motor On can be achieved (if the Clock mechanics are robust enough) by attaching a small magnet to the end of the Clock second hand which then passes over a 'reed switch' (a magnetically operated switch - as used in 'door open' sensors of typical Burglar alarms) once per minute.

Close up of the Reed Switch - it is just possible to see the two switch contacts (one above the other in the middle of the glass envelope), but not the tiny gap between them :-)

	A typical modern AA battery operated quartz type clock mechanism comes with a tiny second hand (see photo left). Plainly the magnet that came with the security alarm reed switch will be much too heavy .. so a smaller type will need to be obtained. ( or some other means used (such as a photo-detector, positioned so that the second hand sweeps past the photo 'eye' to generate the required 'trigger' (a pair of photo-detectors can be found in any old 'ball' type PC Mouse).
One source of tiny powerful magnets is the children's 'Magnetix' toy ('ball & bar' construction set) - another source is you wife's magnetic earrings (or handbag clasp :-) )

Anyway, we wire up a circuit so that each time the Clock second hand passes the detector (once per minute) it will trigger the 'Motor On' relay.

The Drive nut is then fitted with a similar switch circuit. However this is wired up to generate a 'Motor Off' trigger as each turn of the nut is completed (in fact, the Motor power circuit has to be wired up so as to keep driving until the 'Off' switch is cleared .. otherwise the Motor will become stuck on the 'Off' switch and the Clock 'On' triggers ignored).

If the Motor speed is slightly faster then 1 rpm, then, after it has been triggered by the Clock 'Motor On' it will always finish one complete turn first, thus triggering the Drive nut 'Motor Off', stopping, and then wait to be triggered again. When the Clock second hand 'catches up', it will re-trigger the Motor to start it's next turn.

Thus, so long as the Nut turns faster than the Clock second hand, it will always complete each turn first and will always be 'waiting' for the Clock second hand to 'catch up' again (hence the name "hurry up and wait" drive :-) ).

Changes in Motor speed will not lead to any accumulating error (for example it does not matter too much if the Drive nut completes one rotation in 50 seconds or 55 seconds), so long as it doesn't run too slow - should the Motor take more than 60 seconds to complete any turn, then, unless some sort of 'speed up' circuit is fitted, it will never catch up !

Nut (wheel) position Error

The main disadvantage of the above method is the 'default' speed margin required. Since the motor must drive the Nut faster than one rpm, there will always be some unavoidable positional error.

Assuming a 10% default speed margin - in which the Motor drives the Nut to complete one turn in 54 seconds, after which it waits for 6 seconds for the next trigger - the max. positional error (at the end of each turn) will be 1/10 th of a turn. This will not accumulate, so even after a 10 mins. running, the max. positional error will still be the same 6 seconds (1% in 10 mins).

The further you look from the pole, the faster the stars appear to move (at the pole itself, the stars are 'stationary'). The position of a star relative to the pole is called 'declanation'. The pole is at '90 degrees Dec' (='straight up'). Stars at '0 degrees' are (more or less) on the horizon ('flat'). Thus, the further the stars are from the pole, and the higher the magnification, the shorter the time before 'star trails' are seen. The diagram (left) shows how fast star trails will become visible depending on star position (Declanation) and magnification (lens focal length, mm) It can be seen that at the '6 second' line, to avoid visible 'trails' a lens of 135mm or less must used. If a 200mm (or higher) lens is used, then trails will be seen in stars close to the horizon after 6 seconds. With a 300mm lens, stars at 67 degrees or lower (i.e. more than 23 degrees from the pole) will show 'trails' after 6 seconds.

Plainly we would like to use higher magnification lens than 135mm (and use them to photograph stars close to the horizon). Since manually turning the Drive nut (by following  the Clock second hand) should be able to equal (or better) a 1/10th turn (6 second) error, we really should try to do better with our automated system.

NB  If you use a compass for initial 'find north', it has to be kept well away from any magnets :-)

Advanced 'clockwork' Drive

It was noticed that fitting a small magnet to the Clock 'seconds hand' arm unbalanced it .... to rebalance, a second magnet was fitted to the opposite end of the arm.

