|This has been
going on a long time in some form or another. Near the end of the first automation phase while
we were deciding where to go after ditching the C14 there became a need
to design a telescope drive that met our needs. While our efforts at
remote controlling the Celestron Compustar were quite successful it
gave us a good insight into what we didn't want in a telescope drive.
Some of the 'features' of the Compustar that we particularly wanted to
The broad specification for a new drive was this:
- Excessive heat dissipation in linear drivers.
- No remote comms while slewing.
- No uncalibrated low level axis control.
- No usable slow motion by remote control.
- Drive stepping motors of at least 5A/phase,
preferably up to 10A.
- Employ switch-mode current control to reduce power
- Stepping of at least 25,000 microsteps/second to
permit tracking and slewing with one motor.
- Fully configurable acceleration, deceleration,
maximum velocity, tracking velocity, etc.
- Soft-configurable motor current setting.
- Logarithmic joystick control
- Low-level remote control.
- Limit stop support.
- Absolute encoder feedback support.
- Single box design.
Based on a 68HC11
microcontroller and home-brew motor controller, this
worked great from a software viewpoint (ie the digital parts of the
system were good) but I couldn't tune the bugs out of the current
The basic problem which was never solved was stable current control
below 10% of maximum phase current which caused non-linear microstep
motion and loss of torque at moderate speed.
basically an evolution of the first version with only slightly modified
digital circuitry but a completely revised power output stage with a
new chipset (L6203 bridge drivers and L6506 controller). There were
definite improvements but nowhere near good enough performance to drive
was time to give up with the current control problems since this was
obviously outside my skill radius! Attempts to outsource this section
of the drive eventually led me to the GeckoDrive.
A well sorted and surprisingly compact sequencer and microstepping
driver. Virtually a drop-in replacement (functionally at least) for the
part of the drive which never worked before. :-)
A different control method from my design required upgrading the CPU to
something that could handle the higher processing workload. For this I
selected the Atmel
AVR AT90S8535 because there was a readily available
development board I could use to commission a quick test.
with a 7A/phase 2.1Nm motor.
working prototype splayed all over the desk.
4 was basically an evolution of version 3 and is the current
version which can actually be turned into a working manufacturable
product. The principles are the same as v3 but redesigned bearing
things in mind like monitor and control of the power supply, overall
mechanical design and thermal considerations. Of course the
microcontroller is designed into the system instead of using a
logarithmic joystick control mentioned in the specification may or may
not be an obvious feature. The linear control of the Compustar
had several speed modes (i.e. maximum speed attained by maximum
joystick deflection) and switching from coarse to fine required
pressing a button (albeit conveniently located on the joystick) to
'change gear'. Because telescope speed has a dynamic range of some 31dB
(1200:1), I find a logarithmic range is generally more useful because
small movements near the joystick's centre provide good low-speed
control and pushing it right to the edges can go all the way up to
maximum slew speed. At high speed, velocity resolution is quite poor
but when the scope's doing 10 degrees per second and you can't see
anything in the eyepiece what's the point?
TC400 prototype complete with box.
TC400 driving our mount.
same V4 prototype fitted into a 2U 19" rack case.
drive the we actually use right now is shown to the left. It is
essentially the same drive but mounted in a rack-mountable box which
fits in our control rack (see control
). A significant change from a usability point of view (but not from
the design point of view) is the that this uses a different controller
for the primary axis motor. It is a very fine stepping controller from IMS which, while
considerably more expensive than the Gecko,
can microstep to a 25 times higher resolution. A stepper motor
cannot actually be positioned to 1/256th of a step without careful
per-motor calibration but the point is that it drives the motor
extremely smoothly and in virtual silence. This was quite a relief
after being concerned about how much audible noise there was when
tracking with relatively large motor steps. The overall mechanical
design of this version is not that flash but what do you expect from a
software engineer. :-)
This drive has its own page.