Friday, August 31, 2012

Astro-Physics 600E QMD Go-To Conversion

Getting Started
Around my birthday last year, I was able to score an Astro-Physics 600E QMD (Quartz Micro Drive) mount from Astromart.  It is fifteen years old (or more!) and does not have any Go-To capability but was still a great birthday present! (and only cost as much as a typical leather handbag..)

After using the mount for some time (it does have Agilent axis encoders and so can be used in Push-To mode with a Digital Setting Circle) I decided that I wanted to convert it to Go-To. Initially my plan was to use an Arduino, but eventually I decided to use the Littlefoot Elegance Photo, a stepper motor controller developed in Germany. The other options were the FS-2 and Boxdoerfer (both also German), the Sidereal Technology servo controller, or the Losmandy Gemini II.

The SiTech controller costs about the same as the Littlefoot Elegance Photo, but finding proper servo motors is hard - and buying servo motors new is an expensive proposition. We can't use cheap Tamiya motors. The FS-2 is much more expensive than the Littlefoot and has a bad aura about it. Sigurd Boxdoerfer never answered my emails.

And the Gemini II controller is pretty expensive - add the cost of the AP 600E and the Gemini II, and I could have gotten an AP600E GTO (the Go-To version).

An additional, and important consideration is that the conversion must require no modifications to the mount, hence it should be 100% reversible in case I wanted to sell the mount and the buyer wanted it in an original state.


Opening Up the Mount

First step was to open up the mount. It is trivial to get the front plate off, they are held by four screws:

With the front plate off, the two stepper motors are visible. They are attached to a flat aluminum plate, which is fastened to the inside of the mount with four socket-head bolts. A long socket driver is a necessity for getting at those bolts!

The two motors are Nippon Pulse, 48 steps per revolution, with a 150:1 spur gearhead. These are incidentally almost identical to the MT-1 motors of the Vixen Great Polaris (which are the exact same model, but with a 120:1 gearhead). As an aside, look at the lot number on that motor! it was made in February 1981! and still works fine. And Astro-Physics still carries spares! one nice thing about stepper motors is that they are extremely reliable.

Notice the two transfer spur gears. These are 48-pitch Imperial gears with 64 teeth. An important gotcha: they have a 5mm bore (because the Nippon Pulse steppers have a 5mm shaft). A 1.27mm hex key is required to get the spur gears off the shaft.

You cannot buy a 48-pitch, 64-tooth Imperial gear with a 5mm bore! I have been told by Astro-Physics that these started out as 3/16-inch bore, and were bored out on a lathe. Resist the temptation to bore these out yourself - it is impossible to bore them out concentrically while maintaining perfect centering with a drill (even a drill press). If the bore is not centered, the mount's periodic error will worsen. I found out about this the hard way. Therefore, you should not buy any stepper motor with 6mm or 8mm shafts, because this will require modifying the spur gears.

The QMD motor control board uses an Ericsson PBD3517 stepper motor controller chip (actually two of them). This is an old device, and is not capable of micro-stepping. However it can half-step on the RA axis, which contributes to smoother tracking.

Keep the old motors, old motor mounting plate, and motor controller board, in case you want to restore the mount to its original glory.


Stepper Motor Selection

The problem with the existing Nippon Pulse steppers is twofold: first, they are unipolar steppers, and the Littlefoot is a two-phase stepper controller. Second, and far more important, the stock steppers have a 150:1 gearhead. Since steppers have a maximum rotational speed, the large gear reduction imposes a limit on how fast the mount can slew. For these particular motors, the maximum slew speed is around 20X sidereal, which is glacial. It would take 36 minutes to slew from horizon to horizon at that speed.

The only way to improve the slew speeds is to replace the motors. Specifically, motors with a lower gearhead reduction are needed.  After some trial and error, I discovered that you need to use Vexta PX243, PK243, PK245, or PK543 steppers (NEMA 17 size) with a spur (SG) gearbox. The common and dirt-cheap Vexta PK266 (which are NEMA 23 size) are too large to fit inside the mount.

Also important, the motors must be with spur gearbox (SG part number). Vexta steppers are also available with planetary and even Harmonic Drive gearheads. These are not usable because the gearhead output shaft is coaxial with the stepper shaft.  For a motor to fit in the AP 600E, it must have an offset output shaft like the original Nippon Pulse steppers.

Handily, the SG gearheads have a 5mm output shaft, so you can use the existing AP spur gears.

Vexta makes SG gearhead NEMA 17 motors with several gear ratios - 3.6:1, 7.2:1, 10:1, 18:1, and 36:1. The first gear ratio probably does not have enough torque, and based on Vexta's data sheet the maximum torque of the 18:1 and 36:1 gearheads is the same. Hence it makes no sense to buy motors with the 36:1 gearhead (which would suffer from painfully slow Go-To slews - although not as bad as the stock motors).

