Background
I 'inherited' the Societies old pillar mount that was equipped with an existing motor drive - however this was mains voltage driven and modern safety requirements prevent it's use in public.
The existing motor consisted of an integrated mains 50Hz synchronous AC motor plus gearbox producing 'exactly' 1 rev every 10 minutes at the RA worm shaft. As expected, the RA gear has 144 teeth - i.e. the same count as the old EQ3 etc. series mounts (the more modern EQ3-2 has 130 teeth). No feedback grating or sensor system existed and no 'speed up' / 'slow down' adjustment was possible - the accuracy of this system relies totally on the stability of the UK 50Hz mains frequency.
Whilst it may still be possible to obtain drive motors for the older EQ3's, they will all be of the 'open loop' design, with manual speed adjustment. The need for adjustments during public viewing was seen as 'unacceptable', so some thought went into producing a 'DIY' solution.
Due to limited funds, I decided to go with a digital solution utilising scrap parts where-ever possible.
Planing a solution
I first looked at obtaining a precision grating and then working backwards from that.
An old 'ball' mouse was dismantled and the sensor grating extracted. The grating was found to have 36 teeth. Direct connection to the RA worm gear shaft would thus generate 36 signals per 10 minutes = or 16.66r seconds per 'pulse', corresponding to 0.06 Hz. This was felt to be much too 'slow' to provide sufficiently accurate positional feedback. This was the first problem.
Turning to the motor, the relationship between RA shaft turn and motor 'clock' is 1 shaft turn = 10 minutes. Unfortunately, 1 rev per 10 minute geared motors are not generally available. It would thus be necessary to use a faster motor plus some intermediate gearing - but what gear ratio ? This was the second problem.
Fortunately, addressing the second problem can provide a solution to the first - if I added gearing to the motor shaft, I could position the grating on the motor shaft and 'multiply up' the grating signals by the gearing ratio. There would, of course, be some loss of accuracy since this would be before the gearing (and thus be unable to take gear errors into account)
Next I looked at possible gearing. I found two 'Meccano' brand gear wheels in the 'bits box' - one of 57 teeth and the other of 95 teeth. This would allow a motor/gearbox output of 5.7 or 9.5 rpm = and a quick check showed that both 6rpm and 10rpm motors are available on eBay.
I first considered the 5.7 rpm option. Using the 'mouse' grating here would produce 36 signals x 5.7 rpm = 205.2 per 10 minutes or 2.924 s per signal, = 0.3420 Hz. I then looked at the larger gear wheel .. 36 x 9.5 gives us 342 per 10 mins = 1.754... seconds per, 0.570 Hz. Both these frequencies would be fast enough to produce timely feed-back so both would appear to be suitable.
Next I started looking for a controlled reference clock. It quickly became obvious that not only are there no 'clocks' that would generate 0.342 or 0.57 Hz but that all generally available crystals are in the MHz range and would require long chains of 'custom' divider circuits to generate anything other than simple "hours/minutes/seconds".
It turns out that the real constraint is not the gears or the grating but the 'reference clock' !
Using the standard clock crystal
To keep divider circuits down to manageable 'counts' we must use the only sub-MHz clock crystal generally available = the 32.768 kHz 'watch' crystal.
This, together with the 'off the shelf' CD4060 clock chip (which incorporates both oscillator drive circuitry and a 14 stage binary divider), gives us a 2 Hz signal (0.5s). Also output from this chip are 9 other intermediate binary dividers (4Hz, 8Hz, 16Hz etc) as well as a (buffered) 32.768 kHz.
We can now address the gearing and sensor grating. It is immediately obvious that the gearing must be chosen first, since a 'custom' grating is simple to make, whilst custom gears are not.
Assuming that motors are not a constraint, it is apparent that [Grating count * Gear teeth] must be some value that is an integer divider of 600. For example, 20 teeth (so 2 rpm motor) would work with a grating count of 1 (30s), 2 (15s), 3(10s), 5(6s), 6(5s), 10 (3s), 15 (2s), 30 (1s) or 60 (0.5s)
A 20 teeth gear used to be available in the original Meccano set = part 26 - but part 26 was quickly re-engineered with only 19 teeth and the older 20 teeth version is now a museum piece and simply not generally available (any 'part 26' found on eBay will have 19 teeth).
There is no 30 teeth gear, however a 60 teeth gear (part 27d) is generally available ! To turn this once per 600 seconds, the motor must run at 10s per rev, or 6 rpm. To deliver 1 signal per 0.5s (2Hz) a grating on the motor shaft must have 20 slots (or some digital multiple - i.e. 40 or 80).
Whilst constructing a 20 slot grating is not difficult, a simpler solution is to use a second gear wheel on the same shaft as the worm with 20 (or 40) teeth - and use the teeth as the grating ! I already discovered that no 20t gear is available, however Meccano P96 is a plastic sprocket with 20 teeth. Unfortunately these are quite rare and thus expensive to get hold of :-( There is no 40 teeth gear or sprocket, so I decided to manufacturer a 20 slot grating after all (using a printer to create a template to guide cutting from thick plastic card).
Actual solution
The motor
A 6v motor rated at 6 rpm would drive a worm on a 60 tooth gear to deliver 1 rev per 10 minutes to the RA drive worm. The only drawback is that a 6v motor with a 6v supply would have no 'room at the top' - if the battery voltage dropped (or it was not possible to drive the motor with the full battery voltage) the motor would run at less than 6rpm.
A Chinese eBay seller offers a 7rpm motor, but only at 12v. Whilst the rated speed is ideal, the need for a 12v battery pack is less welcome.
I thus determined to build a 6 cell rechargeable battery pack delivering 6 x 1.2 = 7.2v and be prepared to over-drive the 6v motor if necessary. Fortunately, 6 cell battery packs are available for primary cell use (6 x 1.5 = 9v).
The Control circuit
A 20 slot grating on the motor shaft delivers 20 signals per rev (6 rpm, so 1 rev 10s) = 2 per second (2 Hz) which is fed to the 'down count' of the motor control 'up/down' counter (the 'up count' being driven by the 2 Hz reference clock).
The output of the counter is converted into a 'demand' voltage by the usual R-2R 'D to A' network. The 'demand' voltage level is compared to a 'triangle' wave generated from the CD4060 32kHz reference clock. The resulting 32kHz PWM wave then drives a fast switching power transistor which delivers power to the motor.
The very excellent (and free for non-commercial use) LT SPICE circuit simulation software was used to determine resistor values for the R-2R network and simulate the 'comparator' and motor driver circuit.
For a full explanation of the PWM motor drive circuit, see my previous page
Minor issues
1) The 60 teeth gear is Meccano part 27d. Ideally it should be meshed with the 'narrow' Worm gear (part 32b) - unfortunately, this is not generally available so we will have to 'make do' with the 'standard' worm gear. However this immediately brings up a concern re: worm pitch & need to make sure the worm is not going to 'slip' on the gear teeth. Some sort of spring tensioner would seem a good idea.
2) The off-the-shelf motors available from China via eBay all have very short motor shafts (less than 1 cm) and low 'mm' diameters. Meccano shaft size diameter is a fraction less than 4mm. So to connect the motor to the Meccano gear and grating shaft I decided to use some plastic tubing with an ID of 1mm and OD of 3mm. The pipe would thus 'push fit' over the motor shaft and 'push into' the Meccano gear hub. A bit of epoxy glue at the motor end prevented slippage.