This resulted in 2 triggers per turn, and thus a second trigger had to be fitted to the Nut. As a result, the stop/start system operated twice a minute - and the max. error was halved !

Plainly every time the Drive Nut rotation is split into smaller steps (for example, half turn or quarter turn steps) - then, assuming the speed margin is kept at 10%, the unavoidable positional error is also reduced (halved or quartered).

It is a simple modification to use each corner of the hexagonal Drive Nut to operate a single 'Motor Off' micro-switch, giving 6 'motor off' triggers per revolution. The Clock circuit would then need to generate 6 similar 'Motor On' triggers (once every 10 seconds). In such a case, a 10% default speed 'margin' means each 1/6th turn is completed in 9 seconds and the positional error is now 1/60th of a turn (i.e. 1 second max. in every 10 seconds, but not accumulative, so this is 1 second per turn).

Fitting a micro-switch with an operating 'arm' against the edge of a hexagonal nut (for Motor Off) proved to be quite easy. However to generate a 10 second 'Motor On' trigger is a bit more difficult - but not impossible - for example :-

If the Clock is robust enough, the second hand could be replaced with a thin card or plastic 'disc' with 6 equal spaced 'notches' (or holes) cut into the edge. A 'featherweight' micro-switch arm operating against the edge of this disc (or photo-detector over the holes) would then give the required 6 triggers per minute. It would also be possible to fit 6 magnets around the edge of the card (to operate a single reed switch).

If the Cock is more fragile, a alternative would be to position 6 reed switches in the circle swept by the (single) magnet mounted on the second hand.

We can use the same Relay logic as the simple system outlined above - i.e. the Clock '10 second' trigger is wired to 'Motor On' and the Motor drive Nut switch trigger is wired to 'Motor Off'.

If the Motor is set to the "10% faster"  (i.e. at 66 rpm), then the max. error is now 10% of 1/6th of a rev. (i.e. 1/10th of 10 seconds = 1 second or 1/60th of a turn, 6x better than above).

To smooth out the 'stop/start' application of drive power a decent size capacitor can be wired across the Motor.

This means the supply voltage need not be very accurate (in fact it may be possible to dispense with any voltage regulator completely). 

As the battery runs down there will be a danger that the Motor will run too slow. This will be more of a problem if the initial speed is only 'slightly' too fast.

In the Basic Clock Drive, with the Motor running 10% faster than needed, the maximum error (after 1 minute) is 6 seconds. In the Advanced Clock Drive, if half that maximum error is permitted (i.e. a 3 seconds) then the motor can be run 100% faster (i.e. at double speed) since the error is reset every 10 seconds (and can not accumulate) by the 10 second timer resync.

Practical considerations.

The clock mechanism should not be mounted on the (angled) Barn Door movable arm. Doing so would mean that the magnet weighted second hand has to climb 'uphill' half the time - with the possibility that some seconds may be 'missed'.

Instead it should be positioned on the (Spirit Level adjusted) flat base - this should allow a thin disc ( with 6 small magnets) to be evenly 'balanced' on even a lightweight clock mechanism - see photo.

Even when base mounted, the fitting of multiple magnets to a fragile clock mechanism may not be practical. For balance, it can be assumed that 2 magnets are fitted to the existing second hand, one opposite the other - and then, to obtain the multiple 'On' triggers, multiple reed switches can be mounted around the clock

To obtain 6 'On' triggers, with 2 magnets, requires 3 reed switches. These have to be positioned at exact 0, 30 and 60 degree interval - this is a more complex construction and requires careful construction, since errors in reed switch position (and differences it switch response timing) will effect the trigger timing - although any such errors will be over no more than a half turn (and will not accumulate).

To keep within the 10% target per 1/6 turn error means positioning magnets and setting reed switches to operate within much less than 6 degrees of arc (1/6 th rev = 60 degrees, 10% of this is 6 degrees). If build accuracy is (say) 3 degrees, this leaves only 3 degrees for Motor speed errors.