You can generally get the 7.2:1 or 10:1 motors on ebay for about $90 each. This is much lower than the new price of $220 (buying the motors new would make the Go-To conversion un-economic). However being a stubborn and cheap sort, I found the 18:1 motors for $33 each. The downside of using the 18:1 gearbox is that the maximum Go-To speed is reduced.

Ensure that the motors you get are rated at least 0.6 A, you can also get these in 0.4 A and I'm not sure if those have enough torque. Also ensure that the motors are 2-phase! a unipolar stepper (also called 5-phase) can be wired to work with the Littlefoot but the torque would be much reduced.


Stepper Motor Theory

A hybrid stepper like the Vexta PX243 and friends is limited to about 4000 half-steps per second at a high driving voltage (about 24 volts). Since these motors are 200 steps per revolution, that translates to a speed of (4000 / 2) / 200 = 10 revolutions per second.

Plenty, you say. But if you have a 10:1 gearhead reduction, the speed at the output shaft is only 1 revolution per second. Since the AP 600E has a 192-tooth worm wheel, it moves 6750 arc-seconds with one revolution of the worm.  So if our motor output shaft is rotating at 1 rps, then the mount is moving at 6750" / second - which is 450X sidereal.

Obviously with the 18:1 gearbox, the maximum speed would be 250X sidereal, and with the 36:1 gearbox, 125X sidereal.  All of these speeds are achievable - but there is a catch. To get the maximum motor speed, you need to drive the motor with a high-voltage input. Vexta publishes speed curves with their motors based on a 24V driving voltage, which is almost the best case.

The Littlefoot can take up to 30V DC input (and there is a 40V option, but it costs more).  But if you are driving the Littlefoot from 12V, which most people do, you will not get the maximum speed.

In my case, at 12V and with the 18:1 gearbox, the maximum speed I could get is 160X sidereal (as compared to a theoretical 250X). Above that speed, the stepper motor stalls. 160X sidereal Go-To is slow but usable - at 160X a horizon-to-horizon slew takes 4.5 minutes. By comparison, the original Takahashi Temma II (a comparable stepper motor controller) can slew at 350X at 12V, and 700X at 24V, while the Temma II Jr can only slew at 120X at 12V and 240X at 24V - the same physics is involved.

Incidentally the "new style Temma II" can slew at 700X at 12V. I suspect Takahashi is using an internal DC-DC converter to increase the driving voltage to the steppers.

You might wonder, why not use the 3.6:1 gearhead, that would give 1250X Go-To speed, in theory. The problem is that the gearhead is also a torque multiplier, and it takes lots of torque to turn the worm. The lower the ratio, the lower the torque. And since we are limited to a NEMA 17 motor, the torque we can get from the motor is also limited (larger motors produce more torque - but we can't put a larger motor inside the AP 600E).


Fabricating the Motor Mount

The existing AP motor mounting plate cannot be re-used, because the Nippon Pulse motors do not have the same bolt circle as the NEMA 17 steppers.

A new mounting plate can be constructed from 3mm thick aluminum plate. Take the original mounting plate and trace an outline of all its holes on the new aluminum plate.

The next step is to drill pilot holes. Use a nail (or better, a punch) to create starting holes in the aluminum sheet, so that the drill bit doesn't wander around. Don't drill out the holes with your largest bit, start with a small bit like a 2.5mm and slowly go up. This keeps the holes nice and round. Unfortunately the two large holes for the motor gearhead shafts are 18mm in diameter - and it's pretty hard to get an 18mm diameter bit. You would need a step bit (also pictured below, it looks like the devil's little ice cream cone) to drill the large holes.

Note that I have not yet cut out the mounting plate - it's much easier to drill holes in a large piece of metal than in a small piece of metal (that tends to catch the drill bit, spin around like a circular saw, and mangle your hands).

After drilling with the devil's ice cream cone. A bit of oil on the step bit and on the work will make the drilling much faster and quieter. Any oil will do, I used cooking oil.

Notice that there is a sort of raised "shelf" around the gearhead output shaft, and this shelf is 18mm in diameter. This is why the motor shaft hole in our mounting plate must be 18mm in diameter.

The bolt circle for NEMA 17 motors is 1.220" or about 31.5mm, we need to drill four mounting holes for each motor so that we can attach the motors to the plate. We use M3 bolts to attach the motors to the plate.

After all that drilling, we can attach the motors to the motor plate.

Now for the most annoying part of this activity: the new motors are much larger than the Nippon Pulse motors, so our new assembly won't fit inside the AP 600E.  We need to cut the motor plate in two:

The spur gears have been re-installed - the hub (the little metal protrusion where the grub screw threads in) is facing outward. Make sure not to install the spur gear with the hub inward, as the teeth will hit the insides of the mount and the motor will stall.


Motor Installation

We can then install the Declination motor. Do not install the Right Ascension motor first, you will not have enough clearance to install the Declination motor. Note that the Declination motor is not true inside the mount, it's a bit crooked.

This is because the motor is square, while the original motor was round. The square motor hits this large metal flange inside the mount that secures the declination axis (at the top-left, 11 o'clock position). Luckily there still is enough clearance to install the motor, and the spur gear (barely) meshes with the gear on the worm.

You need to remove the grub screw on the opposite side of the mount and check the engagement of the motor spur gear with the worm spur gear. The engagement must not be too tight, otherwise the motor will stall. My motors were double-shaft and had these nice knobs on the shaft, so it was very easy to turn the knobs with my fingers to check if the motor was turning smoothly.

After the Declination motor is installed, spur gear meshed nicely, and turning smoothly, it's time to install the Right Ascension motor. Same procedure applies.


Face Plate Fabrication

We also need to make a new face plate for the mount, as the existing AP face plate is not suitable (and it would be sacrilege to modify it). The face plate is 136mm x 60mm in size, made of 4mm thick aluminum plate. In our case, we can use the same 3mm plate used for the motor mount.

Simply cut the aluminum to size and trace the four screw holes in the corners from the original plate. For the motor connectors, I decided to use the same type of 4-pin "military style" connectors that the Littlefoot already uses. These are the same type of connectors used in the Gary Bennett mods for the Celestron CGE, and are very solid connectors. They require quite large holes for mounting. Again, the step bit is very useful for drilling these large holes.


Wiring up the motors to the screw connectors is beyond the scope of this article. The pinout of the Littlefoot motor connectors is in the manual, and determining the A, A', B, and B' connections on the 2-phase stepper motors is a simple exercise.


Testing

After wiring up the motor connectors, you must then test the mount.  Configuring the motor power and gear ratios is discussed very clearly in the Littlefoot Elegance Photo manual, so I will not repeat it here. Basically you can define a custom speed for the 2X, 8X, and 32X speeds (three positions of the right toggle switch). You can set a maximum speed of 640X, but the motor will stall (hum but not rotate) above a certain limit.

You want to use the highest reliable speed - meaning the motor should turn and not lose stepper counts. This is also a good time to check that your spur gear meshing is not causing the motor to stall.  Unlike servo mounts like the GTO, steppers are open-loop and the controller cannot detect if the stepper has stalled. So it's quite important not to be too aggressive with the speed settings.

In my case, 160X was the maximum reliable slew speed. I replaced my 12V power supply with a 19V, 5A laptop supply - but the slew speed did not improve. According to Rajiva (the Littlefoot Elegance Photo designer) you really need to use 24V or higher to get a significant improvement over 12V.


Conclusion

After putting everything back together: it doesn't look too DIY and should be quite reliable. No motors dangling off external brackets. I could even stencil in some custom lettering or decals on the front plate. The hand controller is quite strange as it has two toggle switches and four direction switches - no keypad! so navigating it is a bit inconvenient at first but the interface is quite intuitive and very efficient. German..
The hand controller does not do Go-To out of the box (it can only Go-To a specified RA/DEC) but it has an SD card slot. The Littlefoot comes with a DVD containing the necessary software to write the catalogs to an SD card. You can also create your own custom catalogs, alignment star lists, etc.

The Littlefoot does not have any Go-To alignment procedure. You simply point it at something (either using the hand controller direction buttons, or you can push the mount manually, if you have encoders like I do). Once pointed and centered, you can sync the controller to that object, and you are aligned.

Of course if your polar alignment is off, Go-To's will be inaccurate. But if for example you do a meridian flip, you just need to point (again, manually or with the direction keys) at a known bright star or DSO in that area, sync, and Go-To's in the neighborhood will be accurate. It's quite convenient because if you lose power, lose alignment, etc. you don't need to go through a long alignment routine.

With cables on the Littlefoot: RA and DEC motors, power (also the robust screw-type connector), extension port (hidden, and not used); encoder port for Lumicon-style DSC's (this I use), serial/USB port, and LAN.

And the other side of the controller: power switch, video port (for Mallincam or similar video cameras), DSLR shutter control (for Canon cameras), Focus port (compatible with Robofocus stepper-based focusers), ST-4 guide port, and hand controller port.


Next Steps

I'm current hacking on this software called Aspect which is used to manage the Periodic Error Correction (PEC) table of the Littlefoot. Unfortunately the older versions of this controller (i.e. the original Littlefoot, the Littlefoot Vpower, and  the MCU Control) only had 256 PEC cells, while the Littlefoot Elegance Photo has 1024 PEC cells. Aspect currently cannot handle 1024 PEC cells, but I have made a lot of progress in fixing that.

Currently I also have a set of Perl scripts that I'm using to manage the PEC table from the command-line. So far I have gotten the native PE of about 15" peak-to-peak down to 5" peak-to-peak which is quite good.

For my birthday this year - I am looking for an AP 900QMD or a Takahashi NJP (non Go-To versions). The AP 900QMD costs less than half of an AP 900GTO, and can be converted to Go-To using this same procedure. I have yet to find one at the right price, however.

No comments:

Post a Comment